第１部門①

[2A01-06-04] MACHINE LEARNING BASED NONLINEAR PARAMETER IDENTIFICATION FOR HIGH DAMPING RUBBER BEARINGS

*Katrina Mae Santiago Montes¹, Ji Dang¹, Yuqing Tan², Akira Igarashi², Takehiko Himeno³ (1. Saitama University, 2. Kyoto University, 3. Kawakin Core Tech Co.)

Keywords:Machine Learning, High Damping Rubber Bearing , Modified Bout-Wen Model, Nonlinear Parameter Identification

The complicated nonlinear behavior of high damping rubber bearings need deliciated nonlinear hysteresis model with multiple parameters. Those parameters need to be identified from experiments such standard quasi-static loading tests. However, general nonlinear model parameter identification methods such as curve fitting, Newton’s methods et. al. may need to select initial parameter value carefully. Sometimes it is a trial-and-error process and requires engineer’s expertise. This study trained an ANN model that aimed to identify the nonlinear parameters of an HDR-S bearing under Modified Bouc-Wen (MBW) model at ambient and low temperature. The suggested parameters were compared to an actual quasi-static experiment result and hybrid simulation.