Applications of fault detection techniques in industrial environments are increasing in order to improve the operational safety, as well as to reduce the costs related
to unscheduled stoppages. Although there are numerous proposals in the literature about fault detection techniques, most of the approaches demand extensive computational effort or even require too many thresholds or problem-specific parameters to be predefined in advance, impairing their use in
real-time applications. Aiming to overcome these problems, we propose in this paper an approach for real-time fault detection of industrial plants based on the analysis of the control and error signals, using recursive density estimation. Our proposed approach is based on the concept of the density in the data space, which is not the same as probability density function, but is a very useful measure for abnormality/outliers detection. The density can be calculated recursively, which makes it suitable for real-time environments. We define a criterion for density drop integral/sum, which is used as a problem- and user-insensitive (automatic) threshold to identify the faults/anomalies. In order to validate our proposal, we present experimental results from a level control laboratory
process, where control and error signals are used as features for the fault detection, but the approach is generic and the number of features can be significant due to the computationally lean methodology, since covariance or more complex calculations are not required. The obtained results are encouraging when compared with the traditional statistical approach.