Based on the research of the theories of the artificial searching swarm algorithm (ASSA), in order to overcome the shortcomings of the ASSA, the Lévy Flight Artificial searching swarm algorithm is proposed (LFASSA). In order to verify the performance of the improved algorithm, the standard test function is selected for testing. The performance of the algorithm before and after the improvement is compared, and the convergence speed and calculation accuracy of LFASSA are proved. In this paper, ANSYS software is used to establish the parametric model of undervoltage release. The LFASSA and ANSYS software are combined to optimize the undervoltage release. The optimization results show that under the condition of satisfying the suction requirement, the volume of copper and iron of the undervoltage release is reduced, and the magnetic field distribution is more uniform, which achieves the purpose of reducing cost and improving product performance.