Aimed at the issue of bus voltage instability caused by constant-power load in a DC microgrid, a variable-gain approach backstepping sliding-mode control strategy based on a nonlinear disturbance observer (NDO) is proposed. First, the exact feedback linearization method is used to obtain the standard Brunovsky linear model of a bidirectional DC-DC converter. Second, a backstepping sliding-mode controller is designed to enhance the robustness of the system while reducing its adjustment time. The error variables are linearly combined to form a sliding-mode surface, and the sliding-mode approach law of power-exponential variable-gain is constructed to achieve fast and accurate tracking of the reference voltage. On this basis, the NDO is introduced to estimate the system perturbation in realtime, and the estimation results are compensated to the backstepping sliding-mode controller to further reduce the steady-state voltage deviation of the system. Finally, MATLAB simulations and an RT-LAB HIL platform demonstrate that the proposed control strategy can enhance the steady-state performance of the system and is highly robust with respect to system and parameter perturbations.

An interleaved parallel double-tube forward converter is taken as the research object. First, based on the analysis of its working mode, an expression for the system efficiency is obtained, and the relationship between input voltage and modulation frequency is given. In addition, a system model is built and simulated using the PSIM software. Second, through the analysis of power losses of the transformer and switching tube in the system, an expression for the converter efficiency is established by taking into account the power losses. Third, a frequency adaptive control strategy based on input voltage is proposed, so as to optimize the system efficiency based on the closed-loop control of system voltage. Finally, an experimental platform of an interleaved parallel double-tube forward converter with rated power of 3 kW was set up, and results verified the effectiveness of the theoretical analysis and proposed control strategy.

To enhance the accuracy of power estimation for a Boost converter with constant power load (CPL) and improve the dynamic performance of the converter system, a sliding mode control strategy based on continuous finite-time disturbance observer is proposed. First, a nonlinear mathematical model for the converter system is established, which is further transformed into Brunovsky canonical form using the accurate feedback linearization technique. Afterwards, a disturbance observer is designed to estimate the total power value of CPL. To address the issues related to numerous adjustment parameters and significant output buffeting in conventional finite-time disturbance observers, a hyperbolic tangent function is introduced to design a continuous finite-time disturbance observer. Second, a non-singular fast terminal sliding mode controller is designed for the canonical form to enhance the robustness of the converter system, and the stability of the closed-loop system is proved based on the Lyapunov theory. Finally, both the simulation and experimental results validated the correctness and superiority of the proposed control strategy.

Aimed at the issues of a long dynamic tracking time and secondary oscillation during load disturbances in interleaved parallel converters, a variable bandwidth active disturbance rejection control strategy is proposed. First, based on the small-signal model analysis, the influence of parasitic parameter perturbations is taken into account in the transfer function, so as to eliminate internal uncertainty disturbances caused by parasitic parameters. Then, starting from the frequency-domain characteristics, the relationship between bandwidth and anti-interference capability is analyzed in terms of stability and anti-interference characteristics. Finally, Monte Carlo experiments were used to test the robustness of the proposed strategy. Simulation and experimental results show that the proposed strategy is superior to double closed-loop proportional-integral (PI) control in target value tracking and anti-interference quality, thus improving the system’s dynamic performance and anti-interference capability.

The grid-connected photovoltaic power generation technology has developed quickly in recent years, but it also poses risks such as the islanding effect. The realization of real-time monitoring of islanding effect is of significance for ensuring the stable operation of photovoltaic power generation systems. Therefore, aimed at the islanding fault of grid-connected photovoltaic power generation systems, a data-driven subspace-based islanding detection scheme is proposed. First, the fundamental wave from the voltage signal and the secondary harmonic signal of voltage at the point of common coupling are taken as the basis for islanding detection, and a voltage fundamental wave output predictor based on the subspace identification approach is designed. An adaptive filter for islanding detection is designed to track the fundamental wave from the voltage signal in real time, and the fundamental wave from the voltage signal is amplified to generate a residual signal. On this basis, an islanding fault detector is designed according to the residual signal and the secondary harmonic signal of voltage, so as to judge whether the system suffers the islanding fault. Finally, the effectiveness of the proposed approach was verified by simulation results.

Aimed at the problems of slow maximum power point tracking (MPPT) convergence speed, low searching accuracy and susceptibility to falling into local optimum in the traditional grey wolf optimization (GWO) algorithm caused by multi-peak output power of photovoltaic (PV) arrays under partial shadow conditions, an algorithm with an improved Levy-flight GWO and variable-step perturbation & observation (ILGWO-VP&O) method is proposed. During the global searching, a new nonlinear convergence factor is put forward to improve the convergence speed and searching accuracy of the GWO algorithm, and the population with normal distribution is initialized to further improve the searching efficiency. Levy-flight is embedded to increase the randomness of global searching, so as to avoid the local optimum during the searching process. The VP&O method is used in the local searching to quickly capture the maximum power point at the global level. To verify the validity of the proposed algorithm, a PV power generation system was built for different published GWO algorithms and the ILGWO-VP&O algorithm, and these algorithms were experimentally verified. Experimental results show that the proposed ILGWO-VP&O algorithm has the fastest dynamic response speed, as well as the best steady-state control accuracy.

To improve the accuracy of fault diagnosis for photovoltaic arrays, a photovoltaic fault diagnosis method based on an improved dung beetle optimization (IDBO) algorithm which is used to optimize a light gradient boosting machine (LightGBM) is proposed. The IDBO algorithm is used to optimize specific hyperparameters in LightGBM, and a fault diagnosis model for photovoltaic arrays based on IDBO-LightGBM is established. To effectively distinguish various faults, the characteristic points extracted from the current-voltage curves are utilized. To validate the effectiveness of the proposed method, simulation analysis is conducted on the model. A comparison is performed among the proposed method, LightGBM algorithm, DBO-LightGBM algorithm, other decision tree algorithms and support vector machine algorithm, and results verify the stability and accuracy of the novel method.

The permanent magnet synchronous motor (PMSM) is an important component of high-power fan electric independent variable-pitch systems. When PMSMs are driven with a three-level converter, model predictive control will exhibit drawbacks such as a poor steady-state performance and high computational demand. To address these issues, a simplified strategy utilizing variable dead-time is proposed. First, a reference vector is obtained based on deadbeat prediction. Second, the basic vector that has the smallest Euclidean distance to the reference vector is rapidly identified based on the introduction of an algebraic method, and the optimal basic vector is determined in conjunction with the DC-side capacitor voltage. Third, the dead-zone voltage vector corresponding to this optimal basic vector is calculated, and both of them are considered as the output vectors for the next cycle. In addition, the duty cycle that minimizes the current control error (i.e., the optimal dead-time) is calculated. Finally, experimental results demonstrate the effectiveness of the proposed control strategy.

The lead-acid battery is one of the most important DC power sources in a substation, and it undertakes the important responsibility of ensuring the normal operation of a power system. To prolong the battery’s cyclic life and respond to environmental protection policies, the repairing technology for substation decommissioned batteries is gradually becoming popular. A low-cost and convenient repairing method with repair solution was selected, and the electrochemical working environment for the battery’s negative electrode was designed and simulated based on the cyclic voltammetry method. The effects of different types of chemical agents on the charge and discharge capacity of negative electrode were studied, and orthogonal experiments were used to screen out composite activators that have excellent improvement effects on the charge and discharge capacity of negative-electrode. The results are of signify-cance for activating and repairing substation decommissioned lead-acid batteries.

The effects of applied mechanical stress on the electro-mechanical properties of stacked-pouch lithium-ion batteries are comprehensively investigated at the full cell level. A series of characterization experiments of electro-mechanical behaviors were conducted based on two experimental platforms, i.e., a free expansion bench and a constant-displacement fixture. Phase transition-dependent swelling phase diagrams were established by differential voltage, swelling thickness and force curve analysis, effectively correlating the electrochemical phase transition reaction with swelling properties. The effect of C-rate on the swelling thickness and force was concentrated in the Stage 2-2L phase transition phase, while the initial preload force will mainly affect the amount of curve stretching of the swelling force without affecting the shape of the swelling curve. This was mainly attributed to the initial preload force dependence of elastic modulus. As an assemblage of composite porous materials, the cell had stress-relaxed viscoelastic properties, and it was further found that the stress relaxation was correlated with various factors such as SOC (state-of-charge), C-rate and initial preload force. For the electrical properties, it was found that the initial preload force had a significant effect on the maximum available capacity, C-rate capacity and internal resistance. Experimental results show that the ohmic internal resistance decreased with an increase in the initial preload force, but the opposite was true for the maximum available capacity. Therefore, it was inferred that the dependence of the initial preload force for the C-rate capacity was dominated by the maximum available capacity. A full range of electro-mechanical performance analysis plays a key role for batteries in pack applications.