A Novel Method of Material Requirement Forecasting for Discrete Manufacturing System Based on Improved Genetic Algorithm

To solve the problem of low accuracy of material requirement forecasting in discrete manufacturing enterprises, a novel method is being presented by using BP neural network and genetic algorithm. With the proposed method, the collected data of customer orders are cleaned to remove the invalid data. And the processed data are trained using BP neural network to generate the objective function of genetic algorithm. Then, the operators of selection, crossover and mutation are optimized to eliminate inferior populations, increase superior populations and accelerate the iteration rate of superior population genes with adaptive genetic algorithm. In addition, the improved operators and objective functions can be involved in the MATLAB simulation model to achieve a better material requirement forecasting result for the discrete manufacturing system. Finally, empirical research shows that the accuracy of material requirement forecasting has been improved by using the improved genetic algorithm.