Remaining useful life prediction of cutting tools using wavelet packet transform and extreme learning machine

Abstract

Online tool wear prediction is a determining factor to the success of smart manufacturing operations. The implementation of sensors based Prognostic and Health Management (PHM) system plays an important role in estimating Remaining Useful Life (RUL) of cutting tools and optimizing the usage of Computer Numerically Controlled (CNC) machines. The present paper deals with health assessment and RUL estimation of the cutting tool machines based on Wavelet Packet Transform (WPT) and Extreme Learning Machine (ELM). This approach is done in two phases: a learning (offline) phase and a testing (online) phase. During the first phase, the WPT is used to extract the relevant features of raw data computed in the form of nodes energy. The extracted features are then fed to the learning algorithm ELM in order to build an offline model. In the online phase, the constructed model is exploited for assessing and predicting the RUL of cutting tool. The main idea is that ELM involves nonlinear regression in a high dimensional feature space for mapping the input data via a nonlinear function to build a prognostics model. The method was applied to real world data gathered during several cuts of a milling CNC tool. The performance of the proposed method is evaluated through the accuracy metric. Results showed the significance performances achieved by the WPT and ELM for early detection and accurate prediction of the monitored cutting tools.