Application of Genetic Algorithm for Fault Detection in Cracked Composite Structure

Machines and beam like structures used in various industries require continuous monitoring for the fault identification for ensuring uninterrupted service. Different non-destructive techniques (NDT) are generally used for this purpose, but they are costly and time consuming. Vibration based methods can be useful to detect cracks in structures using various artificial intelligence (AI) techniques. The modal parameters from the dynamic response of the structure are used for the purpose. In the current analysis, the vibration characteristics of a glass fiber reinforced composite cracked cantilever beam having different crack locations and depths have been studied. Numerical and finite element methods have been used to extract the diagnostic indices (natural frequencies, mode shapes) from cracked and intact beam structure. An intelligent Genetic Algorithm (GA) based controller has been designed to automate the fault identification and location process. Single point crossover and in some cases mutation procedure have been followed to find out the optimal solution from the search space. The controller has been trained in offline mode using the simulation and experimental results (initial data pool) under various healthy and faulty conditions of the structure. The outcome from the developed controller shows that the system could not only detect the cracks but also predict their locations and severities. Good agreement between the simulation, experimental and GA controller results confirms the effectiveness of the proposed controller.