The predictive model of roundness online measurement based on wavelet network during crankshaft noncircular grinding

Based on the wavelet network, this paper builds a predictive model, which is applied in approximating and prediction of roundness errors for crankshaft roundness online measure system. It aims at enhances the dynamic measuring accuracy and input to the control system as a feedback signal. The experimental result shows the predictive model can faster converge with high accuracy as an alternative to a neural network to approximate a nonlinear system.