Field-circuit Coupled Optimization of Magnetic Latching Monostable Actuator using Multi-objective Jellyfish Search Algorithm and Back Propagation Neural Network

For enhancing the performance of magnetic latching monostable actuators in vacuum circuit breakers, a field-circuit coupled finite element analysis (FEA) model was established, integrating magnetic hysteresis characteristics with transient drive circuit dynamics. A backpropagation neural network (BPNN) predictive model was constructed, where the structural parameters of the actuator and excitation circuitry parameters were selected as the input layer, and the closing holding force, average closing terminal velocity, and opening response time were served as the output layer. Moreover, the error analysis confirmed that the proposed predictive model effectively captures quantitative correlations between input variables and output responses through rigorous statistical validation. Furthermore, the actuator design was optimized through an integrated approach combining the multi-objective jellyfish search (MOJS) algorithm with a BPNN predictive model. And the post-optimization experimental validation revealed performance enhancements: a 12.4% increase in closing holding force, 17.6% reduction in average closing terminal velocity, and 4.5% reduction in opening response time.