Search

South Korea has announced a plan to increase the proportion of renewable energy generation to 20% and reduce traditional energy generation by 2030. Among renewable energy resources, wind power has the advantage of relatively low power generation costs. However, it is difficult to forecast, as the output varies significantly depending on changing wind conditions such as the temperature, wind speed, and wind direction. We believe that short-term wind energy forecasts are the most important part for coping with these fluctuations and minimizing scheduling errors, thereby making the grid more reliable and reducing market service costs. Accordingly, we proposed a practical short-term wind power output forecasting method using a novel ensemble model based on a wavelet transform and autoregressive integrated moving average with explanatory variable (ARIMAX) approach. To demonstrate that the model has a good forecasting performance, we applied historical wind speed and wind power output data obtained from Jeju Island's wind farm to the model, and compared them with forecasted values. The normalized mean absolute error (NMAE) was used as the error metric. The comparison results were described for three supervisory control and data acquisition points. The average NMAE was approximately 3%. In addition, an N-1 contingency analysis was conducted to check the voltage profiles and flow limits in the context of a real power system, to ensure that the power system operated stably even with the forecasted values. The system worked successfully with the forecasted values, and can be deployed as application software for energy management systems in South Korea. • The Ensemble model based on Wavelet transform and ARIMAX modelling is applied for a short-term wind power forecasting. • The practical scenarios based on a short-term wind power forecasting is applied to security analysis of power grids. • The proposed methodology is applied the empirical data of JEJU Island's wind farms and power grids. [ABSTRACT FROM AUTHOR]

With photovoltaic power being increasingly integrated into power grid, accurately forecasting solar irradiance is of critical for ensuring stable and economical operation of power systems. The forecast accuracy in ultra-short-term horizons can be greatly improved by employing ground-based sky images. Although wide range of computer vision methods have been used for modelling, effectively extracting spatiotemporal features from sky image sequence is still a tough task. In this study, a sparse spatiotemporal feature descriptor is introduced to enhance the process of dynamic spatiotemporal information extraction from continuous grayscale sky images, while spatial pyramid pooling is used for feature refinement. Parallelly, dense convolutional network is used to extract static features from the nearest single-frame RGB sky images. Both dynamic and static spatiotemporal features were adequately extracted and subsequently fused for the multi-step prediction of global horizontal irradiance. In addition, various competitive models in object detection are adopted as benchmarks for comparison. The experimental results revealed that the proposed method outperformed baseline models, with up to 5.51% reduction on normalized root mean square error (NRMSE) and 9.38% improvement on ramp event forecast. The proposed method can be widely applied to photovoltaic stations equipped with all-sky-imagers. [ABSTRACT FROM AUTHOR]

The short-term forecast of photovoltaic (PV) power is crucial for the security and economics of power system operations. However, the fluctuation characteristics of the PV power, which are closely related to the meteorological factors, introduce inaccuracies in its forecast. Towards this end, the paper studies the effects of clustering analysis at long time scale and data reconstruction technique at short time scale on capturing PV power fluctuation characteristics. A short-term PV power forecasts method based on multi-scale fluctuation characteristics extraction (MFCE), which employs a path analysis to identify the relevance of meteorological factors with PV power at long time scale and a phase space reconstruction to analyze PV power fluctuation characteristics at short time scale, is proposed in this paper. The proposed MFCE methodology deploys a widely-used extreme gradient boosting (XGBoost) model to output the forecasting results. Both the effectiveness and accuracy of the proposed methodology are verified by using the real data under the conditions of sunny and cloudy days of four seasons compared to traditional methodologies. • The multi-scale fluctuation characteristics are analyzed for PV power forecast. • Identify dominant meteorological factors affecting PV power by path analysis. • The hour-level fluctuation is reduced by using path analysis-based clustering. • The phase space reconstruction is used to reduce the minute-level fluctuation. [ABSTRACT FROM AUTHOR]

Large-scale integration of photovoltaics (PV) into electricity grids is challenged by the intermittent nature of solar power. Sky-image-based solar forecasting using deep learning has been recognized as a promising approach to predicting the short-term fluctuations. However, there are few publicly available standardized benchmark datasets for image-based solar forecasting, which limits the comparison of different forecasting models and the exploration of forecasting methods. To fill these gaps, we introduce SKIPP'D—a SKy Images and Photovoltaic Power Generation Dataset. The dataset contains three years (2017–2019) of quality-controlled down-sampled sky images and PV power generation data that is ready-to-use for short-term solar forecasting using deep learning. In addition, to support the flexibility in research, we provide the high resolution, high frequency sky images and PV power generation data as well as the concurrent sky video footage. We also include a code base containing data processing scripts and baseline model implementations for researchers to reproduce our previous work and accelerate their research in solar forecasting. • A curated sky image and PV generation dataset is released for short-term solar forecasting. • Processed benchmark data and raw data are both provided for flexibility of research. • Reference codes for data processing and baseline model implementations are provided. • Baseline deep learning models are developed to demonstrate the uses of the dataset. [ABSTRACT FROM AUTHOR]

Combining the benefits of an empirical mode decomposition (EMD), a convolutional neural network (CNN), and a long-short term memory (LSTM) neural network, this research proposes a hybrid model for forecasting short-term multistep ahead wind speed that can assist microgrids in achieving their low generation cost and carbon neutrality goals. Prior to EMD, historical wind speed time series were pre-processed, and then the resulting time series of wind speeds were decomposed into intrinsic mode functions (IMFs). Following that, a CNN-LSTM model that had been subjected to optimization was employed to carry out the training and subsequent evaluation (validation and implementation of the model). While CNN is able to directly extract the time series' intrinsic features, LSTM networks can fully leverage the data for improved prediction. In order to demonstrate the superiority of the proposed model, its results were compared to those of six other models (including hybrids and standalones), and the findings of the experiments provide empirical evidence that the hybrid EMD-CLSTM model proposed is superior. The strategy that has been proposed has also been utilized in industrial settings, where it has been found to be successful. The green energy management system (GEMS) of Leonics Co. Ltd. has incorporated the proposed model into their short-term decision-making and scheduling procedures in order to achieve their goals of a better energy mix, low-cost power generation, and carbon neutrality. Last but not least, the energy management system of the modified GEMS microgrid was upgraded to enable a greater penetration of wind energy sources without decreasing the microgrid's stability or reliability, making it possible for the system to eventually become carbon neutral. • Using data decomposition, a hybrid multi-steps forecasting model has been proposed. • The proposed model is applicable to real time microgrid energy management system. • The performance of the proposed model has been validated using two real datasets. • The proposed model can help increase wind energy share and reach carbon neutrality. [ABSTRACT FROM AUTHOR]

Accurate PV power forecasting is becoming a mandatory task to integrate the PV plant into the electrical grid, scheduling and guaranteeing the safety of the power grid. In this paper, a novel model to forecast the PV power using LSTM-TCN has been proposed. It consists of a combination between Long Short Term Memory and Temporal Convolutional Network models. LSTM is used to extract the temporal features from input data, then combined with TCN to build the connection between features and outputs. The proposed model has been tested using a dataset that includes historical time series of measured PV power. The accuracy of this model is then compared to LSTM and TCN models in different seasons, time periods forecast, cloudy, clear, and intermittent days. For one step forecasting, the results show that our proposed model outperforms the LSTM and TCN model. It has carried out a reduction of 8.47%, 14.26% for the autumn season, 6.91%,15.18 for the winter season, 10.22%,14.26% for spring season and 14.26%, 14.23% for the summer season on the Mean Absolute Error compared with LSTM, TCN. For multistep forecasting, LSTM-TCN surpassed all compared models in different time periods forecast from 2 steps to 7 steps PV power forecasting. [ABSTRACT FROM AUTHOR]

Offshore wind turbines are exposed during their serviceable lifetime to a wide range of loads from aero-, hydro- and structural dynamics. This complex loading scenario will have an impact on the lifetime of the asset, with fatigue remaining the key structural design driver for the substructure, e.g. the monopile. The ability to monitor the progression of fatigue life of these assets has recently become an operational concern. To achieve a monitoring alternative to strain gauges (cost-prohibitive farm-wide installation), supervisory control and data acquisition (SCADA) systems, often coupled with acceleration measurements, have been used. Existing work focused primarily on ten-minute fatigue load estimation. However, fatigue accumulates over time and the ability to accurately monitor this accumulation of fatigue over a longer time-window is paramount. In this contribution we investigate a novel approach using nine months of real-world SCADA and acceleration ten-minute statistics as inputs of a neural network model for long-term DEM estimation. This is further enhanced by including physical information relative to the problem at hand into the neural network model, in a so-called physics-informed machine learning approach. Specifically, we employ a custom loss function – the Minkowski logarithmic error – which prioritizes conservativeness (over-prediction of fatigue rates) and to embed the damage accumulation into the machine learning model. In the results and discussion section, we use Tower-TP interface load measurements for three real-world turbines to demonstrate the concept. First a model is trained on one turbine, before being applied to all three locations. The model performance is compared to direct measurements of fatigue progression at all three locations and a control loss function (the mean squared logarithm error) in both the Fore-Aft and Side-Side loading directions. The novel physics-informed model clearly outperforms the control loss function in accumulated fatigue predictions with errors below 3% on accumulated damage equivalent moment estimates. Additionally, the model's long-term performance is checked according to varying timescales (hourly, daily, weekly or monthly accumulation of DEM) and it is seen that the spread on the error rapidly reduces and converges to zero. Finally, the long-term DEM is accumulated into a monthly accumulated fatigue damage. We also see how the error on this begins to converge after 6+ months, which opens the door for extrapolation. • RUL/fatigue estimation on XL monopiles of OWTs using 9 months of real data. • SCADA and acceleration data as alternatives to strain gauge measurements. • Physics-guided learning of neural networks greatly increases model performance. • SCADA and acceleration data can be used for long-term fatigue damage estimation. • Errors on DEM are kept below 3% and on damage below 15%. [ABSTRACT FROM AUTHOR]

As the energy crisis intensifies, wind energy generated by wind turbines, commonly known as a promising renewable energy source, is being more frequently employed. As a result, wind energy forecasting, particularly wind speed forecasting, plays a crucial role for wind energy management. Due to their importance, many wind speed forecasting methods have been proposed. However, many of the traditional prediction models did not account for data pre-processing, or the constraints of an individual forecasting model, resulting in poor prediction accuracy. The purpose of this article is to present a unique forecasting model that incorporates noise-processing methods, statistical techniques, deep learning frameworks, and optimization algorithms to improve on existing methods. The suggested ensemble model was tested using 10-minute wind speed data from real-world conditions. The experimental results show that the mean absolute percentage error of the 10-minute prediction of the proposed model is 5.73%, which is also about 17% improvement compared to the competing model (mean absolute percentage error of 6.71%). [ABSTRACT FROM AUTHOR]

The demand for electrical energy continues to rise at an alarming rate, while supply is not proportional to power demand. Thus, in order to accomplish appropriate preparation and operation of an electrical power system network, it is crucial to properly assess the present and future consumption of electric power systems. Meeting the electric load needs entails actively anticipating the amount of energy required to meet the subscribers' power demand. To forecast energy demand, many algebraic and artificial intelligence schemes have been used. This paper proposes an improved medium-term forecasting scheme that employs an artificial neural network (ANN). The model was developed and trained with the help of an artificial neuro-fuzzy inference system (ANFIS). The data used, which ranged from 2014 to 2019, was obtained from the University of Nigeria Nsukka's (UNN) research and energy Centre, and the works department of the same university. The model was designed and simulated using MATLAB 2018a, and the results were evaluated using MAPE. The proposed system used a robust machine learning based scheme. Unlike conventional systems, the system demonstrated a MAPE of 4.34% and a correlation of 0.91144315 at predicted and actual loads, as shown in Table 2. The proposed system shows that the predicted loads has 25th and 75th percentiles of 0.58MW and 0.98WM, while that of the actual loads are 0.649MW and 0.936MW. [ABSTRACT FROM AUTHOR]

Maritime surveillance is of utmost priority for a nation's security, and hence it's economy. For maritime awareness, coastal surveillance, and maritime activities in the Region of Interest (ROI) should be monitored. One of the ways to keep this in check is to restrain unwanted infiltration. Monitoring unwanted infiltration is feasible through vessel trajectory forecasting and anomaly detection in real time. Most solutions for trajectory predictions are available but require a huge amount of historical data, and high-power computing resources. Here, the requirement is developing a decision support framework consisting of both lite weight approaches for short-term predictions and deep learning-based techniques for long-term forecasting. This paper aims to find the suitability of Linear Stationary Models (LSM) like the Auto-Regressive Integrated Moving Average Model (ARIMA) for predicting and forecasting the Vessel Trajectory as means of lite weight short-term predictions. For this purpose, the Automatic Identification System (AIS) dataset of the U.S. West Coast is used. The significant effort was for data pre-processing to create a robust dataset for model training. An appropriate model after the model-selection process is used for trajectory forecasting. The model's accuracy is validated using Root Mean Square Error (RMSE) performance indices for residual and forecast errors. A window generator model is integrated with the best-fitted ARIMA model for recursive real-time predictions, with varied sizes and visualization. The proposed time-series model provided a very high accuracy as the RMSE value for prediction and 48 hours forecast are 0.023 and 0.017, respectively. [ABSTRACT FROM AUTHOR]

COVID-19 is a pandemic that has resulted in numerous fatalities and infections in recent years, with a rising tendency in both the number of infections and deaths and the pace of recovery. Accurate forecasting models are important for making accurate forecasts and taking relevant actions. As a result, accurate short-term forecasting of the number of new cases that are contaminated and recovered is essential for making the best use of the resources at hand and stopping or delaying the spread of such illnesses. This paper shows the various techniques for forecasting the covid-19 cases. This paper classifies the various models according to their category and shows the merits and demerits of various fore-casting techniques. The research provides insight into potential issues that may arise during the forecasting of covid-19 instances for predicting the positive, negative, and death cases in this pandemic. In this paper, numerous forecasting techniques and their categories have been studied. The goal of this work is to aggregate the findings of several forecasting techniques to aid in the fight against the pandemic. [ABSTRACT FROM AUTHOR]

• Examines resources allocation issue for minimizing the solar power usage in fog computing-enabled wireless IoTs using NOMA. • A typical case is considered to support several IoT devices for offloading computations together to the similar FN. • As a way of achieving an optimal efficiency-complexity balance, a whale optimization algorithm (WOA) is introduced. • The suggested layout is more efficient than current methods in terms of throughput, latency, blackout likelihood, and power usage. Fog computing is used to handle applications that require a lot of calculations at the network's edges for wireless Internet of Things (IoT) networks that have source-limited equipment. In addition to reducing computational time and fronthaul traffic information, fog computing still poses challenges in allocating the existing computing and communication sources, while maintaining the strict quality of service (QoS) standards. The current study examines the issue regarding task management and heterogeneous resource allocation among equipment for wireless IoT networks. As a means of transferring the information to the fog computing nodes (FNs), IoT devices collecting large amounts of information require appropriate offloading decisions. This paper investigates how non-orthogonal multiple access (NOMA) can be deployed for IoT networks in order to enable many devices for transmitting information of the load and the market to a similar FN simultaneously, in a similar frequency range, and in the same code domain. Several IoT devices are optimized together based on their resource allocation and transmission power in this paper while meeting the QoS standards. Additionally, the optimization problem can be expressed in the form of a mixed-integer nonlinear programming problem with the goal of minimizing the system's power usage. The problem would be NP-hard so a whale optimization algorithm (WOA) has been proposed for solving. According to simulation outcomes, the suggested architecture performs well in terms of throughput, delays, blackout likelihood, and power usage. [ABSTRACT FROM AUTHOR]

• A combined approach is proposed according to MMI, deep learning (FCRBM), and MTLA to forecast short-term load quickly. • A FCRBM-enabled prediction and training model is developed for increasing the accuracy. • A digital twin concept is developed to handle the problem complexities. Because users behave randomly and in a nonlinear way, predicting electrical loads proves to be a difficult procedure. Due to the development of the smart grid (SG) and advanced metering infrastructure (AMI), humans will be capable of recording, monitoring, and analyzing these non-linear behaviors. The use of electric load projection layouts is a necessity in order to make decisions, plan, and evaluate contracts in electrical systems. Consequently, there have been several load prediction methods in the research that demonstrate trade-offs among prediction precision and runtime (convergence rate). The current paper presents a method for short-term load prediction of market in trade-by-trade data that would be quick and precise. Modified mutual information (MMI) is used to extract abstractive characteristics from historic information. Learning empowers the factored conditional restricted Boltzmann machine (FCRBM) for predicting the electrical loads. Ultimately, the efficiency has been optimized using the suggested modified teaching–learning algorithm (MTLA). The suggested architecture has the advantage of improving prediction precision and convergence rate. The MMI method and FCRBM layout improve prediction precision. In addition, MTLA has been used to enhance the convergence rates. Based on simulation outcomes, the suggested quick and precise layout performs better than conventional layouts when it comes to forecasting precision and convergence rates, including Bi-level, MI-artificial neural network (MI-ANN), and accurate fast converging short-term load forecast (AFC-STLF). [ABSTRACT FROM AUTHOR]

In this paper, a dynamic ensemble method is proposed to forecast the residential short-term load accurately. The main idea is to utilize the state-space approaches to dynamically adjust the weight coefficients used to combine the base models. The dynamic ensemble method is divided into two stages. In the first stage, the least square method is utilized to estimate the weight coefficients. In the second stage, the particle filter is employed to exclude the estimation error in the first stage and update the weight coefficients dynamically to improve the ensemble accuracy. Then, three heterogeneous models (i.e., vector autoregression model, Gaussian process regression model, and the long short-term memory neural network) are employed as the base models and combined based on the weight coefficients to forecast the residential load. The numerical tests are conducted considering two scenarios in two public datasets to investigate the performance of the proposed method. The first scenario assumes the base models are accurate and stable, while the second scenario considers the dynamics of the base models. Root mean square error, mean absolute error, and mean absolute percentage error are used to evaluate the performances. These indicators show that the proposed method outperforms in both scenarios and datasets. [ABSTRACT FROM AUTHOR]

• Short-term electricity generation data is used to calculate PGF for Khartoum. • Peaks over threshold method is utilized to model the highest power generation. • The obtained PGF equals 4, whereas in PV potential map equals 4.8. • This factor shall facilitate PV system sizing for local engineers. In this research, the authors used the Peaks over Threshold (POT) method alongside short-term electricity generation data belonging to a 5.5 kW p off-grid photovoltaic (PV) system installed on the premises of the National Energy Research Center in Soba district, Khartoum, to estimate the panel generation factor (PGF) of the city that rests within Sudan's capital. PGF is a fundamental technical figure primarily used in the design/sizing of solar PV systems. The addressed problem here is to calculate this coefficient with an acceptable accuracy constrained by insufficient electricity generation data from a PV system operating at its maximum capacity, relying on statistical modeling methods, which is the research novelty. Accordingly, we obtained an approximate PGF of 4 utilizing the POT technique that resulted in a threshold of 19 and Generalized Pareto distribution estimators of 2.2 × 10 - 16 and 2.3 for the shape and scale parameters, respectively. In the literature, the corresponding factor value is 4.8, illustrated in Sudan's PV potential map, based on historical long-term solar irradiance satellite records. Consequently, a 16.67% Percent Error between the two values is present due to the big difference in data amount, favoring the literature. However, in front of relatively abundant data, we believe the Extreme Value Theory methodology applied here will demonstrate robustness in approximating the target factor digits and validating Sudan's map numbers. Finally, this effort shall enable local engineers to blueprint well-assessed PV systems and support the endeavors to scientifically document Sudan's PV energy capacity, which the country lacks very much. [ABSTRACT FROM AUTHOR]

Short-term forecasting of direct normal irradiance (DNI), which refers to photons that did not interact with the atmosphere on their way to the observer, is of great interest in the advanced management of concentrated solar power systems. So, this paper exhibits a hybrid model dedicated to the intrahour forecasting of DNI (horizons are 5, 10, and 15 min). This hybrid model combines a knowledge-based model with a machine learning model: the knowledge-based model is used for clear-sky DNI forecasting from DNI measurements; the machine learning model evaluates the impact of the atmospheric disturbances on the solar resource, through the processing of high dynamic range sky images provided by a ground-based camera. The hybrid model is evaluated by comparing its performance, for different sky conditions, with that of two machine learning models based on past DNI observations only. The results highlight the pertinence of combining knowledge-based models with data-driven models, and of integrating sky-imaging data in the DNI forecasting process: the proposed hybrid model successfully manages clear-sky, overcast, and mixed situations. Regardless of the forecast horizon, the hybrid model outperforms the reference models both in normalized root mean squared error and in DNI ramp detection. • Short-term forecasting of direct normal irradiance (DNI) is obtained using a hybrid model, using both DNI measurements and ground-based sky imagery. • This hybrid model is compared to the smart persistence model and two machine learning models. • The results highlight the pertinence of combining knowledge-based models with data-driven models, and of integrating sky-imaging data. • Regardless of the forecast horizon, the hybrid model outperforms reference models, both in root mean squared error and in DNI ramp detection. [ABSTRACT FROM AUTHOR]

With aggressive government policies driving the electric power systems around the globe towards high renewable energy penetration, methodologies for considering forecast uncertainties in power system analysis and planning are gaining relevance. Probabilistic load flow (PLF) is a useful tool to carry out risk-based power system analysis and planning. This article discusses the need and philosophy of Normal-transformations, elaborating on Nataf transformation and polynomial normal transformation (PNT), to solve PLF and presents a unified approach for representation. Through detailed statistical modeling, the Gaussian mixture (GM) model is found to fit the historical wind and load power forecast error data better compared to widely used parametric models. Sampling of correlated marginal distributions, for scenario generation in PLF, requires the quantile function of the GM distribution, which is not available in a closed expression. A strategy, using quantile functions of Gaussian component sampling and interpolation, to approximate the GM's quantile function is presented. The presented approximation enables the use of GM for solving PLF with correlated marginals, which is not addressed in the literature. This work also contributes to prevailing research questions by two new performance comparisons: between various higher-order PNT embedded to point estimate methods (PEMs) for PLF, and between PEM and Quasi-Monte Carlo (QMC) methods for PLF analysis. When PEMs are employed for PLF, there is no improvement observed in going for higher than third order PNT (3PNT). Results show that estimating the first four moments of the output variables through QMC-PLF is more accurate compared to PEM-PLF. Formulae useful for implementation of various higher-order PNT, for QMC and PEM, are derived and presented for direct use. Finally, the impact of correlated non-normal modeling of inputs on PLF output variables' distribution is demonstrated. [ABSTRACT FROM AUTHOR]

Accurate multi-load forecasting is a crucial prerequisite for the steady and effective on-demand operation of integrated energy systems. Due to the intricate interdependencies between different energy sources and spatial and temporal dynamic variations, however, existing methodologies often struggle with data integration, model complexity, and capturing cross-energy correlations. To address these issues, the multi-energy load forecasting problem is transformed into a hierarchical multi-task learning task, resolved by using a spatiotemporal attention mechanism and a gated temporal convolutional network. Initially, the coupling relationships between different energy sources are analyzed both qualitatively and quantitatively to identify the coupling relationships between different energy loads. Subsequently, the identified couplings serve as guidance for designing a hierarchical multi-task learning network by distinguishing between relevant and less relevant loads. Moreover, by incorporating spatiotemporal attention, a gated temporal convolutional network with hierarchical structure is devised to achieve multi-task learning. Finally, the effectiveness of the proposed method is validated through multi-energy consumption data spanning 14 months from an energy management system. Compared to the several baseline methods, the proposed method reduces the mean absolute percentage error by up to 25.09%, 25.96%, and 17.46% on cooling, heating, and electrical load forecasting tasks. The positive experimental results and interpretable analysis demonstrate the promising application prospects of the proposed method in multi-energy load forecasting and demand-side management. • A hierarchical multi-task learning network is presented for load prediction. • A gated TCN is improved by integrating it with spatiotemporal attention. • A real multi-load prediction issue is solved with some degree of interpretability. [ABSTRACT FROM AUTHOR]

Electricity is fundamental to the development of national economies and societies, reliant on accurate power load forecasting for its stable supply. Ultra-short-term power load forecasting analyzes historical power load data to predict load changes within the next hour. This forecasting is crucial for achieving efficient power dispatching, improving emergency management, and ensuring the stable operation of the power system. However, with the increasingly widespread application of renewable energy, its inherent intermittency exacerbates the complexity and randomness of power loads, posing a challenge for models to accurately capture data features. In addressing this challenge, the study presents a novel method for feature extraction from time series data, aimed at enhancing the accuracy of power load forecasting. By analyzing trend, periodicities, and randomness, it simplifies complex time series data into several stable data features, significantly reducing noise-induced errors and enhancing the identification and understanding of power data features. Moreover, this study applies the feature extraction method to five prevalent deep learning models. Experimental results show that the deep learning models using this feature extraction method reduces the mean absolute percentage error by an average of 54.6905%, 42.6654%, and 51.3868% on datasets from three different substations in China. These results not only affirm the method's efficacy in forecasting power load but also provide new technical foundations for the reliable functioning of future power systems. • A new framework for minute-level ultra short term power load prediction is developed. • A new feature extraction method for time series is proposed. • The method simultaneously considers the complexity and randomness of power load data. • Minute-level power load prediction within an acceptable time frame is achievable. [ABSTRACT FROM AUTHOR]

Accurate and timely photovoltaic (PV) power forecasting is crucial for the stable operation of power systems. To address the issue of sparse PV power data on rainy days, this paper proposes the use of Wasserstein Generative Adversarial Networks with Gradient Penalty (WGAN-GP) to augment rainy-day data. A progressive gradient penalty strategy is introduced to avoid gradient vanishing. A Bidirectional Long Short-Term Memory network is employed for power forecasting, with adaptive hyperparameter optimization achieved through an improved Sparrow Search Algorithm (LSSA). The optimization capability is enhanced using a chaotic mapping strategy. Validation on data from a 1 MW PV plant demonstrates that the LSSA-optimized model achieves the best forecasting results. After augmenting the rainy-day dataset, the Mean Absolute Error and Root Mean Square Error decreased by 15.3 % and 6.15 %, respectively, indicating that WGAN-GP significantly improves forecasting accuracy on rainy days. [ABSTRACT FROM AUTHOR]

With the development of electricity spot markets, accurate electricity load forecasting enables power generation companies to supply the right amount of electricity, greatly avoiding power waste. As a result, time series forecasting in the field of power can bring great benefits. Previously, the Informer model successfully introduced the Transformer into long time series forecasting(LTSF) by proposing the ProbSparse self-attention mechanism, which solved the inherent problem of high memory complexity in self-attention. Recent research has further demonstrated the potential of the self-attention for mining complex dependencies. However, the limited amount of historical data has become one of the main challenges in applying deep learning techniques to power LSTF tasks. Previous researches often add a large number of time covariates to provide more information. In this paper, to address this issue, (i) we design a simple but effective Season-aware Block to enhance the model's ability to mine artificial prior information in temporal covariates; (ii) we conduct experiments using the provincial power data of Zhejiang Province, China, from 2019 to 2022, and our model outperforms other models, achieving a 19 percent MSE relative improvement, which can help power generation companies reduce costs and further reduce prices to benefit consumers; (iii) we conduct ablation experiments to assess the efficacy of the Season-aware Block in extracting temporal periodic features. Furthermore, we elucidate the underlying reasons for the effectiveness of both the self-attention mechanism and the Season-aware Block through visualization experiments through visualization techniques. [ABSTRACT FROM AUTHOR]

• Exploration of the impact of long-term dependence in electricity prices on forecasting model selection. • Proposal of a method to assess long-term dependence using hurst exponent, ACF, and PACF functions. • Development of a SARIMA-LSTM hybrid forecasting model to effectively handle long-term dependencies. • Demonstration that prediction accuracy can be improved by up to 50.98 % and stability by up to 68.09 % using real electricity market data. Day-ahead electricity market is crucial for ensuring the balance of electricity supply and demand. Its electricity price serves as a key guide for market participants and power dispatching agencies, making accurate short-term price forecasting essential. In this paper, a hybrid short-term electricity price forecasting method considering the long-term dependence of wavelet transform (WT) series is proposed. Different from the existing studies that do not distinguish the characteristics of WT series, this paper uses Hurst exponent and autocorrelation function (ACF) and partial autocorrelation function (PACF) plots to analyze the long-term dependence of subseries, and chooses seasonal autoregressive integrated moving average (SARIMA) or long short-term memory (LSTM) method to construct a hybrid forecasting model according to the characteristics of each subseries. Validation using real electricity prices from the NYISO and PJM markets shows that, compared to methods that do not consider long-term dependencies of subseries, the proposed method can improve prediction accuracy by up to 50.98 % and stability by up to 68.09 %. [Display omitted] [ABSTRACT FROM AUTHOR]

Rapid progress in machine learning and deep learning has enabled a wide range of applications in the electricity load forecasting of power systems, for instance, univariate and multivariate short-term load forecasting. Though the strong capabilities of learning the non-linearity of the load patterns and the high prediction accuracy have been achieved, the interpretability of typical deep learning models for electricity load forecasting is less studied. This paper proposes an interpretable deep learning method, which learns a linear combination of neural networks that each attends to an input time feature. We also proposed a multi-scale time series decomposition method to deal with the complex time patterns. Case studies have been carried out on the Belgium central grid load dataset and the proposed model demonstrated better accuracy compared to the frequently applied baseline model. Specifically, the proposed multi-scale temporal decomposition achieves the best MSE, MAE and RMSE of 0.52, 0.57 and 0.72 respectively. As for interpretability, on one hand, the proposed method displays generalization capability. On the other hand, it can demonstrate not only the feature but also the temporal interpretability compared to other baseline methods. Besides, the global time feature interpretabilities are also obtained. Obtaining global feature interpretabilities allows us to catch the overall patterns, trends, and cyclicality in load data while also revealing the significance of various time-related features in forming the final outputs. • A novel deep learning based, post-hoc global interpretability method is proposed based on multi-scale temporal decomposition. • Comparison to baseline load forecasting models demonstrate both superior forecasting accuracy and physically consistent interpretations. • Flexible structure can be utilized for power forecasting with various temporal patterns by adjusting decomposition kernels and the kernel sizes. [ABSTRACT FROM AUTHOR]

— In this study, an image processing-based deep learning approach for short-term forecast of solar radiation has been developed. For this purpose, firstly, cloud movements occurred during the day are tracked and future cloud movements are forecasted, accordingly. Subsequently, using the cloud motion estimation and extraterrestrial solar radiation data, 1-min averaged solar radiation values are estimated for 5-min time horizon. Shi-Tomasi method is employed to determine the feature points to be tracked on the sky images whereas, Lucas-Kanade optical flow method is employed to track the determined feature points on the sequential images. Average cloud velocity and directions are calculated by the help of linear regression method from tracked cloud movements. A hybrid approach including K-means and red/blue ratio is built to classify the pixels of the image whether they are clouds or sky. Finally, short-term solar radiations are estimated using the Long-Short Term Memory (LSTM) deep learning method. The performance of the proposed approach is compared with other methods in the literature. As a result it is concluded that, developed approach outperforms most methods in the literature with RMSE values of 47.576, 53.830, 68.103, and 92.386 for four different days and can be used as an alternative approach. [ABSTRACT FROM AUTHOR]

The biggest wind farm of Turkey is placed at Manisa which is located in the Aegean Region. Electricity is a non-storable commodity for that reason, it is very important to have a strong forecast and model of the potential electricity production to plan the electricity loads. In this study, the aim is to model and forecast electricity production of the wind farms located at Manisa by using a univariate model based on sequence-to-sequence learning. The forecasting range of the study is from short term to midterm. The strength of the proposed model is that; it only needs its own lagged value to make forecasts. The empirical evidences show that the model has high coefficient of variation (R2) in short term and moderate R2 in the midterm forecast. Although in the mid-range forecasts R2 slightly decreases mean squared error and mean absolute error shows that the model is accurate also in the midterm forecasts. The proposed model is not only strong in hourly electricity production forecasts but with a slight modification also in forecasting the minimum, maximum and average electricity production for a fixed range. This study concludes with two fresh and intriguing future research ideas. [Display omitted] [ABSTRACT FROM AUTHOR]

• Overall energy efficiency (OEE) is a crucial factor in tractor-implement systems. • Response surface methodology (RSM) identifies the influence of factorial changes. • Higher correlation value associated with the subsoiler's OEE prediction model. • RSM model accurately predicted the OEE in current study. • Vertical load has a negligible effect on the tested parameters and OEE. The aim of this paper was to predict and optimize the overall energy efficiency (OEE) of a tractor-implement system in semi-deep tillage via response surface methodology (RSM) approach. The OEE was affected by two tillage tines (i.e., subsoiler and paraplow), four forward speeds (i.e., 1.8, 2.3, 2.9 and 3.5 km/h), three operating depths (i.e., 30, 40 and 50 cm) and vertical load imposed on tractor rear wheels at two levels (i.e., 225 kg and no-weight). All tests were replicated four times, forming 192 data points. Field test results revealed all variables were influential on the OEE except vertical load. An increment of speed and depth increased OEE. The RSM approach displayed 3D views with higher accuracy for the OEE change due to changing tine, speed, depth and vertical load relative to regression prediction models. Another feature of the RSM approach was the output graphs with many small pixels. Accordingly, input variables changes and their influences on the OEE are more locational and visible. Moreover, the RSM model accurately predicted the OEE as 35.216% with the desirability of 0.843 at the optimum state as paraplow tine, 30 cm depth, 2.07 km/h forward speed, and 0.01 kg vertical load. [ABSTRACT FROM AUTHOR]

To improve the forecasting accuracy of short-term wind power, an intermittency-trait-driven methodology is proposed in this paper. In the proposed methodology, four main steps, i.e., intermittency-trait-driven data decomposition, intermittency-trait-driven mode reconstruction, intermittency-trait-driven component prediction and intermittency-trait-driven ensemble output, are involved. In particular, the selected distribution function driven by intermittency trait analysis is used to improve the decomposition model, so as to solve the end effect issue in the rolling decomposition. For illustration and verification, the wind power data with 15-min intervals is introduced as the sample data. Compared with the benchmark models, the root mean squared error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) of the proposed model is reduced by 22.357%, 26.457% and 23.828% on average, respectively. The results indicate that the proposed intermittency-trait-driven methodology outperforms all benchmark models in multi-step-ahead forecasting, which can provide valuable insights to effective wind power prediction. [ABSTRACT FROM AUTHOR]

Accurate load forecasting is critical to guarantee the security, steadiness and economic operation of the power system. Therefore, in order to improve the load forecasting accuracy, improved coyote optimization (ICOA) was proposed in this paper to optimizing the Extreme learning machine for load forecasting, which is a novel load forecasting model. The ICOA algorithm effectively improves the accuracy and speed of convergence by introducing the sobol sequence and improving the pack culture trend, which is approved beneath a set of benchmark capacities from the Organized of Electrical and Hardware Engineers (IEEE) Congress on Developmental Computation (CEC) 2017. The result comparisons have moreover shown the exceptional execution of ICOA, since it could rank to begin with within the optimization of 16 benchmark. In addition, the ELM was optimized using to address the problem that the randomly formed weights and thresholds of the ELM lead to an unstable model, as well as experiment with different models using real electricity load data. The results show that the ICOA-ELM model possesses excellent accuracy and stability in short-term load forecasting. Its R2 and RMSE are 0.9970 and 0.3840 MW respectively, which are much better than ELM and LSTM. Thus, the proposed ICOA-ELM is a viable strategy for short-term load forecasting. [ABSTRACT FROM AUTHOR]

This paper projects the risk associated with social benefit gained by Independent System Operator (ISO) due to unpredictability of load demand and intermittent nature of Renewable Energy Sources (RES). The hybrid method i.e. Wavelet Transform–Auto Regression Integrated Moving Average (WT-ARIMA) is used for forecasting the load in day-ahead market. The load demand for 24 hours is forecasted using the historical hourly data of 2 months. The mean absolute percentage error (MAPE) is calculated for both classical ARIMA and WT-ARIMA model. When compared to the ARIMA model, the WT-ARIMA model has a better fit to the time series data. This paper relates the load forecasting with energy market and shows the impact of forecasting error on market parameters. The system parameters like generation cost, social benefit and power loss are analyzed in day-ahead and real-time market. This work also elaborates the effect of intermittent nature of RES on the system parameters. The optimal location of RES based distributed generation is identified by higher Locational Marginal Price (LMP) based ranking method. The optimal energy management problem is formulated with the aim to minimize the total generation cost taking into account the variability in wind power and solar power. The uncertainty modeling of wind and solar is projected by the Rayleigh and Beta distribution respectively. The risk associated with social benefit of ISO due to uncertainty of wind and solar is calculated using indices, Value at Risk and Conditional Value at Risk. The proposed approach is implemented on IEEE 30 bus system. [ABSTRACT FROM AUTHOR]

Accurate prediction models enable the efficacious utilization and integration of solar energy into the power system. Therefore, this study aims to develop novel hybrid prediction models by employing correlation analysis (CA), decomposition techniques, sample entropy (SE), and spatio-temporal attention (STA) based sequence2sequence (S2S) algorithm for accurate prediction of global horizontal irradiance (GHI). The decomposition techniques include variational mode decomposition (VMD), improved complete ensemble empirical mode decomposition with additive noise (ICEEMDAN), and maximum overlap discrete wavelet transform (MODWT). The VMD-STA-S2S hybrid model surmounts the associated hybrid and standalone prediction models by revealing the highest prediction efficiency and the lowest error. Compared to the SARIMAX, SVR, ANN, XGB, GRU, LSTM, and S2S models, the VMD-STA-S2S model reduced RMSE during testing by 80.927 W/m2, 75.426 W/m2, 73.487 W/m2, 62.394 W/m2, 57.811 W/m2, 52.007 W/m2, and 41.836 W/m2, respectively. Similarly, the reductions in RMSE by VMD-STA-S2S model compared to SA-S2S, TA-S2S, STA-S2S, MODWT-STA-S2S, ICEEMDAN-STA-S2S, ICEEMDAN-SE-STA-S2S, and VMD-SE-STA-S2S models are 24.054 W/m2, 20.951 W/m2, 12.702 W/m2, 15.396 W/m2, 9.921 W/m2, 6.103 W/m2 and 0.484 W/m2, respectively, during testing. Furthermore, considering NSE during testing, the VMD-STA-S2S model is 8.66%, 7.72%, 7.41%, 5.71%, 5.07%, 4.31%, 3.11%, 1.43%, 1.19%, 0.64%, 0.81%, 0.47%, 0.35% and 0.07%, more efficient than the SARIMAX, SVR, ANN, XGB, GRU, LSTM, S2S, SA-S2S, TA-S2S, STA-S2S, MODWT-STA-S2S, ICEEMDAN-STA-S2S, ICEEMDAN-SE-STA-S2S, and VMD-SE-STA-S2S models, respectively. The superior performance of VMD-STA-S2S over its counterparts corroborates the integration of the VMD technique and STA-based S2S algorithm for GHI prediction. The multivariate meteorological data of this study is decomposed by VMD into subcomponents more effectively than the ICEEMDAN and MODWT techniques. VMD decomposed subcomponents are further fed to the STA-S2S to efficiently extract and learn the spatial and temporal features, resulting in the enhanced and superior prediction outcomes of the VMD-STA-S2S model compared to all the counterpart models. Besides GHI prediction, the proposed model is also appropriate for other time-series data, including renewable energy, electrical load, and environment monitoring. [Display omitted] [ABSTRACT FROM AUTHOR]

Rainfall is one of the key triggers of severe weather, such as floods and mudslides. It is difficult to forecast accurately and timely. With the extended application of global navigation satellite system (GNSS) into meteorology, existing methods have used GNSS-derived atmospheric parameters to forecast rainfall. However, low success rate and high false alarm rate exist because only a few related predictors have been used. To overcome the disadvantages of existing rainfall forecasting models, this study proposes a short-term rainfall forecast model based on the support vector machine (SVM) algorithm. GNSS-derived precipitable water vapor (PWV), meteorological parameters (such as: pressure (P), temperature (T), relative humidity (RH), and dew point temperature (DPT)); and time-related parameters, such as day of the year (DoY), hour of the day (HoD), and minute of the hour (MoH); are considered as rainfall forecast parameters. The correlation analysis between meteorological parameters/PWV and rainfall is first performed by using a five-minute PWV time series obtained from GNSS observations in Singapore over the period of 2010 to 2012. SVM is then used to establish the rainfall forecast model and forecast the occurrence of rainfall from 10 to 60 min in advance. Experimental results present that (1) various parameters (T, P, RH, DPT, and PWV) show different variation characteristics before the occurrence of rainfall, thereby indicating that good results can be obtained to describe the process of rainfall occurrence by using multiple parameters and (2) the SVM-based rainfall forecast model can forecast rainfall with and average true (TFR) and false (FFR) forecast rates of more than 99% and less than 23%, respectively. In this study, the SVM-based rainfall forecast model with multiple factors performs better than existing methods. TFR and FFR of the SVM-based rainfall forecast model are improved by nearly 10% and reduced by more than 10%, respectively. [ABSTRACT FROM AUTHOR]

A two-state model for short-term forecasting of PV plants output power (further referred as PV2-state) is proposed. PV2-state connects in an inventive way an empirical estimator for clear-sky photovoltaic (PV) power output with a statistical predictor for the sunshine number, a binary indicator stating whether the Sun shines or not. PV2-state shows remarkable features: (1) accessibility (only data series resulted from the PV plant monitoring are processed), (2) universality (no physical models for PV plant components are required) and (3) high-performance (due to a continuous adaptation to the actual atmospheric and PV modules conditions). Based on different error metrics, the model performance is investigated from three perspectives: forecast accuracy, forecast precision and the response to the variability in the state-of-the-sky. The study was conducted with high-quality data collected from a fully monitored experimental micro-PV plant developed on the Solar Platform of the West University of Timisoara, Romania. By processing information about the actual performance of the PV plant, PV2-state proves a notable advance in the forecast precision, becoming a robust competitor in the race for high-accuracy in intra-hour forecast of PV power. • A new two-state model (PV2-state) for nowcasting PV power is proposed. • PV2-state is characterized by simplicity and a wide accessibility. • Tests performed against measured data show high-quality forecasts. [ABSTRACT FROM AUTHOR]

Developing an appropriate model for accurate prediction of energy consumption is very essential for developing an effective energy management system for residential buildings. In view of this, the Short-term Load Forecasting (STLF) of household appliances has been performing an important role in supervising and managing energy in the residential community. In the domain of big data analytics, data-driven load forecasting approaches have realized an amazing performance in the recognition of patterns of residential electric loads and forecasting energy consumption. Nevertheless, current research emphasizes the use of powerful feature-engineering methods, which are ineffective and result in low generalization performance. Further, considering the differences in the consumption behavior of various home appliances, it is unfeasible to discover energy consumption characteristics physically in the power system. Thus, this study addresses the problems of STLF using a novel two-stream deep learning (DL) model called STLF-Net. The first stream is designed with Gated Recurrent Units (GRUs) to learn and capture the long-term temporal representations of the energy utilization data. Simultaneously, in the second stream, the short-term information and positional representations are modeled using a stack of temporal convolutional (TC) modules. The TC module is designated using dilated causal convolutions and residual connection to enable efficient feature extraction while alleviating the gradient vanishing issues. The learned representations from the two streams are fused and subsequently passed to several dense layers to generate the final hour-ahead load forecasts. Experimental assessments on two public energy consumption predictions datasets (IHEPC and AEP) demonstrated the superior performance of the STLF-Net over the recent cutting-edge data-driven approaches. [ABSTRACT FROM AUTHOR]

Short-Term Load Forecasting is of great significance for the economic and stable operation of the power system. Against the background of the breakthrough in generative artificial intelligence based on the Diffusion model, the research on relevant load forecasting methods of the latter is still relatively limited. Therefore, this paper refers to many related excellent works, analyzes the commonalities between image generation tasks and load forecasting tasks, and proposes the Diffusion-based Inpainting Forecasting Method (DIFM). DIFM supports multi-variable inputs and can achieve functions such as load sequence generation, quantile forecasting and missing data imputation, making it a flexible and multifunctional method. The feasibility and performance of this method are validated across multiple datasets, with experimental results revealing that DIFM reduces the mean absolute percentage error by 24.61 % and 17.91 % respectively in short-term load forecasting and load imputation tasks compared to the optimal benchmark models. • Delved into the parallels between forecasting tasks and image inpainting tasks. • Incorporated the Diffusion model into the realm of short-term load forecasting. • Investigated various approaches to condition embedding. • The proposed method achieved generation, forecasting, and imputation. [ABSTRACT FROM AUTHOR]

Building load prediction plays an important role in building energy savings and mechanical and electrical system optimization control. The dynamic energy consumption can be accurately calculated using the traditional physical energy simulation method that entails a complex setup and verification process due to the numerous input parameters required. It is also difficult to change the physical model once it has been determined. The data-mining method is fast in calculation and simple to use, but its prediction accuracy is limited by historical data quality. It is difficult to predict the load of new buildings without historical data. To solve these problems, this study proposes a hybrid building load prediction method for office buildings. The proposed method uses EnergyPlus to generate a building load database that includes than 25.14 million data cases, covering 35 types of building geometry and 2870 building examples. Based on the above database, the LightGBM algorithm was selected to extract feature variables that affect the load and build a load prediction model. The training results show that there are 24 key feature variables for office building load prediction. The hourly MAPE of the cooling load prediction model is 6.95 % and RMSE is 4.31 W/m2, and the hourly MAPE of the heating load prediction model is 7.09 % and RMSE is 11.64 W/m2 compared with EnergyPlus model. Two actual office buildings are selected as case studies to validate the model prediction accuracy. Results show that comparing predicted results with measured data, the hourly cooling load of the MAPE is 12.42 %. Coparing predicted results with actual heating load, daily MAPE is 7.97 %. • A case-rich office load simulation database over 25.14 million pieces is generated and extensible. • A high-precision load forecasting method based on LightGBM for office buildings is proposed. • Eliminate the geometric constraints imposed by simulation model and the reliance on historical data of data-driven model. • The method can be applied to other types of buildings by modifying the parameters. [ABSTRACT FROM AUTHOR]

Accurate characterization of the dynamics of wind power systems under highly unstable conditions is essential for short-term wind power forecasting. Multivariate factors, such as wind speed, wind direction and climatic factors at different heights, as well as the relationships between them, need to be fully considered. To this end, after the Pearson correlation coefficient is used to identify the main influencing variables, the empirical modal decomposition (MEMD) method is utilized to decompose the data in order to capture the deep coupling relationship and the spatio-temporal pattern of change among the variables. Combined with a bidirectional long and short-term memory network (BiLSTM) optimized through the RIME algorithm, a short-term wind power prediction model was constructed. Data analysis of two wind farms in Xinjiang and Gansu, China, shows that the model has higher prediction accuracy compared to the baseline model. The effectiveness of RIME and MEMD is verified through component ablation experiments and comparison experiments with other optimization algorithms and signal decomposition algorithms. To cope with large-scale data, dimensionality reduction of IMF features using Elastic Net (EN) is explored to reduce computational requirements and maintain good prediction performance. The method provides strong support for power system scheduling and efficient utilization of renewable energy. • RIME-MEMD-BiLSTM hybrid model for precise wind power prediction. • MEMD excelled in multivariate data decomposition. • Explored wind power data's temporal and seasonal traits. • Validated RIME's superior hyperparameter optimization. • High-dimensional feature dimensionality reduction with EN. [ABSTRACT FROM AUTHOR]

Short-term load forecasting is vital for efficient power system planning. To better extract features expanded by the signal decomposition method and enhance the performance of load prediction models, this paper proposes an iterative Boruta feature extraction method based on variational modal decomposition (VMD-IB) and a bi-directional long short-term memory network assisted by convolutional neural network and self-attention mechanism(CNNA-BiLSTM). Firstly, VMD decomposes the load data to capture different patterns and scales. An iterative Boruta method is proposed to identify the most valuable features for load prediction through an adaptive iterative strategy during feature selection. VMD-IB shows an advantage in extracting correlated features at multiple scales, enabling the model to better capture complex load fluctuations. Secondly, CNNA-BiLSTM enhances the ability of BiLSTM to fit the nonlinear data using CNN for temporal feature extraction and SAM for attention and weight allocation, improving the feature expressiveness and prediction performance. Lastly, extensive experiments on New York Independent System Operator(NYISO) data validate the framework. Compared to the state-of-the-art method, the proposed method achieves a 12% improvement in mean absolute percentage error and an 8% improvement in the Theil inequality coefficient, showcasing the superiority of the proposed framework in accurately predicting short-term load demand. • Capturing multi-scale patterns and features using VMD. • Adaptive selection of key features through Boruta iteration. • CNNA-BiLSTM enhances nonlinear data fitting and prediction. • Proposed model effectively improves the prediction accuracy. [ABSTRACT FROM AUTHOR]

• This paper combines deep learning and reinforcement learning. • Adaptive hyperparameter optimization based on deep reinforcement learning algorithm. • The prediction ability of the proposed model is stronger than that of pure GRU. • Polar coordinate encoding method is used for date data that affects load. • The prediction ability of the proposed model outperforms the state-of-the-art method. Short-term load forecasting (STLF) is critical to optimizing power system operation. Deep learning (DL) methods can provide extremely high accuracy for STLF. However, most models in existing research lack adaptive optimization capabilities in the prediction process and suffer from performance degradation. To resolve the above difficulties, we propose a hybrid model (DDPG-GRU) based on gated recurrent units and deep deterministic policy gradients for STLF. First, the GRU network has the advantage of processing multiple time series inputs and can simultaneously consider multi-dimensional load characteristics, thereby making the model more efficient. Since the GRU model structure is relatively complex, choosing a good set of hyperparameters is very difficult. Therefore, the purpose of using DDPG is to optimize the hyperparameters of the GRU model adaptively. The proposed model is a combination of DL methods and reinforcement learning. In order to prove the superiority of the proposed model, it is applied to the load data of Area 1 in China to perform single-step and multi-step load forecasting, respectively. The results show that DDPG-GRU has a better fitting effect than the baseline method. Taking the multi-step prediction results as an example, compared with the classic GRU network, the MAPE, MAE, and RMSE of the proposed model are reduced by 22.75 %, 14.44 %, and 14.02 %, respectively, while the R 2 coefficient is increased by 13.23 %. At the same time, we use the China Area 2 data set to verify the universality of the proposed model. Furthermore, we compared the proposed method with state-of-the-art methods and achieved better accuracy. [ABSTRACT FROM AUTHOR]

• A multi-step forecasting method combining long and short interval time series data is proposed. • A one-dimensional convolutional structure is designed. • A hybrid CNN-LSTM-AM model outperforms a single model on the public ACN-dataset. • Our method performs short-term electricity load forecasting up to 96 steps in the ACN dataset. Electric vehicle (EV) load forecasting is becoming increasingly important for power system operation. Accurately multi-step-ahead forecasting EV loads is challenging. The correlation between the series at different time intervals and the key points in forecasting the time series will affect the results of EV load forecasting. Therefore, in this paper, a method is presented for the combination of time series of different length intervals into a hybrid CNN-LSTM-AM model for multi-step-ahead forecasting. The input matrix consists of combining time series of different lengths. A designed CNN network with a one-dimensional convolutional structure is used to extract features. After the convolutional layer, the temporal features remain. Finally, LSTM Encoder-Decoder and Attention Mechanism (AM) are combined to solve the problem of forgetting multi-step-ahead forecasting. Through the validation of the public ACN-data, it is demonstrated that the proposed method achieve accurate prediction results. According to error metrics, MAE, RMSE and R2 outperform other models with a value of 0.5268, 0.9519 and 0.9138 respectively. The maximum number of multi-step-ahead prediction reaches 96 steps. This provides a reference for longer multi-step predictions in the future. It is also confirmed in the ACN-data that the accuracy of the hybrid model is better than the single model in EV load prediction. [ABSTRACT FROM AUTHOR]

• Addressed the challenge of load forecasting in IES with limited data. • Multi-task learning effectively extracts multi-energy coupling relationships. • The two-stage gradient descent optimizes training and enhances prediction efficiency. • The proposed MTL-NNGP model reaches a mean accuracy of 97.93% in only 6.28 s. Multi-energy load forecasting forms the foundation of the operation and scheduling of integrated energy systems. Nevertheless, insufficient data and underutilization of the coupling relationship between the multi-energy load limit the accuracy of load forecasting. This paper presents a predictive model combining neural network Gaussian processes and multi-task learning. The approach is tailored to enhance forecasting accuracy in environments with small-sample datasets. This model capitalizes on the advantageous properties of infinitely wide neural networks for handling small-sample data. Simultaneously, the model effectively extracts the interconnected dynamics of cooling, heating, and electricity loads within the integrated energy system through multi-task learning. In addition, the model applies concrete dropout, enhancing robustness to irregular loads while maintaining the synergistic benefits of the multi-task framework. Furthermore, this paper employs a two-stage gradient descent approach to replace kernel matrix computations of Gaussian processes, reducing the computational cost of parameter optimization and yielding superior forecasting performance in shorter training durations. The simulation results indicate that the proposed model attains a mean accuracy of 97.93% for a 3-day forecasting horizon. Compared with alternative forecasting models, this model exhibits higher accuracy and enhanced generalization capabilities in multi-energy load forecasting for small-sample integrated energy systems. [ABSTRACT FROM AUTHOR]

An accurate multivariate load combination forecast is essential for the planning of the energy dispatch and operation of integrated energy systems. An integrated decision model for multivariate load combination forecasting of integrated energy systems is proposed to address various problems of multivariate load combination forecasting of integrated energy systems, which is capable of evaluating the advantages and disadvantages of different forecasting methods under the same forecasting scenario. First, historical load data, meteorological factors, calendar rules, load combination indicators and integrated meteorological indicators are correlated using Pearson's correlation coefficient. Then, input features are extracted. Subsequently, four single prediction models and three combination prediction models, including a relatively novel combination prediction model of mathematical computation, were employed to predict and preliminarily analyze the electricity, cooling, and heating loads of the integrated energy system. Finally, the three loads are assigned using Best-Worst Method combined with Entropy Weight Method combinatorial weighting model, and two comprehensive evaluation indices are reconstructed and integrated decision making for the seven models is performed by considering the three loads comprehensively. The case study demonstrates that the integrated root mean square error of the adopted mathematical computational combination forecasting model in winter and summer is 7.03 and 4.71, respectively, and the integrated mean absolute percentage error is 0.94 and 1.2, respectively, which is significantly higher than that of the other models in terms of forecast accuracy. • A newer mathematical computational combinatorial prediction model is validated. • An integrated decision-making model for multivariate load combination forecasting of integrated energy systems is proposed. • Established the best-worst-entropy subjective-objective combination of empowerment model. • Seven models for electricity, cooling and heating load forecasting. [ABSTRACT FROM AUTHOR]

Electricity price prediction is essential for the optimal dispatch in power markets, with accurate prediction models being critical for efficient power system operations and market trading decisions. Deep learning networks, with their powerful nonlinear modeling capabilities, have shown promising results in electricity price forecasting. However, their design techniques, especially the selection of network parameters, remain challenging. This indicates that the optimization and exploration of deep learning networks in electricity price forecasting models require further investigation. This paper innovatively proposes a forecasting model that uniquely integrates Variational Mode Decomposition (VMD), Grey Wolf Optimization (GWO), Attention Mechanism (ATT), and Long Short-Term Memory Network (LSTM), optimizing the model from three different perspectives. First, during the data preprocessing phase, the training set is subjected to VMD to reduce noise, thereby enhancing the capture of multi-scale characteristics inherent in electricity price time series. The ATT layer is integrated to adaptively allocate weights, enhancing the model's focus on significant features. The GWO is applied to optimize hyperparameters of the LSTM, accelerating convergence and improving iteration accuracy, thereby reducing model error. A series of experiments were conducted using multiple regional electricity price datasets, evaluated with several metrics including RMSE. The results validated the effectiveness of the proposed three modules in improving the performance of the time series network, with VMD making the most significant contribution. Among all models, VMD-GWO-ATT-LSTM consistently outperformed others, demonstrating its effectiveness and robustness in electricity price forecasting. [ABSTRACT FROM AUTHOR]

Medium-term feeder load forecasting plays a pivotal role in efficiently operating and planning electrical distribution systems. It provides valuable insights into future electricity demand trends, enabling utilities to make informed decisions regarding infrastructure upgrades, resource allocation, and energy management strategies. However, few studies have been done on medium-term load forecasting for the distribution network's operational planning at the medium voltage level. Moreover, conventional load forecasting techniques mainly consider the impact of limited external factors, which is typically challenging to forecast accurately. In this work, multiple influential features have been utilized for accurate medium-term load prediction. Accurate load forecasting is paramount for efficient operation and planning in power systems. This study proposes a novel approach for medium-term feeder load forecasting that enhances peak accuracy using the Prominence-guided Weighted Peaks (PWP) in conjunction with the eXtreme Gradient Boosting (XGBoost) model. To evaluate the performance of the proposed model, we conduct experiments using real-world load data from a distribution feeder. Comparative analysis with traditional forecasting methods demonstrates the superior accuracy of the PWP-XGBoost model, particularly in predicting peak loads. Enhanced peak accuracy is crucial for utilities to effectively manage peak demand, optimize resource allocation, and ensure grid stability. • One year-ahead load prediction at the feeder level using multiple machine learning models. • Uses multiple influential features for prediction and observing the explainability of the model to understand the features in prediction. • Proposed PWP-XGBoost model to improve peak accuracy of medium-term load forecasting. • Our proposed method aided in improving the peak accuracy up to 5%. [ABSTRACT FROM AUTHOR]

Efficient and accurate multi-energy load forecasting is crucial for achieving supply–demand balance in regional integrated energy systems(RIES). Existing research primarily focuses on modeling the coupling between different loads, with limited consideration of modeling from the perspective of sequential temporal information. Excessive emphasis on load coupling can lead to reduced accuracy in long-term load forecasting for RIES. To address this issue, this paper proposes a trend and periodicity feature-enhanced decoupling prediction framework to fully capture the long-term dependency relationship within sequences. Firstly, load data usually has significant trends and seasonal variations, and by decomposing load data into trend and seasonal features, it is possible to forecast for different patterns of change. Next, a decoupling prediction framework is proposed that adequately considers sequential temporal information, trend information, and seasonal information. The proposed framework enables accurate predictions for electricity load, cooling load, and heating load. Finally, comprehensive and extensive validation of the proposed prediction method is conducted using actual data from RIES. The comparative results demonstrate that the method proposed in this paper outperforms other Transformer-based model algorithms, with average improvements of 20.8%, 24.5%, and 21.7% in the evaluation metrics for electricity load, cooling load, and heating load forecasting, respectively, indicating superior predictive accuracy and wider applicability. • A novel multi-branch prediction framework based on time information modeling is proposed, which allows the model to focus more on the temporal information of the sequences. • A novel load coupling modeling approach based on time steps is proposed, which effectively captures the coupling relationships among sequences and the local temporal dependencies within the sequences. • A novel global convolution method based on multi-scale sub-kernels is proposed to capture the global temporal dependencies of the sequences and complement the local temporal dependencies. • By comparing the model's performance on real-world test cases, it is demonstrated that the proposed model exhibits superior performance in long-term multi-energy loads forecasting. [ABSTRACT FROM AUTHOR]

The cyclical behavior of electricity users facilitates the prediction of future electricity loads through the analysis of historical data, in addition to considering the spatial correlation among various electricity units. Building on these observations, this study transforms the power loads generated by regional-level user units into graph signals for a more effective visualization of spatial–temporal electricity consumption patterns. Specifically, the graph-based model harnesses its unique spatial–temporal dependencies to deduce the dynamics of electricity consumption patterns and enhance the precision of electricity generation predictions. Introducing a novel deep learning model, the Spatial-Temporal Fusion Graph Convolutional Network (FSTGRNN), which integrates temporal and spatial correlations for the prediction of electricity load values at future time steps. FSTGRNN is an end-to-end model that, firstly, delineates the spatial and temporal patterns of electric loads by constructing spatial relationship graphs using the FC-LSTM and GC-GRU models designed within the spatial–temporal block (ST-Block). Secondly, the spatial–temporal fusion block (STF-Block) in FSTGRNN, equipped with a gating mechanism, encodes the graph signals to capture both long-term dependencies and short-term correlations. The resulting integrated representation is utilized for the forecasting process. Experimental validation is conducted on two electricity load datasets, and the forecasting results clearly demonstrate the superiority and efficiency of the proposed deep learning system. • Recognizing the regional level spatial–temporal characteristics of power loads. • Proposed a novel forecasting approach utilizing spatial–temporal fusion graph neural networks. • Provides an implementable strategy for training reliable load forecasting model in spatial–temporal scenarios. • Provides a foundation for modeling and decision making in production and manufacturing system. [ABSTRACT FROM AUTHOR]

Effective short-term wind power prediction is crucial to the optimal dispatching, system stability, and operation cost control of a power system. In order to deal with the intermittent and fluctuating characteristics of the wind power time series signals, a hybrid forecasting model is proposed, based on Complementary Ensemble Empirical Mode Decomposition (CEEMD) and Whale Optimization Algorithm (WOA)-Kernel Extreme Learning Machine (KELM), to predict short-term wind power. Firstly, the non-stationary wind power time series was decomposed into a series of relatively stationary components by CEEMD. Then, the components were used as the training set for the KELM prediction model, in which the initial values and thresholds were optimized by WOA. Finally, the predicted output values of each component were superimposed, to obtain the final prediction of the wind power values. The experimental results show that the proposed prediction method can reduce the complexity of the prediction with a small reconstruction error. Furthermore, performance is greater, in terms of prediction accuracy and stability, with lower computational costs than other benchmark models. [ABSTRACT FROM AUTHOR]

The wind is a natural source of energy and wind energy occupies an important share in the global energy structure. Compared with offshore wind energy, the economy and availability of onshore wind energy make it dominate the wind energy industry at this stage. However, the instability and inconsistency of the ultra-short-term onshore wind speed will cause inefficiency with the turbines. Therefore, accurate onshore wind speed forecasts can help onshore wind farms enhance the efficiency of wind turbines and improve the accuracy of energy assessment. Most existing forecasting methods usually suffer from insufficient precision and high complexity. To improve these deficiencies, a multi-component hybrid onshore wind speed forecasting system based on predictability recognition framework, κ point modified multi-objective golden eagle optimizer, and weight hybrid kernel extreme learning machine is proposed in this study. According to the experiments of four onshore wind speed series collected from an onshore wind farm located in the US, the proposed system shows superior performances on one-step ahead and multi-step ahead forecast and it outperforms some typical methods. And the statistical significance of superior forecasting performance is fully demonstrated. Overall, the proposed forecasting system can offer a more reliable forecast for onshore wind speed. [ABSTRACT FROM AUTHOR]

• Data-driven finite element modeling using universal motion concept (UMC) machine settings. • New dynamic loaded tooth contact analysis (DLTCA) focusing on local loaded contact effect. • Geometric boundary forecasting to dynamic meshing impact by using DLTCA. • Forecasting including transmission error, torque, angular velocity and acceleration. • Semi-FEM dynamic meshing impact forecasting model for spiral bevel and hypoid gear transmission. Focusing on the accurate size, direction and position of dynamic meshing impact, an innovative semi-finite element method (semi-FEM) forecasting model is proposed for spiral bevel gear transmission under both high speed and heavy load conditions. To distinguish with the traditional rotor dynamic analysis focusing on vibration effect for global system, a new dynamic loaded tooth contact analysis (DLTCA) focusing on loaded contact effect for local contact area is proposed for the geometric boundary forecasting. Where, DLTCA shows a systematically time-varying characteristics including the dynamic loaded contact pattern, pressure, transmission error. Considering tooth flank flexural behaviors, data-driven tooth flank simulation and finite element modeling is developed by using universal motion concept (UMC) machine settings. At each finite element in a local geometric boundary, the dynamic compatibility and equilibrium equations are established as a semi-FEM dynamic meshing impact forecasting model. Moreover, for a dynamic transmission system, a basic judgement criterion is firstly developed. Then, the improved simplex method is applied to solve the established model for the dynamic meshing impact size evaluations including transmission error, torque, angular velocity, and acceleration. Finally, numerical and test instances are provided to verify the proposed method. [ABSTRACT FROM AUTHOR]

• Incorporation of geographical and socioeconomic data for MSW forecasting. • Modification of kNN algorithm to include exogenous data in a hierarchical setting. • Providing forecasts for comparatively small or incomplete data sets. • Demonstration of MSW forecasting and its composition for Lithuanian municipalities. Forecasting municipal solid waste (MSW) generation and composition plays an essential role in effective waste management, policy decision-making and the MSW treatment process. An intelligent forecasting system could be used for short-term and long-term waste handling, ensuring a circular economy and a sustainable use of resources. This study contributes to the field by proposing a hybrid k-nearest neighbours (H-kNN) approach to forecasting municipal solid waste and its composition in the regions that experience data incompleteness and inaccessibility, as is the case for Lithuania and many other countries. For this purpose, the average MSW generation of neighbouring municipalities, as a geographical factor, was used to impute missing values, and socioeconomic factors together with demographic indicator affecting waste collected in municipalities were identified and quantified using correlation analysis. Among them, the most influential factors, such as population density, GDP per capita, private property, foreign investment per capita, and tourism, were then incorporated in the hierarchical setting of the H-kNN approach. The results showed that, in forecasting MSW generation, H-kNN achieved MAPE of 11.05%, on average, including all Lithuanian municipalities, which is by 7.17 percentage points lower than obtained using kNN. This implies that by finding relevant factors at the municipal level, we can compensate for the data incompleteness and enhance the forecasting results of MSW generation and composition. [ABSTRACT FROM AUTHOR]