Intelligent Classification and Prognostics of Discharge Pulses and Wire Breakage in WEDM of Ni-Based Superalloys using Advanced Machine Learning Techniques

Wire Electrical Discharge Machining (WEDM) offers substantial benefits when cutting difficult-to-cut materials, such as Nibased superalloys. However, common concerns, including wire breakage and unpredictable discharge pulses, offer significant difficulties to productivity, surface integrity, and process sustainability. This paper proposes an advanced Machine Learning (ML) framework for intelligent discharge pulse categorisation and wire breakage prediction. Raw voltage and current data were captured during WEDM to extract essential pulse properties, which were then utilised to train classification models such as Artificial Neural Networks (ANN). The ANN model obtained a 98% pulse classification accuracy, efficiently discriminating normal, short circuit, and arc discharges in wire breakage prediction based on input data such as peak current, spark gap voltage, and pulse lengths. Major Findings: The study found that high arc and brief discharges increase surface roughness and the chance of wire rupture. This ML-driven strategy provides a strong platform for real-time monitoring, fault prediction, and process optimisation in WEDM of superalloys.

View article page

Download PDF