Construction of Automatic Replenishment Dynamic Model based on LSTM Neural Network and Simulated Annealing Algorithm

Due to the strong cyclical and seasonal special training of vegetable products in the sales process, it is very necessary to develop flexible pricing and marketing strategies for consumers. This paper mainly studies the formulation of reasonable replenishment strategy, establishes the pricing model of vegetable products, to reduce the loss of the supermarket and bring higher profits. Firstly, after descriptive statistics of six dishes, the correlation is obtained by gray correlation analysis, and the model, the nonlinear function of total sales and cost-plus pricing is determined, and the LSTM model is predicted to change, the simulated annealing algorithm is used to solve the optimal profit in the next week. T hen, the entropy right comprehensive evaluation method is used to select vegetables with high scores, and use the LSTM model to predict the capture and pricing strategy of commodities in the next month. Finally, more variables are included in the replenishment decision and pricing model to further optimize the model. Studies have found that the established LSTM model and the sales of optimized model vegetables have a lot to do. Reasonable establishment of models to control prices can help the sales of vegetables.