Key points are not available for this paper at this time.
Abstract Wire Electrical Discharge Machining (Wire-EDM) is renowned for its precise material removal and superior surface roughness (Ra). However, in-process variations can lead to defects in mass production, necessitating time-consuming and labor-intensive post-production quality checks. This research aims to develop an online surface roughness prediction model using both multi-linear regression and Fuzzy Logic. Duplex stainless steel 2507 was selected as the test material, and a design of experiment was conducted with three controllable parameters: Wire Tension, Supply Voltage and Table Feed Rate, to simulate the variance and uncertainties in Wire-EDM operations. The top three real-time data inputs retrieved from the in-machine monitoring system - maximum voltage, current difference, and maximum feed rate - were identified through regression analysis. Predictive models using both Multi-linear regression and Fuzzy Logic were developed, achieving prediction accuracy of 94.47% and 95.75%, respectively. The results show that Fuzzy Logic exhibits a stronger correlation between predicted and actual surface roughness. This system offers a fast and reliable means of predicting surface roughness during Wire-EDM operations, thereby enhancing production efficiency and product quality. Future research could explore the integration of additional sensors or optimizing the prediction model further with alternative methodologies.
Yu et al. (Fri,) studied this question.