Research Projects

Micro-Electrical Discharge Machining (µEDM) is uniquely suited for shaping hard-to-cut conductive materials like tungsten and titanium alloys without inducing mechanical stress. However, its efficiency and accuracy are often constrained by limited understanding of plasma dynamics and material removal processes. Plasma channels, which are central to µEDM operations, govern heat transfer and material removal but involve complex interactions that are often oversimplified in existing models, reducing predictive accuracy for machining outcomes. Incorporating ultrasonic vibrations into µEDM has demonstrated potential to enhance performance by improving molten material and debris evacuation, yet the interactions between ultrasonic effects, plasma stability, and dielectric properties remain underexplored. Additionally, the application of ML in µEDM is in its infancy, with existing models struggling to generalize across diverse operating conditions. These challenges limit the reliability and optimization of the process. This project, Enhancing Micro-EDM Precision with ML-Based Prediction and High-Fidelity, proposes a comprehensive framework to address these limitations focusing on improving µEDM by integrating analytical, numerical, and machine learning (ML) techniques. Analytical models will be developed to describe plasma channel behavior and heat transfer using fundamental physics. Finite Element Modeling (FEM) simulations will explore the effects of parameters such as capacitance, voltage, and spark gap on plasma pressure and heat flux distribution under air, deionized water, and hydrocarbon-based dielectrics. Ultrasonic vibration effects will be dynamically incorporated into FEM simulations to evaluate their influence on material removal rate (MRR) and surface quality. High-speed imaging at over 100,000 frames per second will capture plasma dynamics and crater morphology, generating data for training ML models. These models will predict machining outcomes such as plasma channel geometry, heat-affected zones, and crater dimensions. Image processing techniques like Sobel edge detection and Otsu thresholding will enhance data quality, while optimization algorithms such as Genetic Algorithms and Particle Swarm Optimization will identify optimal machining settings for improved efficiency and precision. The project’s objectives include developing analytical and numerical models for plasma channel dynamics, investigating the role of ultrasonic vibrations in enhancing µEDM, and creating ML-based predictive models to optimize process parameters. The anticipated outcomes of this project include enhanced precision and efficiency in µEDM, broader applicability across industries, and contributions to advanced manufacturing science through better understanding of plasma behavior and innovative ML integration. This work sets the stage for establishing new standards in micro-manufacturing and supporting precession industries producing miniaturized products.