Prediction of cutting parameters based on improved neural networks

The prediction and control of cutting process was a complicated problem and the suitable cutting parameters are instrumental to cutting process, the prediction models for cutting parameters realized with artificial neural networks was proposed in this paper. Artificial neural networks have strong non-linear modeling ability which can express the nonlinear mapping relation of input and output, but the traditional BP neural networks has many shortcomings such as easily step into local minimum, with weak generalization ability, especially the middle layer neuron are hard to determine, so the correlation pruning algorithm was applied to resolve the problem, the neural networks prediction models was created and the correlation of hidden layer nodes was analyzed and the algorithm was programmed, the simulation taken under MATLAB software which compared new algorithm with traditional BP algorithm, which shows the new methods is effective and can provide a guidance to optimizing turning parameters and turning process control.