Fully learnable deep wavelet transform for unsupervised monitoring of high-frequency time series.

High-frequency (HF) signals are ubiquitous in the industrial world and are of great use for monitoring of industrial assets. Most deep-learning tools are designed for inputs of fixed and/or very limited size and many successful applications of deep learning to the industrial context use as inputs extracted features, which are a manually and often arduously obtained compact representation of the original signal. In this paper, we propose a fully unsupervised deep-learning framework that is able to extract a meaningful and sparse representation of raw HF signals. We embed in our architecture important properties of the fast discrete wavelet transform (FDWT) such as 1) the cascade algorithm; 2) the conjugate quadrature filter property that links together the wavelet, the scaling, and transposed filter functions; and 3) the coefficient denoising. Using deep learning, we make this architecture fully learnable: Both the wavelet bases and the wavelet coefficient denoising become learnable. To achieve this objective, we propose an activation function that performs a learnable hard thresholding of the wavelet coefficients. With our framework, the denoising FDWT becomes a fully learnable unsupervised tool that does not require any type of pre- or postprocessing or any prior knowledge on wavelet transform. We demonstrate the benefits of embedding all these properties on three machine-learning tasks performed on open-source sound datasets. We perform an ablation study of the impact of each property on the performance of the architecture, achieve results well above baseline, and outperform other state-of-the-art methods.

Interpretable Detection of Partial Discharge in Power Lines with Deep Learning.

Partial discharge (PD) is a common indication of faults in power systems, such as generators and cables. These PDs can eventually result in costly repairs and substantial power outages. PD detection traditionally relies on hand-crafted features and domain expertise to identify very specific pulses in the electrical current, and the performance declines in the presence of noise or of superposed pulses. In this paper, we propose a novel end-to-end framework based on convolutional neural networks. The framework has two contributions: First, it does not require any feature extraction and enables robust PD detection. Second, we devise the pulse activation map. It provides interpretability of the results for the domain experts with the identification of the pulses that led to the detection of the PDs. The performance is evaluated on a public dataset for the detection of damaged power lines. An ablation study demonstrates the benefits of each part of the proposed framework.

Contrastive Learning for Fault Detection and Diagnostics in the Context of Changing Operating Conditions and Novel Fault Types.

Reliable fault detection and diagnostics are crucial in order to ensure efficient operations in industrial assets. Data-driven solutions have shown great potential in various fields but pose many challenges in Prognostics and Health Management (PHM) applications: Changing external in-service factors and operating conditions cause variations in the condition monitoring (CM) data resulting in false alarms. Furthermore, novel types of faults can also cause variations in CM data. Since faults occur rarely in complex safety critical systems, a training dataset typically does not cover all possible fault types. To enable the detection of novel fault types, the models need to be sensitive to novel variations. Simultaneously, to decrease the false alarm rate, invariance to variations in CM data caused by changing operating conditions is required. We propose contrastive learning for the task of fault detection and diagnostics in the context of changing operating conditions and novel fault types. In particular, we evaluate how a feature representation trained by the triplet loss is suited to fault detection and diagnostics under the aforementioned conditions. We showcase that classification and clustering based on the learned feature representations are (1) invariant to changing operating conditions while also being (2) suited to the detection of novel fault types. Our evaluation is conducted on the bearing benchmark dataset provided by the Case Western Reserve University (CWRU).