Wind power systems participating in primary frequency regulation have become a novel trend. In order to solve the problem of the over-speed deloading (OSD) control of wind power systems failing to provide reserved capacity for primary frequency regulation while under turbulent winds, this paper analyzes the influence mechanism of turbulence characteristics on the OSD control and the relationship between the reserve capacity of OSD control and the deloading power coefficient under turbulent wind speeds, while also quantifying the relationship between the turbulence characteristic index and deloading power coefficient. The range of the deloading power coefficient is obtained accordingly, based on which improved OSD control is proposed to dynamically optimize the deloading power coefficient according to the turbulence characteristics, which improves the frequency regulation performance of wind power systems under turbulent wind speed. According to the simulations and experimental results, the improved method proposed in this paper has good effectiveness and superiority in frequency regulation effect and rotor speed performance.

With the continuous improvement of wind power penetration in the power system, the volatility and unpredictability of wind power generation have increased the burden of system frequency regulation. With its flexible control mode and fast power adjustment speed, energy storage has obvious advantages in participating in power grid frequency regulation. Therefore, this paper studies the control strategy of wind energy storage combined with primary frequency regulation and proposes a control method of wind energy storage combined with primary frequency regulation based on the operation state of wind turbines. This paper analyzes the reserve power and rotor reserve kinetic energy of wind turbines operating in different wind speed ranges, introduces the fan frequency regulation operation state coefficient to quantify the real-time frequency regulation ability of the unit, regards the fan and energy storage system as independent frequency regulation sources, and designs the control strategy of joint frequency regulation of wind and storage in different wind speed ranges. The energy storage system is employed to participate in frequency control in the low-wind-speed range, thereby addressing the “blind spot” issue of wind turbine unit frequency control alone. In the medium- and high-wind-speed ranges, the real-time complementary output of wind and energy storage power is achieved by assigning weights based on the frequency control operation status coefficient. Finally, the effectiveness of the joint frequency modulation control strategy of wind storage in low-, medium-, and high-wind-speed regions is verified in the simulation model.

Aiming at the load fluctuation problem caused by a high proportion of new energy grid connections, a reactive power optimization method based on deep reinforcement learning (DRL) considering topological characteristics is proposed. The proposed method transforms the reactive power optimization problem into a Markov decision process and models and solves it through the deep reinforcement learning framework. The Dueling Double Deep Q-Network (D3QN) algorithm is adopted to improve the accuracy and efficiency of calculation. Aiming at the problem that deep reinforcement learning algorithms are difficult to simulate the topological characteristics of power flow, the Graph Convolutional Dueling Double Deep Q-Network (GCD3QN) algorithm is proposed. The graph convolutional neural network (GCN) is integrated into the D3QN model, and the information aggregation of topological nodes is realized through the graph convolution operator, which solves the calculation problem of deep learning algorithms in non-European space and improves the accuracy of reactive power optimization. The IEEE standard node system is used for simulation analysis, and the effectiveness of the proposed method is verified.

With large-scale wind and solar power connected to the power grid, the randomness and volatility of its output have an increasingly serious adverse impact on power grid dispatching. Aiming at the system peak shaving problem caused by regional large-scale wind power photovoltaic grid connection, a new two-stage optimal scheduling model of wind solar energy storage system considering demand response is proposed. There is a need to comprehensively consider the power generation cost of various types of power sources, day-ahead load forecasting information, and other factors and plan the day-ahead output plan of the energy storage system with the minimum system operation cost as the optimization objective of day-ahead dispatching. The demand response strategy is introduced into the time-ahead optimal scheduling, and the optimization of the output value of the energy storage system in each period is studied with the goal of minimizing the system adjustment cost. The particle swarm optimization algorithm is used to solve the model, and the IEEE33 node system is used for an example simulation. The results show that using the demand response and the collaborative effect of the energy storage system can suppress the uncertainty of wind power and photovoltaic power, improve the utilization rate of the system, reduce the power generation cost of the system, and achieve significant comprehensive benefits.

In order to reduce the influence of abnormal data on load forecasting effects and further improve the training efficiency of forecasting models when adding new samples to historical data set, an ultra-short-term load dynamic forecasting method considering abnormal data reconstruction based on model incremental training is proposed in this paper. Firstly, aiming at the abnormal data in ultra-short-term load forecasting, a load abnormal data processing method based on isolation forests and conditional adversarial generative network (IF-CGAN) is proposed. The isolation forest algorithm is used to accurately eliminate the abnormal data points, and a conditional generative adversarial network (CGAN) is constructed to interpolate the abnormal points. The load-influencing factors are taken as the condition constraints of the CGAN, and the weighted loss function is introduced to improve the reconstruction accuracy of abnormal data. Secondly, aiming at the problem of low model training efficiency caused by the new samples in the historical data set, a model incremental training method based on a bidirectional long short-term memory network (Bi-LSTM) is proposed. The historical data are used to train the Bi-LSTM, and the transfer learning is introduced to process the incremental data set to realize the adaptive and rapid adjustment of the model weight and improve the model training efficiency. Finally, the real power grid load data of a region in eastern China are used for simulation analysis. The calculation results show that the proposed method can reconstruct the abnormal data more accurately and improve the accuracy and efficiency of ultra-short-term load forecasting.

The development of electricity spot markets necessitates more refined and accurate load forecasting capabilities to enable precise dispatch control and the creation of new trading products. Accurate load forecasting relies on high-quality historical load data, with complete load data serving as the cornerstone for both forecasting and transactions in electricity spot markets. However, historical load data at the distribution network or user level often suffers from anomalies and missing values. Data-driven methods have been widely adopted for anomaly detection due to their independence from prior expert knowledge and precise physical models. Nevertheless, single network architectures struggle to adapt to the diverse load characteristics of distribution networks or users, hindering the effective capture of anomaly patterns. This paper proposes a PLS-VAE-BiLSTM-based method for anomaly identification and correction in load data by combining the strengths of Variational Autoencoders (VAE) and Bidirectional Long Short-Term Memory Networks (BiLSTM). This method begins with data preprocessing, including normalization and preliminary missing value imputation based on Partial Least Squares (PLS). Subsequently, a hybrid VAE-BiLSTM model is constructed and trained on a loaded dataset incorporating influencing factors to learn the relationships between different data features. Anomalies are identified and corrected by calculating the deviation between the model’s reconstructed values and the actual values. Finally, validation on both public and private datasets demonstrates that the PLS-VAE-BiLSTM model achieves average performance metrics of 98.44% precision, 94% recall rate, and 96.05% F1 score. Compared with VAE-LSTM, PSO-PFCM, and WTRR models, the proposed method exhibits superior overall anomaly detection performance.