Identification of Asymmetric Bouc–Wen Hysteresis Under Intense Noise by Only Measuring Acceleration

Abstract Parameter identification of hysteretic models is significant for predicting structural dynamic response in vibration isolation structures. However, quasi-static testing and displacement measurement methods are not convenient for assembly structures and sensor layouts. Moreover, the methods based on evolutionary optimization need to provide appropriate boundary conditions for convergence and efficiency. Therefore, a novel hybrid identification method that takes the advantage of physics-informed parameter constraints and only acceleration measurement is proposed to identify the asymmetric Bouc–Wen hysteresis model. The restoring force surface is constructed for hysteretic force extraction based on the measurement of base excitation and isolated mass acceleration. The polynomial fitting and limit cycle approach are utilized for physical information given of an improved Bouc–Wen model. Furthermore, the evolutionary algorithm based on parameter constraints is implemented for final parameter estimation. A numerical simulation of an asymmetric Bouc–Wen model shows that the proposed method can keep an normalized mean square error (NMSE) of 0.19% under the noise level of 30 dB. The reconstructed hysteresis loop keeps in good agreement with the theoretical one.