Search

Accurate and fast event identification in power systems is critical for taking timely controls to avoid instability. In this paper, a synchrophasor measurement-based fast and robust event identification method is proposed considering different penetration levels of renewable energy. A difference Teager-Kaiser energy operator (dTKEO)-based algorithm is first proposed to improve multiple-events detection accuracy. Then, feature extractions via the integrated additive angular margin (AAM) loss and the long short-term memory (LSTM) network are developed. This allows us to deal with intra-class similarity and inter-class variance of events when high penetration renewable energy occurs. With the extracted features, a multi-stage weighted summing (MSWS) loss-based criterion is developed for adaptive data window determination and fast event pre-classification. Finally, the re-identification model based on feature similarity is established to identify unknown events, a challenge for existing machine learning algorithms. Simulation results on the IEEE 39-bus, Kundur 2-area, and an actual large-scale power grid system are used to demonstrate the advantages of the proposed method over others. [ABSTRACT FROM AUTHOR]

As an efficient tool for uncertainty quantification of renewable energy forecasting, prediction intervals (PIs) provide essential prognosis to power system operator. Merely improving the statistical quality of PIs with respect to calibration and sharpness cannot always contribute to the operational value for specific decision-making issue. In this paper, the cost-oriented prediction intervals are firstly proposed to achieve the joint improvement of forecasting quality and decision performance. In order to bridge the gap between forecasting and decision, a novel cost-oriented machine learning (COML) framework is established, which unifies nonparametric renewable power PI construction and decision-making. Formulated as a bilevel programming model, the COML minimizes the operational costs of decision-making process by adaptively adjusting the quantile proportion pair of PIs resulting from extreme learning machine based quantile regression. The hierarchical optimization model of the COML is equivalently simplified as a single level nonlinear programming problem. Then an enhanced branch-and-contract algorithm with innovative bounds contraction strategy is devised to efficiently capture the optimum of the single level problem with bilinear nonconvexity. Numerical experiments based on actual wind farm data simulate the online forecasting and decision process for wind power offering. Comprehensive comparisons verify the substantial superiority of the proposed COML methodology in terms of forecasting quality, operational value, as well as computational efficiency for practical application. [ABSTRACT FROM AUTHOR]

While there are growing interests in using pumped hydro storage to facilitate the integration of renewable resources, the flexibility of storage is not being fully utilized by existing energy and market management systems. Today, one common approach to operate pumped hydro storage is to determine schedules for a future time horizon based on a look-ahead operational planning stage, with limited adjustment in real-time. However, as renewable penetration levels increase, such approaches do not fully utilize the flexibility of pumped hydro storage. In this paper, a policy function approach is proposed to enhance the utilization of pumped hydro storage in real-time operations. The performance of the approach is evaluated and compared with other benchmark approaches using the IEEE RTS 24-bus system. The result shows that the policy function approach is effective in utilizing the flexibility of pumped hydro storage and has minimal added computational complexity to the existing dispatch process. [ABSTRACT FROM AUTHOR]

Dynamic Security Assessment (DSA) for the future power system is expected to be increasingly complicated with the higher level penetration of renewable energy sources (RES) and the widespread deployment of power electronic devices, which drive new dynamic phenomena. As a result, the increasing complexity and the severe computational bottleneck in real-time operation encourage researchers to exploit machine learning to extract offline security rules for the online assessment. However, traditional machine learning methods lack in providing information on the confidence of their corresponding predictions. A better understanding of confidence of the prediction is of key importance for Transmission System Operators (TSOs) to use and rely on these machine learning methods. Specifically, from the perspective of topological changes, it is often unclear whether the machine learning model can still be used. Hence, being aware of the confidence of the prediction supports the transition to using machine learning in real-time operation. In this paper, we propose a novel Conditional Bayesian Deep Auto-Encoder (CBDAC) based security assessment framework to compute a confidence metric of the prediction. This informs not only the operator to judge whether the prediction can be trusted, but it also allows for judging whether the model needs updating. A case study based on IEEE 68-bus system demonstrates that CBDAC outperforms the state-of-the-art machine learning-based DSA methods and the models that need updating under different topologies can be effectively identified. Furthermore, the case study verifies that effective updating of the models is possible even with very limited data. [ABSTRACT FROM AUTHOR]

In a smart grid, state estimation (SE) is a very important component of energy management system. Its main functions include system SE and detection of cyber anomalies. Recently, it has been shown that conventional SE techniques are vulnerable to false data injection (FDI) attack, which is a sophisticated new class of attacks on data integrity in smart grid. The main contribution of this paper is to propose a new FDI attack detection technique using a new data-driven SE model, which is different from the traditional weighted least square based SE model. This SE model has a number of unique advantages compared with traditional SE models. First, the prediction technique can better maintain the inherent temporal correlations among consecutive measurement vectors. Second, the proposed SE model can learn the actual power system states. Finally, this paper shows that this SE model can be effectively used to detect FDI attacks that otherwise remain stealthy to traditional SE-based bad data detectors. The proposed FDI attack detection technique is evaluated on a number of standard bus systems. The performance of state prediction and the accuracy of FDI attack detection are benchmarked against the state-of-the-art techniques. Experimental results show that the proposed FDI attack detection technique has a higher detection rate compared with the existing techniques while reducing the false alarms significantly.

Inspired from recent insights into the common ground of machine learning, optimization and decision-making, this paper proposes an easy-to-implement, but effective procedure to enhance both the quality of renewable energy forecasts and the competitive edge of renewable energy producers in electricity markets with a dual-price settlement of imbalances. The quality and economic gains brought by the proposed procedure essentially stem from the utilization of valuable predictors (also known as features) in a data-driven newsvendor model that renders a computationally inexpensive linear program. We illustrate the proposed procedure and numerically assess its benefits on a realistic case study that considers the aggregate wind power production in the Danish DK1 bidding zone as the variable to be predicted and traded. Within this context, our procedure leverages, among others, spatial information in the form of wind power forecasts issued by transmission system operators (TSO) in surrounding bidding zones and publicly available in online platforms. We show that our method is able to improve the quality of the wind power forecast issued by the Danish TSO by several percentage points (when measured in terms of the mean absolute or the root mean square error) and to significantly reduce the balancing costs incurred by the wind power producer. [ABSTRACT FROM AUTHOR]

Numerous research works establish the high efficacy of Machine Learning (ML) based power system fault diagnosis over conventional analytical methods. The ongoing integration of renewable energy sources (RES) into the existing transmission networks alters the system topology, potentially resulting in significant changes in fault signatures depending on the size of the newly added RES. However, there is a notable absence of studies in the literature analyzing the impact of new RES integration on the fault diagnosis performances of ML models. Therefore, to assess the fault classification and localization performance of potential ML models, this paper proposes to analyze two practical scenarios arising from new RES integrations: 1) when no-fault data is available for the changed system and 2) when the changed system data is available over time. The proposed performance and adaptability testing of potential ML models has been conducted by optimally integrating different sizes of RES into ‘IEEE 9-Bus System’. The integrated solar-based RES has been modeled incorporating standard temperature and irradiance variations. A diverse fault database was generated considering actual field variations of fault attributes. Impact analysis revealed significant degradation in the fault diagnosis performances of all tested models post RES integrations. The adaptability testing was performed by extensive analysis of the learning trends of ML models with gradual data availability. The proposed Bayesian ridge regression has emerged as the fastest learning model for locating transmission line faults, whereas XGBoost, Extra Tree, and Random Forest classifiers gave comparable results for fault classification.

There is a rapid increase in the utilization of renewable sources such as solar and wind to provide power and electricity. The reason for this trend is to reduce costs and preserve the environment. However, the challenge is to efficiently use and store the energy from these sources. One approach is to optimize the decision when to charge or discharge a battery. The objective is to generate the greatest monetary gain. More charge/discharge cycles will reduce the life of a battery, thereby increasing the cost. A strategy to reduce the number of charge cycles while maintaining the effectiveness of electricity distribution from battery storage will improve battery life. With the inevitable proliferation of electric vehicles (EVs) in the market, strategies specific to electric vehicle battery profitability will be explored. An additional concern which threatens the financial feasibility of battery energy storage systems (BESSs) is the requirement of secure operation. There is currently little research study on strategies to detect cyberattacks on such systems. In this paper, we have presented a novel taxonomy for battery optimization, survey representative BESS utilization strategies, and classify these schemes within the taxonomy. Within our classification, we outline the battery optimization methods that have been discussed, analyze their ability to address issues that arise when implementing a BESS, and describe alternative research that could be explored. Future research could refer to this information to create unique battery optimization schemes to provide more efficiency and optimal revenue for a BESS when compared to current strategies.

Heavy renewable penetrations and high-voltage cross-regional transmission systems reduce the inertia and critical frequency stability of power systems after disturbances. Therefore, the power system operators should ensure the frequency nadirs after possible disturbances are within the set restriction, e. g., 0.20 Hz. Traditional methods utilize linearized and simplified control models to quantify the frequency nadirs and achieve frequency-constrained unit commitments (FCUCs). However, the simplified models are hard to depict the frequency responses of practical units after disturbances. Also, they usually neglect the regulations from battery storage. This paper achieves FCUCs with linear rules extracted from massive simulation results. We simulate the frequency responses on typical thermal-hydro-storage systems under diverse unit online conditions. Then, we extract the rules of frequency nadirs after disturbances merely with linear support vector machine to evaluate the frequency stability of power systems. The algorithm holds a high accuracy in a wide range of frequency restrictions. Finally, we apply the rules to three typical cases to show the influences of frequency constraints on unit commitments.

Potential malicious cyber-attacks to power systems which are connected to a wide range of stakeholders from the top to tail will impose significant societal risks and challenges. The timely detection and defense are of crucial importance for safe and reliable operation of cyber-physical power systems (CPPSs). This paper presents a comprehensive review of some of the latest attack detection and defense strategies. Firstly, the vulnerabilities brought by some new information and communication technologies (ICTs) are analyzed, and their impacts on the security of CPPSs are discussed. Various malicious cyber-attacks on cyber and physical layers are then analyzed within CPPSs framework, and their features and negative impacts are discussed. Secondly, two current mainstream attack detection methods including state estimation based and machine learning based methods are analyzed, and their benefits and drawbacks are discussed. Moreover, two current mainstream attack defense methods including active defense and passive defense methods are comprehensively discussed. Finally, the trends and challenges in attack detection and defense strategies in CPPSs are provided.

The intermittent and stochastic nature of solar energy generation systems, climate change, and the inefficiency of modern power systems due to zero inertia have created many challenges for on-grid operators. Solar forecasting systems based on machine learning algorithms are an emerging and effective solution that uses Big Data related to weather phenomena. However, the predictive ability of these algorithms is hampered by the sporadic nature of solar energy generation. In this article, a robust hybrid machine learning system that utilizes multiple linear regression (MLR) and a Pearson correlation coefficient (PCC) was tested on solar power plant sites of varying capacities in Germany (100–8500 kW). The volume of Big Data features can be reduced by focusing on the features that significantly improve the reliability of the mid-term forecasting system. In this way, drastic fluctuations in the prediction of photovoltaic (PV) power generation can be avoided. The results of our approach are evaluated regarding real-world data using the extreme gradient boosting (XGBoost) with feature engineering, and principal component analysis (PCA), in order to forecast PV energy, rank, and track the importance of feature engineering for different PV capacities. Furthermore, we found that the need for selectivity and reduction of performance error was supported by ridge regression. In addition, the proposed novel XGBoost forecast system decreased the root-mean-square error (RMSE) and mean absolute error (MAE) by 30% and 18%, respectively, compared to the Autoencoder and long short-term memory (LSTM) for same datasets. Furthermore, the CoD determination coefficient ($R^{2}$) increased by 85% compared to the statistical model's autoregressive integrated moving average (ARIMA). [ABSTRACT FROM AUTHOR]

Smart networks such as active distribution network (ADN) and microgrid (MG) play an important role in power system operation. The design and implementation of appropriate protection systems for MG and ADN must be addressed, which imposes new technical challenges. This paper presents the implementation and validation aspects of an adaptive fault detection strategy based on neural networks (NNs) and multiple sampling points for ADN and MG. The solution is implemented on an edge device. NNs are used to derive a data-driven model that uses only local measurements to detect fault states of the network without the need for communication infrastructure. Multiple sampling points are used to derive a data-driven model, which allows the generalization considering the implementation in physical systems. The adaptive fault detector model is implemented on a Jetson Nano system, which is a single-board computer (SBC) with a small graphic processing unit (GPU) intended to run machine learning loads at the edge. The proposed method is tested in a physical, real-life, low-voltage network located at Universidad del Norte, Colombia. This testing network is based on the IEEE 13-node test feeder scaled down to 220 V. The validation in a simulation environment shows the accuracy and dependability above 99.6%, while the real-time tests show the accuracy and dependability of 95.5% and 100%, respectively. Without hard-to-derive parameters, the easy-to-implement embedded model highlights the potential for real-life applications.

Deep neural networks have revolutionized many machine learning tasks in power systems, ranging from pattern recognition to signal processing. The data in these tasks are typically represented in Euclidean domains. Nevertheless, there is an increasing number of applications in power systems, where data are collected from non-Euclidean domains and represented as graph-structured data with high-dimensional features and interdependency among nodes. The complexity of graph-structured data has brought significant challenges to the existing deep neural networks defined in Euclidean domains. Recently, many publications generalizing deep neural networks for graph-structured data in power systems have emerged. In this paper, a comprehensive overview of graph neural networks (GNNs) in power systems is proposed. Specifically, several classical paradigms of GNN structures, e. g., graph convolutional networks, are summarized. Key applications in power systems such as fault scenario application, time-series prediction, power flow calculation, and data generation are reviewed in detail. Further-more, main issues and some research trends about the applications of GNNs in power systems are discussed.

We aim to provide a tool for independent system operators to detect the collusion and identify the colluding firms by using day-ahead data. In this paper, an approach based on supervised machine learning is presented for collusion detection in electricity markets. The possible scenarios of the collusion among generation firms are firstly identified. Then, for each scenario and possible load demand, market equilibrium is computed. Market equilibrium points under different collusions and their peripheral points are used to train the collusion detection machine using supervised learning approaches such as classification and regression tree (CART) and support vector machine (SVM) algorithms. By applying the proposed approach to a four-firm and ten-generator test system, the accuracy of the proposed approach is evaluated and the efficiency of SVM and CART algorithms in collusion detection are compared with other supervised learning and statistical techniques.

As typical prosumers, commercial buildings equipped with electric vehicle (EV) charging piles and solar photovoltaic panels require an effective energy management method. However, the conventional optimization-model-based building energy management system faces significant challenges regarding prediction and calculation in online execution. To address this issue, a long short-term memory (LSTM) recurrent neural network (RNN) based machine learning algorithm is proposed in this paper to schedule the charging and discharging of numerous EVs in commercial-building prosumers. Under the proposed system control structure, the LSTM algorithm can be separated into offline and online stages. At the offline stage, the LSTM is used to map states (inputs) to decisions (outputs) based on the network training. At the online stage, once the current state is input, the LSTM can quickly generate a solution without any additional prediction. A preliminary data processing rule and an additional output filtering procedure are designed to improve the decision performance of LSTM network. The simulation results demonstrate that the LSTM algorithm can generate near-optimal solutions in milliseconds and significantly reduce the prediction and calculation pressures compared with the conventional optimization algorithm.

Lightning is one of the most common transient interferences on overhead transmission lines of high-voltage direct current (HVDC) systems. Accurate and effective recognition of faults and disturbances caused by lightning strokes is crucial in transient protections such as traveling wave protection. Traditional recognition methods which adopt feature extraction and classification models rely heavily on the performance of signal processing and practical operation experiences. Misjudgments occur due to the poor generalization performance of recognition models. To improve the recognition rates and reliability of transient protection, this paper proposes a transient recognition method based on the deep belief network. The normalized line-mode components of transient currents on HVDC transmission lines are analyzed by a deep belief network which is properly designed. The feature learning process of the deep belief network can discover the inherent characteristics and improve recognition accuracy. Simulations are carried out to verify the effectiveness of the proposed method. Results demonstrate that the proposed method performs well in various scenarios and shows higher potential in practical applications than traditional machine learning based ones.

Uncertainty modeling of renewable energy sources, load demand, electricity price, etc. create a high volume of data in smart grids. Accordingly, in this article, a precise forecasting method based on a deep learning concept with microclustering (MC) task is presented. The MC method is structured based on hybrid unsupervised and supervised clustering tasks by K-means and Gaussian support vector machine, respectively. In the proposed method, the input data sequence is clustered by the MC task, and then the forecasting process is employed. By applying the MC, input data in each hour are categorized into different groups, and a distinctive forecasting unit is allocated to each one. In this way, more clusters and forecasting networks are earmarked for the hours with higher fluctuation rates. The bi-directional long short-term memory (B-LSTM), which is one of the newest recurrent artificial neural networks, is proposed as the forecasting unit. The B-LSTM has bidirectional memory—feedforward and feedback loops—that helps us to investigate both previous and future hidden layers data. The optimal number of clusters in each hour is determined based on the Davies–Bouldin index. To evaluate the performance of the proposed method, in this study, three forecasting tasks including the wind speed, load demand, and electricity price are studied in different periods using the Ontario province, Canada, data set. The results are compared with other benchmarking methods to verify the robustness and effectiveness of the proposed method. In fact, the proposed method, which is equipped with the MC technique and B-LSTM networks, significantly promotes the forecasting results, especially in spike points. [ABSTRACT FROM AUTHOR]

Estimating the potential load curtailments as a result of hurricane is of great significance in improving the emergency response and recovery of power grid. This paper proposes a three-step sequential method in identifying such load curtailments prior to hurricane. In the first step, a twin support vector machine (TWSVM) model is trained on path/intensity information of previous hurricanes to enable a deterministic outage state assessment of the grid components in response to upcoming events. The TWSVM model is specifically used as it is suitable for handling imbalanced datasets. In the second step, a posterior probability sigmoid model is trained on the obtained results to convert the deterministic results into probabilistic outage states. These outage states enable the formation of probability-weighted contingency scenarios. Finally, the obtained component outages are integrated into a load curtailment estimation model to determine the expected potential load curtailments in the grid. The simulation results, tested on the standard IEEE 118-bus system and based on synthetic datasets, illustrate the high accuracy performance of the proposed method.

Fast and accurate forecasting of schedulable capacity of electric vehicles (EVs) plays an important role in enabling the integration of EVs into future smart grids as distributed energy storage systems. Traditional methods are insufficient to deal with large-scale actual schedulable capacity data. This paper proposes forecasting models for schedulable capacity of EVs through the parallel gradient boosting decision tree algorithm and big data analysis for multi-time scales. The time scale of these data analysis comprises the real time of one minute, ultra-short-term of one hour and one-day-ahead scale of 24 hours. The predicted results for different time scales can be used for various ancillary services. The proposed algorithm is validated using operation data of 521 EVs in the field. The results show that compared with other machine learning methods such as the parallel random forest algorithm and parallel $k$-nearest neighbor algorithm, the proposed algorithm requires less training time with better forecasting accuracy and analytical processing ability in big data environment.

The protection of intertie zone between wind farm and grid line is critical for stable and safe operation of both the grid line and wind farm in the event of fault within or outside the intertie zone. As a reliable source of renewable energy doubly fed induction generator (DFIG)-based wind farms have been increasingly integrated to the power grid over the last two decades. Nowadays with the enormous penetration of large-scale DFIG wind farms, the commonly used distance relays are no longer reliable due to their incapability of providing accurate impedance measurement during internal and external faults. Thus, it results in maloperation, false tripping, and/or delayed operation. Therefore, in this article a digital differential-based protective relay (DBPR) scheme is designed and developed to provide reliable protection for wind farm intertie zone. Additionally, a new Bayesian-based optimized support vector machine (SVM), as a supervised machine learning classifier approach, is developed to take into account both the dynamic behaviors of wind speed and the current measured by the current transformers. Thus, the proposed hybrid SVM-DBPR scheme can distinguish among the normal operation, internal and external faults correctly that helps to avoid any false tripping. In a laboratory environment the proposed DBPR is implemented in realtime using FPGA DE2-115 board equipped with Cyclone IV-E device (EP4CE115F29C7). It is found from both simulation and experimental results that the proposed hybrid SVM-DBPR is able to provide reliable, efficient, and robust protection for the intertie zone of wind farms with 97.5% accuracy rate. [ABSTRACT FROM AUTHOR]

Timely and accurate forewarning of solar power ramp events (SPREs) is crucial for a power system operation. In this article, a novel forewarning method for SPREs is proposed based on credal network (CN) and imprecise Dirichlet model (IDM). A new expression of SPRE is proposed, which focuses on the power change caused by meteorological fluctuation. Considering that the single-valued probability may not provide convincing results in case of insufficient ramp event records, probability interval is adopted to reflect the ambiguous correlation between SPREs and meteorological conditions. The meteorological evidences are mapped to ramp events by a CN to enhance the sensitivity of SPRE identification. The structure of the CN is established by the learning algorithm with respect to the relations between SPRE and meteorological conditions, as well as different meteorological conditions. Furthermore, an extended IDM is developed to estimate the interval-valued parameters in the CN. Then, a credal classifier is established to output the ramp forewarning result. The effectiveness of the proposed method is verified through case studies, and obvious improvement on the accuracy of ramp forewarning can be seen. [ABSTRACT FROM AUTHOR]

Large-scale datacenters are facing increasing pressure of capping their carbon emission and power cost. Many leading-edge studies have started to explore server clusters running on multiple power sources. Existing approaches do not sufficiently consider the fine-grained power delivery to satisfy diverse requirements in datacenter, especially in the multi-tenant/colocation datacenter, which may yield low energy utilization. To address the emerging trend and new requirements, this article proposes a novel Datacenter inner Power Switch Network (DiPSN) to improve datacenter power efficiency and user satisfaction.DiPSNis a reconfigurable and easy-to-scale-out power architecture, which enables datacenter to distribute various power sources in a fine-grained manner. Moreover, a tailored machine learning based power source management framework is proposed forDiPSNto dynamically optimize user customized performance metrics and maximize datacenter revenue. Compared with conventional single-switch power distribution system, ourDiPSNcan be configured to improve solar energy utilization by 39.6 percent, reduce utility power cost by 11.1 percent and improve workload performance by 33.8 percent. Meanwhile, our design can extend battery lifetime by 9.3 percent. This work could provide valuable guidelines for designing heterogeneous power distribution architecture and management methodology in datacenters for improving user-customizable efficiency, sustainability and economy. [ABSTRACT FROM AUTHOR]