Abstract: In order to reduce the vibration of the core reactor effectively, a vibration optimization method of the core reactor considering the magnetostrictive characteristics of the material is proposed in this paper. Firstly, based on the classical J-A model and the quadratic domain rotation theory, an improved magnetostrictive model considering stress is established, and the particle swarm optimization-simulated annealing (PSO-SA) algorithm is used to identify the characteristic parameters of the improved model. The root-mean-square error after identification is less than 3.5, which verifies its accuracy. Then, the simulation model of the electromagnetic field and structure force field of the core reactor is established, and the magnetic field and vibration distribution of the core reactor are calculated. Based on multi-physical field calculation and Latin hypercube sampling test method, combined with sensitivity analysis technology, the influence law of each parameter on vibration is obtained, and the optimization targets and variables are stratified. Finally, response surface methodology (RSM) method and Kriging method are used to optimize the parameters of the core reactor, and the best structural parameters of the core reactor under different methods are obtained. The results show that, compared with the initial parameters, the maximum vibration displacement of the core is reduced by 13.93% and 24.64% respectively by using RSM and Kriging optimization methods. And the loss and metal conductor consumption are also significantly reduced. Therefore, on the premise of ensuring that the performance parameters of the core reactor meet the requirements, the vibration displacement of the core reactor can be significantly reduced by using Kriging optimization method, which provides important guiding significance for the vibration reduction optimization of the core reactor.