Abstract: The P-U curve of photovoltaic array presents multiple peak values in complex shading environment,resulting in the failure of maximum power point tracking（MPPT）algorithm. Therefore,a two-layer control model is proposed. In the upper model,Levy flight and polynomial mutation strategies are introduced into GWO algorithm to form a Levy-polynomial grey wolf optimization algorithm（LPGWO）to search for the global maximum power points. In the underlying model,the perturbation observation method is used to partly track the maximum power points to stabilize the power oscillation in the dynamic process. Simulation results show that in the multipeak MPPT control,the proposed dual-layer control model has the characteristics of fast-tracking speed,high convergence accuracy and small overall power oscillation,thus improving the tracking efficiency and accuracy of maximum power output of photovoltaic array in complex shading environment.