Machine learning‐based approach for fatigue crack growth prediction using acoustic emission technique

Abstract

In this study, a general machine learning‐based approach is proposed for fatigue crack growth rate (FCGR) prediction using multivariate acoustic emission (AE) online monitoring data. To improve the prediction accuracy, a backpropagation neural network optimized by genetic algorithm (GA‐BPNN) is developed to describe the intricate link between the FCGR and multivariate input data. Several conventional machine learning models and the traditional FCGR prediction method based on the linear relationship between AE energy rate and FCGR are also used to examine the effectiveness of the proposed GA‐BPNN model. The results indicate that the developed GA‐BPNN exhibits higher accuracy and superior adaptability in predicting FCGR from unseen data than other methods. The findings of this study will provide a strategy for developing and optimizing a machine learning solution for FCGR prediction based on AE monitoring data and also aid in determining the most suitable feature for AE monitoring studies.

Review of the Uses of Acoustic Emissions in Monitoring Cavitation Erosion and Crack Propagation

“…The support vector machine (SVM) is an algorithm that constructs a hyperplane with the shortest possible distance between it and the sample points. It has exhibited a good generalisation capability and high prediction accuracy in cases of limited training datasets [41].…”

“…The authors highlighted that the data points were obtained from AE waveforms recorded during the entire fatigue load cycle without any load-based AE data filtering procedure. The same authors proposed, in a further study [41] that was carried out with the results of the same tests, a back propagation NN model optimised with a genetic algorithm (GA-BPNN) for fatigue crack growth rate (FCGR) prediction. This model is based on AE parameters that can be ordered according to how significant they are relative to the FCGR prediction model.…”

Characterization of Fatigue Damage in Hadfield Steel Using Acoustic Emission and Machine Learning-Based Methods

“…In the field of SHM, machine learning has the potential to deal with complex problems related to non-linear relationships and can help to improve prediction accuracy [ 24 ]. As such, many researchers started using a machine learning approach for the health monitoring of various structural systems in the last decade [ 23 , 25 , 26 , 27 , 28 ]. For instance, Sikdar et al [ 25 ] proposed a convolutional neural network (CNN)-based approach for identification of the region of damage in a composite panel.…”

“…Garret et al [ 23 ] utilized a combination of Choi–Williams transform (CWT) and convolutional neural network to predict fatigue-crack length in aluminum plates. Chai et al [ 28 ] developed a back propagation neural network optimized by genetic algorithm (GA-BPNN) for FCGR prediction based on AE monitoring data. It is noticed from the literature that despite the significance of identification of fatigue sub-stages, research is still missing that uses machine learning-based methods to identify fatigue sub-stages.…”

Structural reliability analysis of aircraft wing rib fatigue cracking using surrogate dynamic Bayesian network

“…Fatigue crack growth is a complex degradation process that is challenging to predict. 2 Furthermore, prediction of fatigue crack growth is affected by factors such as inspection reliability and prediction uncertainty. For these reasons, in-service aircraft rely on the overly conservative predictions of traditional aircraft structural integrity programs (ASIP).…”