Under dynamic shading, the PV array output P-U curve will appear multiple power extreme points, and the traditional maximum power point tracking will fall into local optimal. Therefore, a compound algorithm (ICS-INC) based on the improved cuckoo (ICS) algorithm and incremental conductance (INC) was proposed. The algorithm proposed adaptive discard probability and adaptive step size factor, combined with differential variation to carry out random preference walk in the early stage, so as to improve the search and development ability of the algorithm and avoid falling into local optimal effectively. By improving the Lévy flight formula to reduce its randomness and reduce the number of iterations of the algorithm, the tracking time can be shortened, and the maximum power point area can be quickly and efficiently located. In the later stage, INC can achieve local fast search and stable output of maximum power. A simulation model was established in MATLAB/Simulink, and experiments were carried out on various algorithms. The simulation results show that compared with other algorithms, the algorithm's tracking speed, global search ability and ability to adapt to environmental changes are improved.