Parameter identification and loss prediction of modified J-A dynamic hysteresis model based on chaotic genetic particle swarm optimization

The accurate calculation and prediction of dynamic hysteresis losses are of great significance for the design and optimization of energy consumption in high-frequency electromagnetic devices. This paper modifies the existing J-A dynamic hysteresis model based on the nonlinear relationship between dynamic coefficients and magnetic flux density. Aiming at the problems of low computational accuracy and susceptibility to local optima in traditional particle swarm optimization algorithms, a modified hybrid algorithm (GA-CAPSO-NLDIW) is proposed by introducing Tent chaotic mapping function, adaptive velocity factor and learning factor, T-distribution function, and genetic algorithm crossover and mutation factor. Theoretical and experimental verification shows that the hybrid algorithm has low dependence on initial values and has better convergence speed and computational accuracy than other single algorithms. And this modified model, and method can be used for frequency dependent hysteresis loop identification, covering multi-component magnetic material systems such as ferromagnetic materials and low saturation magnetic flux density soft magnetic materials. In addition, this method can also predict dynamic losses under other unknown magnetic densities by using hysteresis loops under known magnetic densities, with errors within a reasonable range. The modeling method and hybrid optimization strategy proposed in this study can provide theoretical basis for the design of motor and magnetic bearing structures, accurate evaluation of dynamic losses in electromagnetic equipment, and other fields.