A fuzzy systems approach to structural damage detection and identification
Although improved design methodologies have significantly enhanced the reliability and safety of structures in recent years, it is still not possible to build structures that are infallible. There is an increasing interest in the development of smart structures with built-in fault detection systems that would provide failure warnings. This thesis presents general methodologies for structural damage detection and assessment using fuzzy systems theory. The approach to damage detection is based on monitoring various system responses to determine the health status of a structure or machine. Fuzzy logic rules encoded in a fuzzy associative memory are used to identify the location and extent of the damage in terms of stiffness reduction. The use of both global and local responses, including eigenvalues, static displacements and strain measurements are investigated. The damage detection methodology is demonstrated through examples including a simple fixed-free beam, truss structures and an aircraft wing model. Once damage is detected, it must be identified and assessed. Fuzzy computational methods for translating fuzzy stiffness reduction into fuzzy geometric damage parameters are presented. These methods are based on fracture mechanics for crack damage and a simple hole damage model for more general forms of damage. Several examples of determining a fuzzy crack length for a given fuzzy reduction in local stiffness are presented, including edge cracked beams and plates. The application of the hole damage model is demonstrated in the analysis if damaged flat rectangular and cylindrical shell membrane components. Damage tolerance techniques based on fuzzy parameters are developed for damage assessment. The criticality of fuzzy fractures is evaluated through comparisons of fuzzy stress intensities and fracture toughness measures. Similarly, the residual strength of damaged structures are assessed through comparisons of fuzzy stress and strength values. The notion of a fuzzy factor of safety is also introduced. The proposed methodologies represent unique approaches to damage detection, identification and assessment that can be applied to a variety of structures used in mechanical, aerospace and civil engineering applications. ^
Major Professor: Singiresu S. Rao, Purdue University.
Engineering, Aerospace|Engineering, Civil|Engineering, Mechanical|Artificial Intelligence