Distributional Smoothing in Bayesian Fault Diagnosis

Abstract

Abstract-Previously, we demonstrated the potential value of constructing asset-specific models for fault diagnosis. We also examined the effects of using split probabilities where prior probabilities come from asset-specific statistics and likelihoods from fleet-wide statistics. In this paper, we build upon that work to examine the efficacy of smoothing probability distributions between asset-specific and fleet-wide distributions to improve diagnostic accuracy further. In the current experiments, we also add environmental differentiation to asset differentiation under an assumption that data is acquired in the context of online health monitoring. We hypothesize that overall diagnostic accuracy will be increased with the smoothing approach relative to a fleet-wide model or a set of asset-specific models. The hypothesis is largely supported by the results. Future work will concentrate on improving the smoothing mechanism and in the context of small data sets.

Knowledge-Driven Board-Level Functional Fault Diagnosis

“…However, since different components (with nonuniform failure rates) are used in different production lines, and equipment used for testing can vary, training a diagnosis system using these aggregated data may lead to incorrect learning. Therefore, correction techniques based on conditional probabilities are used to alleviate data corruption [34].…”

Self-learning and adaptive board-level functional fault diagnosis

“…Several classification techniques (a sub-domain in machine learning) have been used for diagnosis system to achieve high diagnosis accuracy, e.g., artificial neural networks (ANNs) [16], [17], [23], Bayesian networks [13], [18], [27], [28], support-vector machine [23], [24], [29], decision tree [11], [26], weighted majority voting [23], etc. Work has been carried out on model-update (i.e., refinement) solutions for reasoning-based diagnosis engines [21], [27], [30].…”