Stable cutting zone prediction in computer numerical control turning based on empirical mode decomposition and artificial neural network approach

Stable cutting zone prediction is the key requirement for retaining high-productivity with enhanced surface quality of work-piece. Tool chatter is one of the factors responsible for abrupt change in surface quality and productivity. In this research work, an optimum safe cutting zone has been predicted by analyzing the tool chatter so that higher productivity can be achieved. Initially, chatter signals have been recorded by performing experiments at different combinations of cutting parameters on computer numerical control trainer lathe. Further, these recorded signals have been preprocessed by empirical mode decomposition technique, followed by the selection of dominating intrinsic mode functions using Fourier transform. The preprocessed signals have been used to evaluate a new output parameter, that is, chatter index (CI). Artificial neural network (ANN) based on the feedforward backpropagation network has been proposed for predicting tool chatter in turning process. The input machining parameters considered are depth of cut, feed rate and cutting speed. It has also been deduced that from available different transfer functions, the Hyperbolic Tangent transfer function in ANN is best suitable to predict tool chatter severity in turning operation. Moreover, the safe cutting zone has been assessed by evaluating the dependency of CI on cutting parameters. Finally, more experiments have been conducted to validate the obtained cutting zone.