Pipeline leak diagnosis based on leak-augmented scalograms and deep learning

This paper proposes a new framework for leak diagnosis in pipelines using leak-augmented scalograms and deep learning. Acoustic emission (AE) scalogram images obtained from the continuous wavelet transform have been useful pipeline health diagnosis, particularly when combined with However, background noise has significant impact on AE signals, which can reduce accuracy of identification classification models. To address this issue, type called is introduced, enhances variation colour intensities images. The are by pre-processing them image-enhancing Gaussian Laplacian filters. proposed method utilizes convolutional neural networks (CNNs) autoencoders (CAEs) feature extraction. CNN extracts patterns specific to local changes, while CAE holistic scalograms. resulting susceptible indicators merged into single pool provided as input shallow artificial network (ANN) evaluate conditions. achieves high well accuracy, precision, F-1 Score recall, compared existing state art methods.