A Hybrid SVR Model Based on Support Vector Regression and Differential Evolution for Milling Force Prediction of Titanium Alloy

A Hybrid SVR Model Based on Support Vector Regression and Differential Evolution for Milling Force Prediction of Titanium Alloy

Prediction of milling force plays an important role in milling process of titanium alloy. In this paper, the milling process of titanium alloy is studied, and the material quality is affected directly by the milling force. Support vector regression (SVR) has shown a prominent performance for many practical applications. Although there is some literature about parameter optimization techniques of SVR model, it still needs further research and improvement on the performance and accuracy of this model. We present a hybrid milling force prediction model, namely DE&SVR, which hybridizes the SVR with differential evolution (DE) to enhance the prediction accuracy for milling force of titanium alloy. The main advantage of hybrid model is that the DE is adopted to optimize the kernel parameters of the SVR. The main parameters affecting milling force, such as the milling depth, feeding speed, and cutting speed, are considered in this study. The results have shown that the hybrid model yields better prediction accuracy, and the percentage prediction milling force deviation is found to be less than 3.5% for all the cases tested, NRMSE is only 0.0200, and MAPE is only 1.4791%. Thus, this methodology can be widely applied to the fields of material processing optimization.