Research on the Prediction Model for Abrasive Water Jet Cutting Based on GA-BP Neural Network

Abstract

In combination with genetic algorithm (GA), the prediction in abrasive water jet (AWJ) cutting depth and process parameters optimization are implemented aiming at the disadvantages of slow convergence rate and prone to trap in local optimal solution of BP neural network. Moreover, the prediction unit for AWJ cutting is developed to make the network model visible for convenience in guiding the operation during actual machining. The neural network, when combined with GA, makes full use of nonlinear mapping ability of neural network and global optimization capability of GA. Prior to further optimization of BP algorithm, this paper adopts GA to optimize the initial weight and threshold to the area near to global minimum point. The contrast between prediction and experimental results shows that the GA-BP neural network can implement the prediction of cutting depth and optimization of process parameters effectively.