【Objective】 A grid thinning model based on forest spatial structure optimization was constructed to obtain an efficient and convenient thinning method. 【Method】 Based on the random thinning model, the grid thinning method was proposed, and the random thinning method was used to test the grid thinning method. The grid thinning method indicates that the rasterized forest land is divided into several continuous square grids with the length of integer A, the cutting amount K, and square grid edge length A. Using these variables, all possible thinning methods were enumerated to simulate thinning. Finally, the applicability of the model was verified using the secondary forest of the Jinpenshan Mountain. 【Result】The spatial structure of the secondary forest is not ideal and needs to be improved in terms of the uniform angle index, opening degree and competition index. The optimal thinning intensities of the 33, 40 and 45 years old secondary forests were 34.69%, 21.14% and 29.90%, respectively, and the evaluation index increased by 47.94%, 53.94% and 79.73%, respectively. The 33 and 40 years old forests were suitable for using the grid thinning method, with cutting methods as follows: three trees with the smallest DBH were removed from the grid having a length of 7 m; three trees with the smallest DBH were removed from the grid having a length of 6 m. 【Conclusion】The grid thinning model can effectively improve the spatial distribution pattern of stands, but it is only applicable to the secondary stands with a more uniform forest distribution. For unevenly distributed stands, the model can only determine the interlogging and thinning intensities for the sample plots.