A fault detection approach based on machine learning models
Chapter in Scopus
- Additional Document Info
- View All
We present a new approach for process fault detection based on models generated by machine learning techniques. Our work is based on a residual generation scheme, where the output of a model for process normal behavior is compared against actual process values. The residuals indicate the presence of a fault. The model consists of a general statistical inference engine operating on discrete spaces. This model represents the maximum entropy joint probability mass function (pmf) consistent with arbitrary lower order probabilities. The joint pmf is a rich model that, once learned, allows one to address inference tasks, which can be used for prediction applications, In our case the model allows the one step-ahead prediction of process variable, given its past values, The relevant past values for the forecast model are selected by learning a causal structure with an algorithm to learn a discrete bayesian network. The parameters of the statistical engine are found by an approximate method proposed by Yan and Miller. We show the performance of the prediction models and their application in power systems fault detection. © Springer-Verlag Berlin Heidelberg 2005.