Aiming at the multi-peak output of photovoltaic arrays in partial shading, this paper proposes a new parallel combination evolutionary algorithm (GA-DE-PSO), which solves the disadvantages of instability and lack of precision of traditional single evolutionary algorithm to track the maximum power. The method randomly divides all feasible solution individuals into two sub-populations, and uses the differential evolution model of vectors and the genetic model of chromosomes to generate new individuals and candidate individuals. Then through the particle swarm algorithm for hybrid selection, the guidance information of the more effective feasible domain is obtained, so as to converge to the best advantage more quickly and achieve maximum power point tracking. The simulation results show that the combination algorithm retains the advantages of three evolutionary algorithms and has high precision and stability. This work is supported by National Natural Science Foundation of China (No. 51708194) and Key Research Project of Hunan Provincial Department of Education (No. 18A120).