Research on the Price Prediction of Agricultural Products Based on GSAA-BP Neural Network

China is a large agricultural country, and changes in the sales prices of agricultural products affect the country's economic development. As the prices of agricultural products are subject to external factors such as climate change, people's eating habits and national economic regulation, prices vary greatly from period to period, which causes great distress to agricultural operators and consumers. Careful analysis of price trends and accurate forecasting of agricultural prices are conducive to maintaining the stability of agricultural markets and the improvement of people's living standards. In this paper, we collect daily price data from 8 April 2021 to 26 February 2022, using potatoes as the research object, and introduce an agricultural price forecasting model combining GA and SA to optimize BP. The model first uses GA as the framework, introduces annealing operators after the GA selection, crossover and variation operations to improve the GA local search capability, optimizes the initial weight threshold of the BP using the GSAA algorithm, and assigns the optimized weight threshold to the BP neural network for training prediction. The GSAA-BP algorithm was used for the analysis of potato price forecasting, and the empirical results showed that the average absolute error was around 1.5%, the average percentage error was around 0.03% and the root mean square error was around 0.05%, which was a great improvement in accuracy compared with the traditional BP. The results show that the combined GSAA-BP forecasting model has good forecasting ability and can provide a reference for agricultural price forecasting.