Structural Damage Identification Using Machine Learning Techniques: A Critical Review

Abstract

The importance of structural health monitoring has grown in response to the growing requirement for the safety and functionality of civil structures. Recent advances show that computational digital signal processing have enormous potential for developing more efficient, reliable, and robust structures damage identification systems. Artificial Neural Networks (ANNs) are computational models made up of many and highly interconnected processing elements that process information, establish complex, highly nonlinear relationships and associations from large datasets. The ability of the ANN is to train a given data set and predict missing data on that basis to makes an appealing proposition for knowledge acquisition for problems where there is currently not be acceptable theory. In the contemporary era of science and engineering, convolutional neural networks (CNN) and recurrent neural networks (RNN) are two of the most powerful tools. The present literature on the application and development of conventional neural networks (ANNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs) is important for structural health monitoring and damage detection is reviewed. Some significant issues with traditional damage identification systems can be solved, and damage detection accuracy can be enhanced, by using CNNs and RNNs.