Optimisation of superstore single product replenishment and pricing strategy based on LSTM neural network and weighted multi-objective planning

Abstract

Vegetable items in supermarkets usually have a short shelf life and their quality gradually deteriorates over time. Therefore, supermarkets usually need to make daily vegetable replenishment decisions based on historical sales data and demand. The purpose of this paper is to investigate the replenishment strategy and pricing strategy of individual items under many constraints in order to reach the goal of maximising supermarket revenue and market demand. Firstly, the original sales volume data is preprocessed to calculate the sales volume of each single item within 24-30 June, excluding the single item with 0 sales volume, the remaining 49 single items; secondly, the LSTM neural network model is established to predict the non-discounted sales volume, the discounted sales volume, and the cost of each single item on 1 July, respectively, and excluding the single item with total sales volume lower than 2.5 kg on 1, the remaining number of saleable single items is 32; finally, considering the three possibilities of the existing supply and demand relationship, a weighted multi-objective planning model is established to solve for each single item separately, and the optimal solution is selected from it. The results show that the maximum profit of the superstore on 1 July is 886.56 yuan.