SEM-guided surface morphology classification and intelligent optimization of EDM parameters for AISI D2 steel

This research investigates electrical discharge machining of AISI D2 steel, a high-chromium, high-carbon tool steel widely used in industrial applications such as dies, molds, and cutting tools. The post-machining surface characteristics were evaluated using scanning electron microscopy. A convolutional neural network was employed to classify scanning electron microscopy images based on surface morphology, and a pixel intensity histogram-based approach was adopted for efficient feature extraction. The convolutional neural network model achieved a classification accuracy of 93.9% across six surface categories. The results indicate that increased discharge energy leads to the formation of larger craters, globules, and debris, resulting in higher surface roughness values. Particle swarm optimization combined with the technique for order of preference by similarity to ideal solution was used to optimize electrical discharge machining process parameters. The optimal input values for I _p , Ton, Tau, and V were identified as 5, 75, 85, and 60, respectively, and the results were validated experimentally with an error margin of less than 8%. The proposed framework establishes a quantitative relationship between scanning electron microscopy-based morphology and surface roughness, enabling effective electrical discharge machining process optimization.