Parameter Calculation Research Of The Jiles-Atherton Hysteresis Model Using Intelligent Optimization Algorithms

Hysteresis phenomenon is a usual nonlinear characteristic in physical systems and electromagnetic device. It plays a key role in the safety of the system operation and the stable running of the equipment. In recent years, with the development of new materials, for example magnetoelectric composite materials, because of their distinct physical properties, which have latent adhibition in all kinds of microdevices and integrated units such as micro sensors, microelectromechanical systems devices, and high density information storage devices. As a new type of intelligent material, magnetically shape memory alloy (MSMA) has a good performance in its applications in the actuator manufacturing. In addition, magnetic hyperthermia is a promising cancer treatment technique, which brings new hope for cancer patients. These new materials and the application of the technology are both involved in hysteresis nonlinear phenomena. So the design and analysis of new intelligent materials and the research and development of high technology, which depend on the modeling of hysteresis and the calculation accuracy of its parameters. It has a great significant to understanding and analysis of the hysteresis characteristics.It generally contains many parameters to be determined after a hysteresis model is set up. It represents the different physical states when parameters taking different values. The validity of a model design depends largely on the accuracy of the parameters extraction. No matter how perfect a hysteresis model is, the operability of the model is not guaranteed if there is no effective calculation method of these parameters.At present, researchers have proposed many models to describing the magnetic hysteresis, which from the angle of physical and mathematical. The more common models are Bouc-Wen hysteresis model, Preisach hysteresis model, Jiles-Atherton(JA) hysteresis model. In numerous hysteresis model, JA model has a clear physical meaning, less parameters, contains only a first order ordinary differential equation. But the complexity and difficulty of the model parameters identification have been bothering people. Many of the traditional optimization methods have been used to calculate the parameters of JA model. But these methods are vulnerable to the influences of the initial value selection, as a result, the convergence of the algorithm are not guaranteed, easy to fall into local minimum and increasing the computing burden. More worryingly, a lot of optimization methods, especially the optimization methods based on derivative are becoming more and more appear "overwhelmed" when facing the model equations of discontinuous, discrete, unimodal and multimodal mathematical properties.Recently the new optimization techniques based on swarm intelligence have been payed great attention to the calculation of parameters on various areas. Intelligent optimization algorithms are the stochastic optimization algorithms based on the bionics. The typical methods are that Eberhart and Kennedy have put forward the Particle Swarm Optimization(PSO) algorithm, Genetic Algorithm(GA) algorithm, Differential Evolutionary(DE) algorithm and Dorigo has put forward the ant colony algorithm and so on. These methods are widely used in scientific research and the practical problems, which have obtained the very good results that can not be replaced by the traditional optimization methods.Intelligent optimization algorithms have a strong applicability, they have no any requests to the continuity of the objective functions, and are not sensitive to the selection of initial solutions. So it has practical significance to treat the intelligent optimization algorithms as the candidate algorithms for solving complicated optimization problems.The main problems we will concern in this thesis including:(1) This paper proposes a method of the particle swarm optimization algorithm combination with MATLAB/Simulink dynamic simulation integration environment to calculate the parameters of JA hysteresis loop model. By means of noise-free and noisy simulation data, the numerical experiments of JA model with the three groups of different parameter values have been carried out. The new method is proposed based on material magnetization mechanism from the physical perspective, in fully familiar with domestic and foreign research present situations, the deep understanding and mastering the intelligent optimization algorithm and MATLAB/Simulink dynamic simulation integration environment.(2) In addition this paper constructed the JA model equation based on Simulink module. The accurate B-H hysteresis loops have been obtained by means of solving the model equation. The present paper makes the JA model equation of Simulink module and PSO algorithm realizing the seamless integration. It provides guarantee for the smooth running of the algorithms optimization.(3) It has a great influence on the precision, computing time and convergence when the control variables of the algorithms taking different values. So according to different problems, we need to focus on how to choose the optimal parameters configuration in this paper.(4) Finally we made a comparison with genetic algorithm for the computed results. Upon the analysis we found that PSO algorithm shows good robustness, and it is better than genetic algorithm from the calculation accuracy and convergence property for the complex parameters calculation of the Jiles-Atherton nonlinear hysteresis model. Obviously, in the research on material hysteresis characteristics in view of the complexity and difficulty of the model parameters identification that the particle swarm optimization algorithm combination with MATLAB/Simulink dynamic simulation integration environment is a kind of effective and feasible research method.