Delamination Diagnosis in Composite Beam Using AIS and BGA Algorithms Based on Vibration Characteristics

Abstract

In this study vibration-based delamination detection is achieved using the artificial immune system (AIS) method. The approach is based upon mimicking immune recognition mechanisms that possess features such as adaptation, evolution, and immune learning. The identification of the delamination location and its length in the composite beam is formulated as an optimization problem. The cost function is based on differences between analytical or experimental natural frequencies and predicted natural frequencies by AIS method. Analytical natural frequencies of delaminated beam are obtained from Euler-Bernoulli beam theory with constrained delamination mode. Also, in this paper the binary genetic algorithm (BGA) was applied to compare the predicted locations and lengths with those obtained from AIS method. Errors of predicted location and length are 2.16 % and 0.1968 % respectively, using the AIS and these values become $-$3.83 % and 1.76 % for the BGA. The proposed approach (AIS) showed significantly better performance in detecting failures in comparison with the other method (BGA). In addition, detection accuracy and prediction errors, calculated with variance account for (VAF) and mean square error (MSE) concepts are compared with different artificial neural networks (ANNs) methods. To investigate the accuracy of the proposed method, some experimental results were obtained. A laser vibrometer was used to identify natural frequencies change in delaminated carbon-fiber-reinforced polymer (CFRP) composite beam case study. The average error values of predicted location and length in experiment test are 18.7 % and 9.8 %, respectively.