Fault detection and diagnostics using data mining

Abstract

The purpose of data mining is to find new knowledge from databases in which complexity or the amount of data has so far been prohibitively large for human observation alone. Self-Organizing Map (SOM) is a special type of Artificial Neural Networks (ANNs) used in clustering, visualization and abstraction. In modern process automation systems, it is possible to collect and store huge amounts of measurement data. In this paper, SOM is used successfully to discover the base models from the automation system. Strategies based on data mining techniques are further developed for efficient fault detection and diagnostics. A semi-supervised anomaly detection technique is used with classification rules based on standardized data and domain experts’ analysis to construct the condition monitoring system.