Statistical topology optimization scheme for structural damage identification

Abstract

Statistical topology optimization (TO) scheme, which statistically records and analyzes local optima at random frequencies of the excitation force, is developed for material damage and open crack. To identify the alternations of stiffness, mass, and strength due to the damages, the structural responses of healthy and suspicious systems can be compared and optimized by TO for fault identification. Owing to the local optima problem in TO, slight modifications in responses caused by relatively minimal damage, and inadequate comparison of insufficient vibration information, the application of TO to a damage identification problem is often complicated. To overcome this complication, this study establishes a new approach that records local optima (layouts) and performs a statistical process (clustering and averaging of layouts). To illustrate the concept of the present scheme, several examples with material damage and open crack are solved.