Demand Prediction of Emergency Supplies in Campus Public Health Emergencies Based on GM-GA-BP Neural Network

Abstract

Medical alcohol is the most widely used and highly demanded emergency supply in both campus public environments and medical settings. It has a sharp contradiction between supply and demand in the event of a public health event. For this reason, forecasting the demand for medical alcohol is crucial for averting and managing unexpected public health issues on campus. An enhanced GM-GA-BP neural network prediction model was put out in this article. The benefits of the equal dimension and new information complement GM(1,1) model's new and old information replacement and small sample data prediction are combined. Following data fitting and modification, a BP neural network based on conjugate gradient algorithm is built to carry out training and prediction. Meanwhile, the connection weight and threshold were globally optimized through genetic algorithm.To forecast the demand for medical alcohol at a certain university in Yibin City from 2022 to 2023, a thorough comparison of a combination model vs the conventional BP neural network model was made. The results showed that the MAPE had been decreased by roughly 3% and the absolute percentage error was as low as 0.12686%. These findings have provided a scientific reference for the university to improve its emergency material reserve strategy represented by medical alcohol.