<p>The existing optimization methods of memory machine (MM) with more variable parameters usually require two rounds of multi-objective genetic algorithm (MOGA) optimization, which greatly sacrifices the optimization efficiency. In response to this issue, a novel efficient optimization methodology integrating MOGA and fuzzy inference Taguchi (FIT) method is proposed in this article. Specifically, the overall optimization accuracy is ensured through the first round of MOGA optimization. Subsequently, the comprehensive correlation coefficients of each parameter to all optimization objectives are analyzed, and more accurate parameter value ranges are obtained according to those coefficients. The second round of FIT optimization constructs orthogonal experiments based on the latest parameter value ranges to greatly reduce the required computational cases, thereby improving the overall optimization efficiency. The complete workflow of the proposed methodology is described in detail. The multi-mode field-circuit coupling model and fuzzy inference system, which are required to be established for completing the optimization, are mainly introduced. Furthermore, the advantages and effectiveness of the proposed methodology in improving the electromagnetic characteristics and optimization efficiency are verified by applying it to the optimization of a hybrid magnetic circuit MM. Some experimental measurements on a 1.2 kW prototype validate the correctness of the above analysis.</p>