Research on Supply Chain Demand Prediction Based on Bp Neural Network Algorithm

Demand prediction is a hot research field in markets management, especially for fresh agricultural products prediction based on supply chain management. Based on BP neural network, a new demand prediction algorithm for fresh agricultural products is presented in the paper. First, the structure and data indicators of BP neural network algorithm are redesigned and the training function is selected for the fresh agricultural products prediction algorithm. Second, the improvement of excitation function, (including trigonometric function and sigmoid function) and orthogonalizable design, are presented and analyzed to speed up the calculation and improve the prediction accuracy of ordinary BP algorithm. Finally, data from certain fresh agricultural product corporations are taken for example and the simulation results show that not only the problem of convergence speed has been solved, but also the prediction accuracy is ensured when the improved algorithm is used in demand prediction for fresh agricultural products .