Prediction of EDM process effect based on GA-BP neural network

Addressing the intricacies of electrical discharge machining (EDM), a GA-BP neural network model has been developed to enhance the genetic algorithm of SKD61 steel EDM technology, incorporating variables like current, pulse width, pulse stop electrical parameters, machining depth, machining area, and electrode shrinkage machining parameters as inputs, and the roughness of the workpiece as outputs. The BP neural network model's relative forecast error ranges from 1.57% to 26.25%, with the mean square error (MSE) of 18.4599. In contrast, the GA-BP neural network model, refined through a genetic algorithm, exhibits a relative prediction error between 0.09% and 1.01%, and the MSE is 0.0497. The GA-BP neural network model demonstrates a notably high level of predictive precision, aligning with real-world scenarios.