Product Schemes Evaluation Method Based on ImprovedBP Neural Network

摘要

Improved back propagation (BP) neural network evaluation method for product schemes took the main index data as input vector, took the sample comprehensive scores as output by using the analytic hierarchy process (AHP). The network was separately trained by momentum factorial algorithm, Gauss-Newton algorithm and Levenberg-Marquardt algorithm. With the application and verification in Haier refrigerator schemes, the comparison of speed and mean absolute error show that the BP neural network trained by Levenberg-Marquardt algorithm is reliable.