Optimization of EDM process parameters based on variable-fidelity surrogate model

To balance the cost and accuracy in the optimization of EDM process parameters, the variable-fidelity surrogate model (VFM) was introduced in this paper to replace the implicit relationship between the EDM process parameters (peak current, duty cycle, pulse period) and performance functions (material remove rate, MRR, and surface roughness, Ra). Low-fidelity (LF) samples were obtained by using the low-cost EDM heat flow simulation, and high-fidelity (HF) samples were obtained by using the EDM machining experiments. By fusing the LF and HF samples, it is possible to establish an accurate expression of the implicit relationship between the EDM process parameters and performance functions in the variable-fidelity framework. Finally, the EDM process parameter model was established, and sequential quadratic programming (SQP) was used to obtain the optimal solution by calling the VFM. Through the verification experiments, the results showed that the parameter combination obtained by VFM is 15.1% higher than the MRR of the low-fidelity model (LFM) and 13.0% higher than the high-fidelity model (HFM), which reaches 88.285 mg/min, while the Ra has slightly decreased and is within the constraint range. From the above research, it can be concluded that the proposed technology has significant advantages and application potential in the field of EDM machining optimization.