Dynamic Relationship Analysis and Decision Optimization of Key System Factors Based on LSTM Model

Abstract

Within the setting of a vegetable laboratory,the short shelf life of vegetable samples poses a challenge as their quality tends to deteriorate rapidly. This study delves into the relationship between the usage of various vegetable categories and individual samples with time, as well as the interdependence among different projects, using descriptive statistical analysis and correlation coefficient matrix calculations. To more accurately predict the overall usage of vegetable samples in various categories for the upcoming seven days, an advanced Long Short-Term Memory (LSTM) model is introduced. This deep learning algorithm successfully captures the dynamic trends in sample usage through the learning of time-series data, providing a robust predictive tool for laboratory management.Considering laboratory requirements, the total number of different vegetable categories, and the usage of samples in each category, the focus of the study shifts from vegetable categories to individual items. In order to determine the precise replenishment quantity of experimental materials for July 1st, a genetic algorithm model is employed. This unique model design, achieved through the continuous evolution of populations, identifies the optimal solution that best fits the experimental conditions. Such a distinctive modeling approach enables laboratory managers to plan laboratory resources more effectively, ensuring optimal experimental outcomes throughout the process.