The energy storage system is an important part of photovoltaic microgrid. However, the traditional battery energy storage system has many problems such as a low power density and a short service life. A supercapacitor hybrid energy storage system based on photovoltaic power generation is established, and its control strategy is proposed, under which the DC-DC converter connected to the energy storage element is controlled through the coordinated operation of model predictive control (MPC) and low-pass filter, and the DC-AC converter connected to the AC side is controlled through a virtual synchronous generator. MPC is a real-time control method, and it can determine the constraints in certain time domain in the future according to the system state at present and optimize the control objectives in the predicted time domain, which has good stability and robustness compared with other control methods. Under changes in the parameters including the intensity of solar radiation, the DC bus voltage can be kept stable, and the dynamic response capability of DC bus voltage can be improved. Finally, the system is simulated in MATLAB/Simulink, and the effectiveness of the proposed method is verified.

With a focus on the research of energy storage configuration and aimed at the problems such as incomplete data and poor data quality of distributed photovoltaic output, a configuration method for the basic responsiveness capability of energy storage based on the extreme value theory is proposed. First, a probabilistic analysis method for distributed photovoltaic output based on generalized extreme value distribution is put forward, and the power data under different granularities of extreme value is fitted and analyzed. Second, with two goals of reducing the photovoltaic abandonment rate and smoothing the output fluctuation normally, the configuration method for the basic responsiveness capability of energy storage is proposed based on the return period and return level of distributed photovoltaic output. Finally, the feasibility and extensibility of the proposed method are verified by simulations of an example based on the output data of a distributed photovoltaic power station in a certain area. Results show that the generalized extreme value distribution can achieve a good fitting of distributed photovoltaic output distribution, and the dependence on data accuracy is minor. Meanwhile, the configuration results of the basic responsiveness capability of energy storage given by the proposed method agree with the minimum storage capacity requirements for photovoltaic projects in some provinces of China.

The overall development statuses and trends of hydrogen-related industries are briefly introduced at first, and different types of electrolytic water hydrogen production technologies are compared in detail. Then, the development statuses of photovoltaic (PV) power hydrogen production, wind power hydrogen generation and wind-PV complementary hydrogen generation technologies are summarized, and the development trends in the future are investigated. Finally, the research framework system of “Green Hydrogen Plant” for a large-scale new energy hydrogen production system is constructed. The research and development ideas for key technologies involved in new energy hydrogen production are provided, the demand analysis for products and solutions required by relevant projects is conducted, and suggestions for the research direction of “wind-PV-hydrogen-storage” energy comprehensive utilization in the future are put forward.

Lithium-ion batteries are the most commonly used batteries in an energy storage system, and their state-of-health (SOH) is an important battery parameter monitored in a battery management system. In addition to the accuracy of prediction, the simulation of a user’s random charging behavior for realizingreal-time evaluation at any time is also very important. Based on the user’s random charging behavior and from the perspective of obtaining the data integrity, the SOH prediction methods are divided into three types, which are based on complete charging cycle data, segment of charging data and segment of charging data restored complete data, respectively. The mean and variance of battery capacity changes during the charging process of NASA battery aging experimental data are used as feature inputs of different prediction models, and the prediction results obtained using three methods are compared and analyzed. Results show that the complete charging data features have high prediction accuracy, and the quality of features extracted from segment data is related to the segment collection time. The complete data features can be obtained in a short time through the restoration network, but the prediction accuracy is affected by the restoration network. In addition, the actual prediction needs to take into account the randomness of charging behavior and the real-time performance of prediction results.

The estimation performance of state-of-charge (SOC) of lithium battery plays a vital role in abattery management system. The recursive least squares method with forgetting factor was used to identify the parameters of a battery model. Aimed at the problem that the sampling of a filter in the existing SOC estimation algorithm cannot well simulate the probability distribution of the system state, a rank sampling method was introduced, and an adaptive square root rank filter algorithm was proposed. The covariance matrix of the system state was updated iteratively in the form of square root, and the observation noise was adjusted adaptively according to the innovation in each iteration, so as to improve the SOC estimation accuracy and its adaptive capability with respect to noise. The performance of the proposed algorithm was verified by simulation experiments under FUDS and DST conditions. Results show that the proposed algorithm can improve the convergence speed and estimation accuracy compared with extended Kalman filter, unscented Kalman filter and rank filter.

The use of traditional constant balance coefficient for the state-of-charge (SOC) of batteries cannot reach a compromise between the balancing speed and circulating current. To overcome this limitation, an adaptive balance optimization control strategy for battery SOC is proposed, which is based on modular multilevel converter-battery energy storage system (MMC-BESS). A detailed analysis of the power components within the MMC-BESS bridge arm is given, and the design of an SOC equalization controller based on direct control of battery power on the DC-DC side is optimized. A standard deviation factor based on a constant proportional balance coefficient is introduced, and the interphase-interarm-intraarm cascaded SOC balance algorithm is improved, which can adaptively determine the proportional balance coefficient according to the SOC difference in each module. In addition, a feedforward control quantity is added to the outer power loop of the DC-DC controller, so as to improve the response speed of battery power charging and discharging. The simulation results based on MATLAB/Simulink show that the proposed SOC balance control strategy can reduce the circulating current at the initial stage of balance and shorten the total balance time.

The analysis method based on incremental capacity (IC) achieves effective analysis of the aging mechanism of lithium-ion batteries by converting voltage plateaus into characteristic peaks. However, the immune capability of IC curves with respect to noise is poor, and the extracted health features are susceptible to interference from the noise peaks. To solve this problem, a state-of-health (SOH) estimation method based on noise immunity probability density function (PDF) feature extraction combined with an ensemble learning model is proposed. First, the PDF curve is obtained by calculating the PDF of the voltage curve, and different peak features extracted from the PDF and IC curves are analyzed and compared by setting different values of signal-to-noise ratio. Second, the learning speed and randomness of extreme learning machine are used to construct an ensemble learning model framework, so as to reduce the prediction error of a single model. Finally, SOH estimation experiments under different noise conditions were conducted on the CALCE dataset, and results demonstrated the high accuracy and reliability of the proposed method.

In a new power system characterized by a “dual-high” structure, the virtual synchronous generator (VSG) technology has been widely applied in the grid-forming control of inverters owing to its inherent response capability to inertia and frequency variations. When the grid voltage is unbalanced, the power injected by grid-connected inverters based on the conventional VSG will include second-harmonic fluctuations. To address this issue, a virtual voltage and current regulation strategy under three-phase unbalanced conditions based on VSG is proposed. Through the analysis of the mechanism of inverter power fluctuations and the virtual phase-current regulation (VPCR) method used in grid-following control, it is pointed out that the existing VPCR method cannot solve the power fluctuation problem under grid-forming control. With the combination of grid-forming control characteristics of VSG, a power fluctuation suppression strategy is proposed, which integrates virtual voltage and current regulation on the basis of the existing VPCR method. This approach only requires adjusting the voltage and current gain coefficients, without the need for additional control loops or positive-negative sequence separation, thereby fully preserving the control characteristics and grid-forming attributes of VSG. The proposed method can effectively suppress power fluctuations under voltage imbalance. Experimental results verified the feasibility of the proposed power fluctuation suppression strategy.

The wind-thermal bundled outgoing system may cause sub-synchronous oscillation, while the inherent low-frequency oscillation of the thermal power unit and the interaction between its controller and sub-synchronous oscillation increase the difficulty in stability control. To solve this problem, a small-signal model of the system is established using the modular modeling idea at first, and the leading factors and participation modules of low-frequency and sub-synchronous oscillations are obtained by analyzing the eigenvalues and participation factors of the system oscillation modes, so as to determine the input signal and installation position of the controller. Then, the robust active disturbance rejection control (ADRC) technology is used to perform linearization, and a multi-channel and multi-frequency band doubly-fed wind farm linear ADRC (L-ADRC) additional damping controller is designed. Finally, considering the influence of the low-frequency oscillation additional damping controller on the damping of sub-synchronous frequency band, a coordinated suppression strategy for the multi-channel and multi-frequency band doubly-fed wind-thermal outgoing system based on an improved particle swarm optimization algorithm is put forward, which is verified by time-domain simulations. Results show that the designed control strategy has good suppression effect on low-frequency and sub-synchronous oscillation, and it also has good immunity and robustness.

The short-circuit and phase failure of a DC flexible offshore wind power transmission system at the receiving end will generate excess power, leading to an incremental voltage of DC bus. One of the effective methods to solve this issue is adding DC energy-consuming devices. To reduce the volume and cost of DC energy-consuming devices, a DC energy-consuming device based on a novel submodule and the corresponding hybrid switching control strategy are proposed. The proposed control strategy combines the advantages of step switching control and cyclic switching control schemes, which can effectively reduce the power impact on the system caused by the step change in excess power and simplify the topology of the new submodule. Meanwhile, the problems of operation imbalance and unequal heat dissipation among submodules can be solved. A design method for the parameters of electrical components in the DC energy-consuming device is also given, and a simulation model of the proposed DC energy-consuming device is built. Simulation results verify the effectiveness and feasibility of the topology of the proposed novel DC energy-consuming device and its hybrid switching control strategy.