Rail crack defect recognition based on a multi-feature fusion algorithm using electromagnetic acoustic emission technique

Multi-feature fusion has been widely used to enhance recognition accuracy for different health stages of rails, which may lead to high dimensionality and information redundancy of signals. In addition, conventional supervised methods require plenty of labeled samples with class information, which can take significant time and involve high economic costs. In order to improve the effectiveness of the electromagnetic acoustic emission technique in rail crack defect recognition, a novel method including multi-feature fusion based on weakly supervised learning and recognition threshold construction is proposed in this paper. First, a mechanism consisting of multi-feature extraction and feature selection is developed to fully reflect the information of different health stages of the rail and avoid interference caused by the ineffective features. Then, the effective features and a novel weakly unsupervised label are input into the self-normalizing convolutional neural network and long short-term memory model to construct the rail health indicator (RHI). Finally, the recognition threshold is calculated based on the characteristics of the RHI to achieve crack recognition automatically. Furthermore, the experimental results under different working conditions demonstrate that the proposed method achieves a higher recognition performance than other existing methods in rail crack defect recognition.