Abstract: To quantify the structural safety status and to perform health evaluation, a Multi-Physics Guided Neural Networks (MPGNN) method based on finite element synergy and on BWBN hysteresis modeling is proposed for the health evaluation of frame structures. The identification of BWBN hysteresis parameters of the finite element model is carried out by Bayesian Updating (BU) technique. Based on the relative stiffness level of the structure, the structural health degree under the finite element simulation and the BWBN hysteresis model is calculated, which is used to construct the finite element synergistic term and the physics theoretical term in the Multi-Physics Guided Block (MPG-Block), so as to guide the neural network to learn the physics laws in the finite element and in the BWBN hysteresis model. A MPGNN is formed. The overall and local monitoring indexes and the actual health degree of the structure are used as the input and output of the MPGNN, respectively, to realize the structural health evaluation. On this basis, the effectiveness of the method is verified by numerical simulation and experimental study respectively, and it has more accurate structural health evaluation effect compared with other classical models.