By taking into account the social behaviors of renewable distributed generations (RDGs) and electric vehicles (EVs), the multi-operator AC/DC hybrid distribution system (HDS) has evolved into a cyber-physical-social system (CPSS), necessitating an appropriate architecture. In this paper, a distributed CPSS architecture is proposed for the multi-operator AC/DC HDS using a cloud-edge-device cooperation. The architecture incorporates a model that distinguishes different types of EVs, enabling the demonstration of the interaction among the cyber, physical, and social systems. To address issues of transmission delay and noisy, a Non-ideal Communication Alternating Direction Method of Multipliers is employed. The proposed framework consists of three stages: day-ahead economic dispatch, intra-day security dispatch, and real-time adjustment. This enables efficient dispatching of EVs based on economic and node voltage considerations while optimizing the performance of renewable distributed generators for each operator. Case studies are conducted to demonstrate the effectiveness of the CPSS structure in achieving distributed dispatching the HDS.

In order to solve the problem of frequency modulation power deviation caused by the randomness and fluctuation of wind power outputs, a method of auxiliary wind power frequency modulation capacity allocation based on the data decomposition of a “flywheel + lithium battery” hybrid-energy storage system was proposed. Firstly, the frequency modulation power deviation caused by the uncertainty of wind power is decomposed by the successive variational mode decomposition (SVMD) method, and the mode function is segmented and reconstructed by high and low frequencies. Secondly, a mathematical model is established to maximize the economic benefit of energy storage considering the frequency modulation mileage, and quantum particle swarm optimization is used to solve the target model considering the charging and discharging power of energy storage and the charging state constraints to obtain the optimal hybrid-energy storage configuration. Finally, the simulation results show that, in the step disturbance, the Δf_max of the hybrid-energy storage mode is reduced by 37.9% and 15.3%, respectively, compared with single-energy storage. Under continuous disturbance conditions, compared with the single-energy storage mode, the Δf_{p_v} is reduced by 52.73%, 43.72%, 60.71%, and 47.62%, respectively. The frequency fluctuation range is obviously reduced, and the frequency stability is greatly improved.

To further improve the frequency regulation stability of wind farm, and optimize the state of charge (SOC) basepoint, charge and discharge rate and recovery capacity of energy storage. In an isolated, off‐grid state, a two‐layer optimization method is proposed, taking into account the frequency regulation reliability and SOC adaptive adjustment of the wind storage. In the upper layer, the model predictive control theory is adopted to optimize the energy‐storage frequency regulation output power with the goal of minimizing the wind storage frequency regulation power deviation. While solving the problem of low‐frequency regulation reliability of wind farm, the SOC recovery basepoint and frequency regulation power of energy storage are optimized. At the SOC optimization layer, based on the upper SOC recovery basis point, we propose a dynamic recovery method for energy‐storage SOC that considers both the SOC recovery demand and grid‐bearing capacity, to determine the energy‐storage recovery power. By determining the frequency regulation or recovery power, we propose a calculation method to optimize the energy‐storage charge and discharge coefficients as per the SOC for avoiding excessive charging and discharging. The simulation results show that, under continuous disturbance, the root‐mean‐square deviations of the proposed method is 80.13% and 62.63% lower than those of the fixed K and SOC methods, respectively. Furthermore, the proposed method exhibits the best SOC maintenance effect.

This study proposes a short-term load prediction method of a bidirectional long short-term memory network based on feature mining of the power consumption big data in combination with the attention mechanism (AT) of Bayesian optimization to address the problems that a considerable amount of feature factors exist and the feature relationship is obscured in the historical power consumption big data. The method comprehensively considers the global features of the power consumption data in space and the local features in time. First, the Cen-CK-means clustering method is used to cluster the electricity consumption data of users, and the statistical, combination, and time category characteristics are extracted according to the meteorological factors related to load over multiple time scales. Second, the Bayesian and bidirectional long and short memory networks are combined to extract the temporal and spatial characteristics of the load data itself. Meanwhile, the AT is introduced to automatically assign the corresponding weights to the hidden layer state of the bidirectional long and short memory. This task is carried out to distinguish the importance of the different time load series, which can effectively reduce the loss of historical information and highlight information about key historical time points. Finally, taking the first type of load as an example, compared with the SVP, RBPNN, BiLSTM, and BO-BiLSTM algorithms, the MAPE index is reduced by 1.05%, 1.75%, 0.52%, and 0.26%, respectively. RMSE decreased by 186.61, 154.93, 91.88, and 15.76 MW, respectively, while R2 increased by 0.04, 0.07, 0.03, and 0.03, respectively. In the one-week forecast time, MAPE index decreased by 1.97%, 2.44%, 1.21%, and 0.6%, respectively; RMSE decreased by 271.18, 305.7, 183.13, and 97.91 MW, respectively; and R2 increased by 0.12, 0.08, 0.04, and 0.03, respectively.