Affiliation:
1. , Xi’an Jiaotong University, , , China
Abstract
The optimization design of complex nonlinear structures mainly relies on expert experiences and trial and error. In this paper, we proposed an optimization design framework for nonlinear structures by combining experimental data and machine learning. The framework can search the entire design space and guide the next experiment by machine learning model until the optimization targets are met. To demonstrate the effectiveness and practicability of this framework, we have optimized the damping efficiency and principal resonance frequency (PRF) of an Electricity Distribution System (EDS) with eight rubber isolators. The results show that the design targets of the optimized structure are consistent with the experimental results after two iterations. This framework is able to guide and accelerate nonlinear structure design and has significant value for engineering applications.
Subject
Electrical and Electronic Engineering,Mechanical Engineering,Mechanics of Materials,Condensed Matter Physics,Electronic, Optical and Magnetic Materials