Search

The dynamic load in the soil directly leads to the damage of underground structures upon explosions. In this study, a method to predict blast load on underground structure surface based on the neural network was developed to study the load distribution under close-in detonation. First, taking the underground utility tunnel as the experimental structure, 52 groups of field blast tests were conducted on the surface load mechanism, and the surface load data samples were obtained. Second, the key influencing parameters of the reflected blast load were obtained through the dimensional analysis method, and the backpropagation neural network model was constructed based on the test data using the Levenberg–Marquardt algorithm to train and optimize the neural network. Finally, the accuracy of load prediction results was compared and evaluated among the neural network, empirical formula, and nonlinear regression analysis (NRA) methods. It is found that the input parameters of combined variables can further improve the prediction accuracy of the neural network compared with the input parameters of single physical variables. Compared with the empirical formula method and the NRA method, the neural network model with input parameters of combined variables provided the most accurate prediction. The load distribution under typical conditions calculated by the neural network showed that the explosive setting parameters impact the uneven shape of blast load on the structure surface. The increase in explosive equivalent and depth reduces the nonuniformity of load distribution, while the decrease in explosion distance increases the nonuniformity of load distribution. [ABSTRACT FROM AUTHOR]

Short-term power load forecasting provides important guidance for the improvement of power marketing and control levels of power enterprises. In this paper, a novel method, named RF-TStacking, is proposed to forecast the short-term load. This study starts from the influence factors of the power load, the random forest is applied to estimate the importance of the influence factors of short-term load. Based on Stacking strategy, the integration of LightGBM and random forest is realized to achieve short-term power load forecasting. To improve the generalization ability of the load model, random put back sampling is used to sample each primary learner, and the average value is taken as the result of each primary learner. The Bayesian optimization is used to adjust the super parameters of the model to improve the accuracy of the selection of influencing factors. The load data of a region in northwest China are used for the testing, and it is found that the model can provide stable prediction results. [ABSTRACT FROM AUTHOR]

With the increasing proportion of solar grid-connected, the establishment of an accurate photovoltaic (PV) power prediction model is very important for safe operation and efficient dispatching of a power grid. Considering the multi-level periodicity of PV power caused by many factors, such as seasons and weather, a short-term PV power prediction model based on transfer component analysis is designed by introducing the idea of transfer learning. In order to measure the uncertainty of numerical weather prediction (NWP) and power sequence, a novel algorithm considering weather similarity and power trend similarity is proposed. First, the intrinsic trend is measured by extracting permutation entropy, variance, and mean from the historical PV power sequence. Second, weighting of NWP is accomplished based on the Pearson correlation coefficient. PV power data are divided into different clusters by K-medoids clustering. At the same time, the transfer component analysis alleviates the time-varying problem of data distribution caused by multi-level time periodicity and effectively improves the prediction accuracy of the model. Finally, simulation experiments are carried out on the PV power output dataset (PVOD). The results show that the prediction accuracy of the proposed method is better than the traditional methods, and the accuracy and applicability of the proposed method are verified. [ABSTRACT FROM AUTHOR]

Short-term wind power forecasting is crucial for power system stability, dispatching, and cost control. Wind energy has the potential to be a viable source of renewable energy. Wind power generation forecasting is vital for resolving the supply and demand challenges of the smart grid. Moreover, one of the most problematic aspects of wind power is its high fluctuation and intermittent nature, which makes forecasting difficult. The goal of this research is to create machine learning models that can properly estimate wind power production. Significantly, the major contributions of this work are highlighted in the following significant elements. First, a data analysis framework for visualizing the gathered dataset from the SCADA system is presented. Second, for forecasting wind power time-series dataset values, we examine the predicting performance of various machine learning models using various statistical indices. The experimental findings demonstrate that with a minor reconstruction error, the proposed forecast approaches can minimize the complexity of the forecasting. Furthermore, in terms of forecast accuracy, a gradient boosting regression model outperforms other benchmark models. According to the analysis, our methodology might be applied in real-world circumstances to assist the management group in regulating the power provided by wind turbines. [ABSTRACT FROM AUTHOR]

Aiming at the problem that the traditional wind power forecasting is difficult to deal with a large amount of strong volatility data and limited processing capacity of time series, a wind power forecasting method based on multi-model combination under stacking framework was proposed. First, the wind turbine data are cleaned by density-based spatial clustering of applications with the noise clustering method. Considering the differences of data observation and training principles, the proposed stacking method embedded multiple machine learning algorithms to utilize their diversified strength. The stacking base-learner includes the CBLSTM model, which has the advantages of deep architecture feature extraction, and takes into account data timing and nonlinear relationship as well as XGBoost and other tree ensemble learning models that were suitable for complex data modeling. The feasibility of the algorithm was verified by using the actual wind power data of two wind farms in Northeast and Western China. Experimental results show that the stacking ensemble learning method proposed has better forecasting performance and stability than other single forecasting models, which is of great significance to guide wind power dispatching operation and improve wind power consumption capacity. [ABSTRACT FROM AUTHOR]

In accordance with the topographic features and other relevant factors, data affecting wind power generation in engineering practice may be difficult to obtain, and three short-term wind power forecasting methods are proposed based on incomplete data. Most wind power forecasting studies are based on wind speed data, but wind power forecasting needs more influential factors in addition to wind speed data, and it is difficult to obtain these data in actual production and life. Therefore, this paper proposes the wind power forecasting under the condition of incomplete data. First, this paper makes theoretical analysis on multi-state space Markov chain wind power forecasting, backpropagation (BP) neural network wind power forecasting, and genetic algorithm (GA)-BP neural network wind power forecasting, and the corresponding wind power forecasting models are constructed. Second, with the actually measured data of a wind farm in the Heilongjiang province as the research object, the historical data are preprocessed first and then imported into three wind power forecasting models for simulation, and the curves of relative error and absolute error of total wind power forecasting in this area are obtained. Finally, the simulation results of three methods based on incomplete data are analyzed and evaluated. The results show that under the condition of incomplete data, the maximum error of the GA-BP neural network wind power forecasting model improved by the genetic algorithm is reduced from 6.8% to 1.6%, and the forecasting accuracy is greatly improved. [ABSTRACT FROM AUTHOR]

As a clean, economical, and renewable energy source, wind energy plays a very important role in easing the shortage of fossil energy, environmental population, and climate change. However, due to the strong intermittency, volatility, and randomness of wind speed, the large-scale connection of wind energy into the power grid is restricted. Therefore, constructing a reliable prediction model to achieve high-accuracy wind speed prediction is necessary. For this purpose, a novel hybrid model for short-term wind speed prediction is proposed in this paper. First, empirical mode decomposition is employed to decompose the raw wind speed time series into a set of subseries. Then, a data augmentation technique is first used to generate more training data to avoid overfitting of the prediction model. Furthermore, a new predictor based on a convolutional neural network (CNN), a long short-term memory (LSTM) network, and an extreme learning machine (ELM) is proposed for deterministic wind speed prediction, where a fuzzy entropy-based partition strategy is implemented to assign subseries to the CNN-LSTM and ELM. To improve the prediction performance, a synchronous optimization method based on an improved hybrid particle swarm optimization/gray wolf optimizer is proposed for feature selection and parameter optimization. Afterward, kernel density estimation is used to estimate the wind speed probability density function for probabilistic prediction. Finally, the performance of the proposed model is compared with seven other models by using three wind speed datasets from four aspects: point prediction, interval prediction, probability prediction comprehensive performance, and prediction reliability. The experimental results show that the proposed method achieves excellent performance on wind speed time series prediction. [ABSTRACT FROM AUTHOR]

High-precision wind power prediction is an important method to ensure the safety and stability of wind power integration. However, because of the intermittent and uncontrollable nature of wind speed sequences, the wind power prediction has become a difficult task. Given this, this paper studies the prediction of wind power from three aspects. First, combining a similar day with long short-term memory is proposed to predict ultra-short-term wind power, and the improved gray correlation analysis method is used to select a similar day. In addition, to study the influence of weather data on the accuracy of the prediction model, a univariable prediction model and a multivariable prediction model are proposed to predict ultra-short-term wind power, and their performances are compared. The experimental results show that all of the above studies help improve ultra-short-term wind power prediction accuracy. Finally, the uncertainty prediction (confidence interval) of wind power is estimated by the nonparametric kernel density estimation based on the Bootstrap-Kernel density method on the result of deterministic prediction, and the upper and lower limits of wind power fluctuation are given at a certain level of confidence. The research results can provide decision-makers with accurate data changes in risk analysis and reliability assessment. [ABSTRACT FROM AUTHOR]

Efficient and accurate photovoltaic (PV) modeling is significant for ensuring the stable operation of the power grid. The nonlinear output characteristics of PV modules are usually described by current–voltage (I–V) curves or power–voltage (P–V) curves. Traditional models normally include a diode model and an analytical model. The process of model parameter's extraction is complex and inconvenient because the equations of the diode model and analytical model are non-linear. In traditional extraction methods using a neural network, all irradiance training data are input into the neural network, which suffer from long training time and insufficient accuracy. In this paper, two neural networks with different training ranges are proposed to replace the whole neural network for predicting I–V curves, P–V curves, and maximum power of the analytical model. Each neural network consists of three layers. In detail, the input is the value of solar irradiance and module temperature, and the output is the shape parameters of the analytical model. The effectiveness of the proposed method is verified by massive experimental data of different PV modules. The output performance estimation of the proposed method is better than that of other methods. [ABSTRACT FROM AUTHOR]

A new method is proposed for ultra-short-term prediction of photovoltaic (PV) output, based on an LSTM (long short-term memory)-ARMA (autoregressive moving average) combined model driven by ensemble empirical mode decomposition (EEMD) and aiming to reduce the intermittency and uncertainty of PV power generation. Considering the superposition of the overall trend and local fluctuations contained in the PV output data, an EEMD adaptive decomposition criterion based on continuous mean square error is proposed to extract the various scale components of the PV output data in the time–frequency domain; an ARMA (autoregressive moving average) model suitable for short correlation analysis is constructed for the intrinsic mode function components that characterize local fluctuations of PV output. Environmental parameters such as solar radiation, temperature, and humidity are introduced to construct a LSTM prediction model with autocorrelation capability and environmental characteristics for the EEMD residual that characterizes the overall trend of PV output. Finally, the overall trend and the local fluctuation forecast results are fused to realize an ultra-short-term forecast of PV output. The training set and test set were randomly selected from the PV microgrid system of Hangzhou Dianzi University and used for PV output prediction according to different seasons and weather types. The maximum MAPE on sunny, cloudy, and rainy days was 23.43%, 32.34%, and 33.10%, respectively. The minimum MAPE on sunny, cloudy, and rainy days was 5.53%, 6.47%, and 19.19%, respectively. The results show that the prediction performance of this method is better than traditional models. The ultra-short-term forecasting method for PV output proposed in this paper can help us to improve the safety, flexibility, and robustness of PV power systems. [ABSTRACT FROM AUTHOR]

Intermittent electrical power output from grid-connected solar farms causes intermittent and uncertain requirements for dispatchable power to balance power supply and demand. Accurate forecasting of electrical power output from solar farms can improve managing power generators connected to the grid. To forecast the electrical power output, a time series model is developed for two solar farms in Australia. The forecast model consists of a Fourier series that models seasonality and an autoregressive moving-average component that models the difference between the observed electrical power outputs and the Fourier series. Persistence detection is added to the model to improve forecast performance on clear days. Using minutely data, the model forecasts the electrical power output seven minutes ahead at every five-minute interval to comply with the requirements of the Australian Energy Market Operator (AEMO). Based on a 30-day testing period, the normalized mean absolute error (NMAE) skills of the time series model are 10.9% and 13.2% lower than those of the clear sky index persistence (CSIP) model. However, the normalized root mean squared error (NRMSE) skills of the time series model are approximately 3% and 12% higher than those of CSIP and the model currently used by AEMO, respectively. As the NRMSE skills are more indicative than the NMAE skills in reducing large forecast errors that would reduce electricity grid stability, the results suggest that AEMO can improve the management of the electricity grid with an inexpensive tool by adopting the developed model to forecast electrical power output of solar farms. [ABSTRACT FROM AUTHOR]

Accurate wind speed forecast plays an important role in the safe and stable operation of large-scale wind power integrated grid system. In this paper, a new hybrid model for short-term wind speed forecasting based on hyperparameter optimization and error correction is proposed, where the forecasting period is 5, 10, and 15 min, respectively, for three sites. The empirical wavelet transform is used to decompose the original wind speed series. Then, the Elman neural network and kernel extreme learning machine, which adopt Bayesian optimization algorithm for hyperparameter optimization, are used as predictors for wind speed prediction and error processing, respectively. In addition, a new error correction model using wind speed as model input is proposed. In order to verify the performance of the proposed model, three datasets collected from different real-world wind farms in Gansu and Xinjiang were considered as a case study to comprehensively evaluate the prediction performance of ten forecasting models. The results reveal that the proposed model has higher prediction accuracy and better prediction performance than the contrast models. [ABSTRACT FROM AUTHOR]

A deep-learning-based hybrid strategy for short-term load forecasting is presented. The strategy proposes a novel tree-based ensemble method warm-start gradient tree boosting (WGTB). Current strategies either ensemble submodels of a single type, which fail to take advantage of the statistical strengths of different inference models, or simply sum the outputs from completely different inference models, which does not maximize the potential of the ensemble. Inspired by the bias-variance trade-off, WGTB is proposed and tailored to the great disparity among different inference models on accuracy, volatility, and linearity. The complete strategy integrates four different inference models of different capacities. WGTB, then, ensembles their outputs by a warm-start and a hybrid of bagging and boosting, which lowers bias and variance concurrently. It is validated on two real datasets from the State Grid Corporation of China of hourly resolution. The result demonstrates the effectiveness of the proposed strategy that hybridizes the statistical strengths of both low-bias and low-variance inference models. [ABSTRACT FROM AUTHOR]

This Focus Issue brings together recent notable advances in the field of dynamical systems and complexity science on the subject of the non-linear collective behavior of power grids and energy systems. SUMMARY OF CONTENTS A. Multistability and bifurcation structure From a dynamical systems point of view, the question of power grid stability is whether the desired operational point of the power system is an attractor of the dynamical system. Together with the randomness in power demand, the real-time power imbalances between input and output lead to the variability in grid frequency that may directly endanger power grids' normal operation. Wolff I et al. i [31] observe the non-Gaussian features in grid frequencies rising from a wind power feed-in and discover its strong dependence on grid heterogeneities. [Extracted from the article]

Accurate and timely short-term traffic flow forecasting plays a key role in intelligent transportation systems, especially for prospective traffic control. For the past decade, a series of methods have been developed for short-term traffic flow forecasting. However, due to the intrinsic stochastic and evolutionary trend, accurate forecasting remains challenging. In this paper, we propose a noise-immune long short-term memory (NiLSTM) network for short-term traffic flow forecasting, which embeds a noise-immune loss function deduced by maximum correntropy into the long short-term memory (LSTM) network. Different from the conventional LSTM network equipped with the mean square error loss, the maximum correntropy induced loss is a local similar metric, which is immunized to non-Gaussian noises. Extensive experiments on four benchmark datasets demonstrate the superior performance of our NiLSTM network by comparing it with the frequently used models and state-of-the-art models. [ABSTRACT FROM AUTHOR]

One of the most important preconditions for guaranteeing a smooth link between wind farms and the power system is to develop an accurate model for forecasting the wind speed. This paper describes a novel wind speed prediction model based on dynamic adaptive variable-weight optimization theory that considers the relevance of historical observations. The model applies signal preprocessing to recorded wind speed observations using ensemble empirical mode decomposition. The decomposed signals are then subjected to a random noise reduction procedure, which improves the robustness of the prediction model. An autoregressive integrated moving average model, general regression neural network, and long short-term memory are used to recognize the different features of each decomposed subsequence. Brain storm optimization is then applied to further promote the forecasting performance by integrating different forecasting models with dynamically adapted variable weights. To evaluate the prediction capacity of the proposed method, three case studies are conducted. The experimental outcomes reveal that the method presented in this paper provides more satisfactory prediction ability and robustness than other models. [ABSTRACT FROM AUTHOR]

A method for mapping color sky images of convective fair-weather cumuli to a scalar irradiance metric is presented. While basic interpretations of irradiance from color sky imagery are sufficient for many irradiance forecasting applications, sky images containing convective fair-weather cumuli can often contain gray-colored regions. Gray regions of a cloud commonly relate cloud thickness, indicating a region that has a high potential to attenuate surface irradiance from the sun. However, existing irradiance metrics are not found to translate such gray regions in accordance with their importance to a clear-sky index for short-term forecasting. This method exploits the special structure of sky images in three-dimensional red-green-blue (RGB) Cartesian space. By applying principal component analysis and basic clustering operations, a planar discriminant for cloudy-to-clear-sky classification is found. Once the {cloudy, clear-sky} identity of all image pixels is determined, the cloudy and clear-sky pixel distributions are separately remapped in RGB space so that gray pixels score a higher optical depth value than white pixels, and clear-sky pixels are mapped to low optical depth values. The resulting clear-sky index is simply the grayscale interpretation of the remapped image. The method is described and demonstrated with two actual sky images, and errors for sequences of several hundred images are tabulated and discussed. Remapped images are shown to represent gray cloud regions with a relatively high cloud optical depth in comparison with white cloud regions. The proposed metric is shown, by comparison with experimentally measured irradiance time series, to produce a more accurate clear-sky index in comparison with other methods. [ABSTRACT FROM AUTHOR]

Short-term load forecasting (STLF) plays a significant role in economic and social development. As a challenging but indispensable task, STLF has become a hot topic in the field of energy. However, the inadequacy of existing methods lies in their inability to capture accurate input features that are highly related to the output, as the main focus of research has been on improving the accuracy of STLF, while ignoring its stability. Therefore, in this paper, a novel, robust hybrid forecasting system was developed, composed of four modules: (1) data preprocessing, (2) forecasting, (3) optimization, and (4) evaluation. In the data preprocessing module, an effective data preprocessing scheme based on singular spectrum analysis and gray correlation analysis was used to produce a smoother time series and to mine the best input and output structure for the model. An extreme learning machine (ELM) optimized using a multiobjective genetic algorithm (MOGA) that considers both the forecasting accuracy and stability was employed to provide the forecasting. Additionally, a generalized regression neural network (GRNN) was also used in the subsequent module to perform forecasting. Moreover, to further obtain accurate results and to overcome the drawbacks of using single models, a simulated annealing (SA) algorithm was utilized to optimize the combined parameters of the MOGA–ELM and GRNN algorithms in the optimization module. To validate the proposed model, half-hourly load data from the New South Wales and Tasmania are provided as illustrative cases. The experimental results show that the proposed hybrid model obtains more accurate and stable results than the reference models used for comparison. [ABSTRACT FROM AUTHOR]

Cloud movement makes short-term forecasting of solar photovoltaic (PV) panel output challenging. A better PV forecast can realize value for both grid operators and commercial or industrial customers with solar assets. In this study, we build convolutional neural network (CNN) based models to forecast power output from PV panels 15 min into the future. Model inputs are the PV power output history and ground-based sky images for the past 15 min. The key challenge is ensuring that due importance is given to each type of input. We systematically explore 28 methods of "fusing" these heterogeneous inputs in our CNN. These methods of fusion (MoF) belong to 4 families. We also systematically explore the many hyperparameters related to model training and tuning. Limited resources preclude an exhaustive search. We apply a three-stage "funnel" approach instead, wherein we narrow our search to the most promising one of these 28 MoF. We find that a two-step autoregression-CNN MoF has the best performance followed closely by a "mix-in" MoF that performs feature expansion and reduction to give appropriate importance to the two types of inputs. The two-step autoregression-CNN model has a forecast skill (FS) of 17.1% relative to smart persistence on the test set comprising 20 complete days (9 sunny, FS = 22%; 11 cloudy, FS = 16.9%). This optimization results in the improvement of FS from 14.1% for a previously published nonoptimized "baseline" model, a CNN wherein the PV history was simply concatenated to the end of the image-sourced vector obtained after convolution, pooling, and flattening operations. [ABSTRACT FROM AUTHOR]

The application of dynamic mode decomposition to clear-sky index forecasting of shadowing effects of convective fair-weather cumulus clouds is presented. Cloud dynamics are captured by sequences of visible-light photographic video frames. This method can be more easily applied to the modeling of cloud evolution than traditional fluid-based methods and can enhance the existing frozen-cloud advection methods. Its use is demonstrated for an actual fair-weather cumulus cloud image sequence and compared to an advection-only forecast. It is concluded that the method shows promise for very short-term clear-sky index forecasting for up to 7 minute horizons. [ABSTRACT FROM AUTHOR]

Skill scores can be used to compare deterministic (also known as single-valued or point) forecasts made using different models at different locations and time periods. To compute the skill score, a reference forecasting method is needed. Nonetheless, there is no consensus on the choice of reference method. In this paper, three classes of commonly used references methods, namely, climatology, persistence, and their linear combination, are studied in a day-ahead solar forecasting scenario. Day-ahead global solar irradiance forecasts with an hourly resolution are generated using research-grade data from 32 sites around the globe, over a period of 1 year, in an operational manner. To avoid exaggerating the skill scores, it is generally agreed that the most accurate naïve forecasting method should be chosen as the standard of reference. In this regard, the optimal convex combination of climatology and persistence is highly recommended to be used as the standard of reference for day-ahead solar forecasting. [ABSTRACT FROM AUTHOR]

With the continuous development of the global electricity market, the electricity market needs the electricity price forecasting result to be accurate, and electricity price forecasting also has more profound practical significance. Accurate price forecasting can not only better reflect the operation of the electricity market but also help to make better decisions about the electricity market. However, the electricity price is affected by a variety of uncertain subjective and objective factors, as well as the constraints of the power system, making it more difficult to obtain accurate electricity price forecasting than electricity load forecasting. The kernel function occupies a very important position in a support vector machine (SVM), which is the key to the mature development of SVM theory. When using SVM for classification and regression, the appropriate kernel function is the basis and precondition for obtaining better classification and approximation effects. In this paper, a new combinatorial kernel function that combines RBF and UKF kernel functions is proposed and is applied to a least squares support vector machine (LSSVM), and based on this, a new hybrid model empirical mode (EMD)-Mixed-LSSVM is proposed to forecast electricity price. The hybrid model puts the electricity price data after the noise reduction by the EMD into the LSSVM with the combined kernel function for calculation. In this paper, we forecast the electricity price in Australia and compare the results of EMD-Mixed-LSSVM with other methods. The results show that EMD-Mixed-LSSVM can effectively improve the accuracy of electricity price forecasting. [ABSTRACT FROM AUTHOR]

Short-term load forecasting is an important part in the energy management of micro-grids, and the forecasting error directly affects the economic efficiency of operation. Considering the strong randomness and high volatility of loads on the user side of micro-grids, a short-term load forecasting method based on Empirical Mode Decomposition (EMD), Extreme Learning Machine with different Kernels, and Switching Delayed Particle Swarm Optimization (SDPSO) is proposed. First, the history load dataset is decomposed into several independent Intrinsic Mode Functions (IMFs) by EMD, and the sample entropy values of the IMFs are calculated. According to the approximation degree of sample entropy values, the IMFs are divided into three categories. Then, ELM with different kernels is adopted to forecast the three categories. Finally, the prediction results are summed to obtain the final prediction result. SDPSO is used to optimize the relevant parameters in the forecasting model. Three micro-grids with different capacities are used to verify the model, and the experimental results demonstrate that the proposed forecasting model can provide better accuracy than the other two methods. The proposed model can provide practical reference for the efficient operation of micro-grids. [ABSTRACT FROM AUTHOR]

In real-world operation conditions, Hybrid Energy Systems (HESs) are exposed to a wide variety of uncertainties, which cause an unexpected operation and performance in the case of neglecting the effects of uncertainties in design and operation processes. This paper presents a multi-objective optimization of the component size and long-term operation of the HES in the presence of multiple uncertainties, considering the Net Present Cost and Energy Not Served as the objective functions. Uncertainties related to load forecasting, wind speed, and components' outage are probabilistically modeled and incorporated into the Multi-Objective Particle Swarm Optimization approach using the Monte Carlo Simulation (MCS) method. MCS generates samples (future scenarios) of uncertain variables based on the probabilistic models of the uncertainties. The optimization algorithm determines the optimal values of component's size as well as the operation parameters to efficiently optimize the HES operation. The applicability and effectiveness of the proposed method are investigated through some numerical analyses. The proposed method can be a useful stochastic optimization tool to consider important uncertainties in the practical design and operation of the HES. [ABSTRACT FROM AUTHOR]