Topology Optimization Using Hyper Radial Basis Function Network

Abstract

Topology optimization has been conventionally used in automobile and aerospace industries to produce lightweight structures for minimum compliance or maximum fundamental frequency. Today, topology optimization finds application in solving diverse types of problems ranging from medicine to consumer products. Popular continuum based method SIMP fails to provide global solution due to too many variables involved. In this work, a novel hyper radial basis function network is presented as topology description function. The proposed approach provides drastic reduction in number of design variables involved in topology optimization. This makes the use of heuristic global solution possible. Parallel processing, reanalysis formulation and automatic selection of optimal volume fraction are proposed to speed up analysis. Application to solve minimum compliance, maximum fundamental frequency, compliant mechanisms and optical medical imaging problems are presented. The proposed approach can be easily incorporated with legacy software. The design process is made efficient by speeding up analysis and decreasing the need for human interpretation.