A Generalized Method for Featurization of Manufacturing Signals, With Application to Tool Condition Monitoring

Abstract

The application of machine learning techniques in the manufacturing sector provides opportunities for increased production efficiency and product quality. In this paper, we describe how audio and vibration data from a sensor unit can be combined with machine controller data to predict the condition of a milling tool. Emphasis is placed on the generalizability of the method to a range of prediction tasks in a manufacturing setting. Time series, audio, and acceleration signals are collected from a Computer Numeric Control (CNC) milling machine and discretized into blocks. Fourier transformation is employed to create generic power spectrum feature vectors. A Gaussian Process Regression model is then trained to predict the condition of the milling tool from the feature vectors. We highlight that this multi-step procedure could be useful for a range of manufacturing applications where the frequency content of a signal is related to a value of interest.

Three-Level NPC Inverter Incipient Fault Detection and Classification using Output Current Statistical Analysis

“…In recent decades, more and more applications in industrial, manufacturing, or the transportation domain have been increasingly electrified for efficiency and environmental issues [1][2][3]. However, due to the reliability, availability, and maintainability requirements [4], the system must remain operational despite fault occurrence and its performances must remain robust to environmental nuisances.…”