Out-of-Distribution Data Generation for Fault Detection and Diagnosis in Industrial Systems

The emergence of Industry 4.0 (I4.0) has transformed modern-day factories into expansive industrial sites through rapid automation and increased access to real-time data. Deep learning (DL) approaches possessing superior capabilities for intelligent data-driven fault diagnosis (FD) have become critical in ensuring process safety and reliability in these industrial sites. However, DL-based FD applications trained exclusively on in-distribution (ID) process data face challenges in the wake of previously unseen out-of-distribution (OOD) data in the real world. This paper addresses the challenge of OOD detection for DL-based FD models by generating synthetic OOD data to simulate real-world anomalies not present in the training set. Our research aims to enhance the robustness of DL-based FD systems against OOD data. We propose Manifold Guided Sampling (MGS), a data-driven method for generating synthetic OOD samples from the ID data-supporting manifold estimated through a deep generative model. Synthetic OOD data from MGS enhances the DL-based FD model capacity for estimating the prediction uncertainty by including insights from OOD data, thereby delivering safe and reliable DL-based FD models for real-world industrial process monitoring. Additionally, the MGS algorithm maintains ID data feature space as a reference point during data generation to ensure the resulting synthetic OOD data is realistic. We analyze the effectiveness of MGS through experiments conducted on the steel plates faults dataset and demonstrate that augmenting training data with synthetic OOD data from MGS enhances DL-based FD model performance in OOD detection tasks and robustness against dataset distributional shifts. The findings underscore the effectiveness of utilizing synthetic MGS-generated OOD data in scenarios where real-world OOD data is limited, enabling better generalization and more reliable fault detection in practical applications.