Genetic fuzzy system for damage detection in beams and helicopter rotor blades

Abstract

Structural damage detection is an inverse problem of structural engineering having three main parts: finding the existence, location and extent of damage. In this study, a genetic fuzzy system is used to find the location and extent of damage. A finite element model of a cantilever beam is used to calculate the change in beam frequencies because of structural damage. Using these changes in frequencies, a fuzzy system is generated and the rule-base and membership functions optimized by genetic algorithm. The output faults of the fuzzy system are four levels of damage (undamaged, slight, moderate, and severe) at five locations along the beam (root, inboard, center, outboard and tip). The genetic fuzzy system developed for a noise level of 0.20 in the data gives a fault isolation success rate of 99.81% when the first eight natural frequencies are used. With noise level of 0.15 and 0.10, accuracy rates of 100% are obtained even when only the first four natural frequencies are used. The fuzzy system also shows excellent robustness with missing measurements and degrades gradually in the presence of faulty sensors/measurements. The genetic fuzzy system allows easy rule generation for different structures and when the number of inputs and outputs increase, thereby avoiding the ‘curse of dimensionality’ that plagues fuzzy systems. Results with a non-uniform beam and a finer output set of damage at 10 locations in the beam also show excellent results. The genetic fuzzy system also gives very good results for BO-105 hingeless helicopter rotor blade for frequency as well as mode shape-based data. The genetic fuzzy logic system in this study is proposed as a method for automatic rule generation in fuzzy systems for structural damage detection.