Geometric influence and multi-objective optimization of WEDM for hardened tool steels using ANN and NSGA-II

Nowadays, due to their exceptional strength, hardness, and resistance to wear, hardened AISI D2 and DC53 tool steels are frequently used in the tool and die-making business. But these materials are difficult to machine conventionally because they include metallic carbides that are abrasive and shorten the life of the cutting tool. Wire electric discharge machining (WEDM) is broadly used to obtain exact details on these hardened steels. However, differences in geometric attributes may result in differences in the machining process. The impact of machining parameters on linear cutting speed (CS) and volumetric material removal rate (MRR) for both flat and curved profiles is examined in this work. These parameters include peak current (I_P), voltage (V), spark gap/pulse (P), and heat-treated material type (D2 and DC53). A Taguchi L18 orthogonal array is employed in the study to methodically evaluate performance disparities associated with various geometric characteristics. Understanding process physics is aided by energy dispersive X-ray analyses (EDX) and scanning electron microscopy (SEM), while statistical variability and parameter importance are investigated using analysis of variance. An artificial neural network (ANN) is designed to predict output parameters and assess correlations between experimental and anticipated values in order to handle inconsistencies and complex interactions. Ultimately, the parameters are improved using a non-dominated sorting genetic algorithm (NSGA-II), which leads to notable gains in CS and MRR for curved and flat profiles by 63.17%, 75.36%, 74.10%, and 73.38%, respectively, over unoptimized settings.