Use of stochastic optimization procedures in the identification of the Generalized Moving Preisach Model

The paper presents an identification method for the Generalized Moving Preisach Model (GMPM) by means of stochastic optimization procedures. The use of the GMPM can overcome some limitations of the Classical Preisach Model, but it lacks of established identification procedure. The use of a factorization of the Preisach Distribution Function and an analytical definition of the resulting one dimensional function allows one to reconstruct BH loops as a function of few parameters. The identification can thus be reduced to a best fit problem where optimization procedures have been widely used Stochastic optimization routines are global optimizers so that they guarantee convergence to the global minimum of the function avoiding spurious solutions tied to local minima of the cost function. Results obtained by different procedures are shown and discussed.