The nonisolated grid‐connected photovoltaic inverter system suffers from the problem of leakage current. Therefore, a leakage current suppression control strategy based on the model predictive direct power control is proposed for the three‐level four‐leg grid‐connected inverter. At first, the system model is established to analyze the factors that affect the leakage current and the potential of the neutral point. Second, the first three legs, inverting part, use model predictive direct power control, and the fourth leg of the inverter is used to maintain a constant common‐mode voltage to achieve the suppression of the leakage current. The fourth leg is indirectly controlled by the switch state of the first three legs; thus, the number of optimization times is reduced. Finally, by simulation, the leakage current is compared and analyzed with the three‐level three‐leg inverter under conditions with different parasitic capacitance values. The results show that the proposed strategy suppresses the leakage current by more than 60% without affecting the neutral point potential and has a fast optimization speed.

Aiming at the problem of multi-objective weight coefficient setting of model predictive control (MPC) for permanent magnet synchronous motor (PMSM), a hybrid particle swarm optimization (HPSO) algorithm with low computational complexity of fitness value is proposed to realize the self-setting of weight coefficient of cost function. In the proposed strategy, good particles update velocity and position through particle swarm optimization (PSO) algorithm, while bad particles not only do the same but generate the offspring by cross and mutation, and then the worse offspring will be replaced by their extremum individuals. It is faster that the adaptive cross and mutation rate makes the offspring get closer to the good particles, and it increases the diversity of particles without destroying the good particles. Experimental results show that compared with other optimization algorithms, the proposed algorithm. Firstly, is more inclined to escape from the local optimum. Secondly, it has higher search accuracy and faster convergence speed. Moreover, with setting weight coefficient, the system speed regulation time is shortened, the current total harmonic distortion (THD) is reduced significantly, and the switching frequency is effectively reduced without affecting the output power quality.