Performance monitoring and failure prediction system for wire electric discharge machining process through multiple sensor signals

Abhilash, P. M. and Chakradhar, Dupadu (2022) Performance monitoring and failure prediction system for wire electric discharge machining process through multiple sensor signals. Machining Science and Technology, 26 (2). pp. 245-275. ISSN 1091-0344 (https://doi.org/10.1080/10910344.2022.2044856)

[thumbnail of Abhilash-Chakradhar-MST-2022-Performance-monitoring-and-failure-prediction-system-for-wire-electric-discharge-machining-process]
Preview
Text. Filename: Abhilash_Chakradhar_MST_2022_Performance_monitoring_and_failure_prediction_system_for_wire_electric_discharge_machining_process.pdf
Accepted Author Manuscript
License: Strathprints license 1.0

Download (2MB)| Preview

Abstract

This study aims to develop a pulse classification algorithm to understand wire electric discharge machining (wire EDM) process stability and performance based on the discharge pulse characteristics. Also, a process data driven failure prediction system is proposed. The wire EDM monitoring system includes high sampling rate differential probes and current probes. The features extracted through pulse train analysis were spark discharge energy, ignition delay time, spark frequency and proportion of various discharges. A pulse discrimination algorithm was proposed, which classifies the discharges into open circuit sparks, arc discharges, short circuit sparks and normal sparks. It was observed that higher proportions of short circuit pulses resulted in inferior part quality. The differences in the pulse cycle during stable and unstable machining were studied based on the extracted features. It was found that the discharge frequency and the proportion of arc and short circuit pulses were extremely high before the wire breakages. An artificial neural network (ANN) model was developed to predict the process responses, like cutting speed and surface roughness, from the process data. Also, an intelligent algorithm was developed based on the extracted in-process data to predict the unstable conditions, leading to machining failures. The accuracy of the algorithm was confirmed to be very high by conducting confirmation tests.