Hierarchical genetic-particle swarm optimization for bistable permanent magnet actuators

Abstract

Bistable permanent magnet actuator (BPMA) has been widely used in industry as a kind of energy converter. However, its power consumption and response performance are in conflict during operating because of the conflict between holding force and initial force (the minimum threshold driving force). An optimization procedure based on a novel hierarchical genetic-particle swarm algorithm (HGP) for a hybrid excited linear actuator (HELA) constrained in a specific volume has been developed. For obtaining the objective function for each parameter combination precisely, three-dimensional finite element analysis (FEA) was employed. Meanwhile, the improvement of optimization efficiency and quality is extremely demanding because the FEA needs much computing time. The proposed hybrid algorithm adopts a hierarchical structure from the concept of “experimental classes” in China. The base level is composed of the majority general individuals of Genetic algorithms (GA), which provides the global search ability of the entire algorithm. The top level comprises the minority elite consisting of the better individual, which performs the accurate local search by using the particle swarm optimization (PSO) algorithm with variable inertia weight and constriction coefficient. Additionally, the performance of HGP has been evaluated through the Rosenbrock function, and the proposed method is superior to other related methods Finally, the proposed procedure was verified by optimizing HELA's parameters in multidimensional parameters space under the design constraint. The results show that both of the holding force and the initial force are improved more than 25% compared with the initial design, which ensures both low power consumption and fast response.