Abstract
Reasonable analysis and formulate relevant policy to control population stable growth is of great significance to social development and economic development in this paper, combining with the relevant data search conducted fertility policy mathematical modeling and analysis, introduces both short-term and long-term memory neural networks with time characteristics (LSTM), considering the population forecast may be is more sensitive to high-dimensional feature, The multi-layer stacked long and short-term memory neural network (DLSTM) was used as the prediction model, and the multi-layer stacked structure significantly improved the robustness and accuracy of prediction.