Complex probabilistic slow feature extraction with applications in process data analytics

Today, in modern industrial processes, thousands of correlated process variables are measured and stored. Dimension reduction techniques are often employed to construct informative features by discarding re-dundant information. Slow feature analysis is one such technique that extracts the slowly varying pat -terns from measured data. Oscillatory behaviour is prevalent in process data due to inadequate control loop tuning and external disturbances such as diurnal temperature variation. Extracting these oscillatory patterns is vital in applications such as control loop monitoring, fault diagnosis. Slow feature analysis may not extract oscillating patterns when the signal to noise ratio is low in process data. This paper pro-poses the complex probabilistic formulation that extracts slow oscillatory features. We also present the Expectation-Maximization algorithm to obtain the optimal parameter estimates. Finally, three case stud-ies are presented to illustrate the efficacy of the proposed formulation in soft sensing and fault detection applications. (c) 2021 Elsevier Ltd. All rights reserved.

Automated summary

In modern industrial processes, numerous correlated process variables are measured and stored, necessitating the use of techniques such as slow feature analysis to process the data and extract useful information while eliminating redundancy. Oscillatory behavior is commonly seen in process data, and extracting these patterns is crucial for control loop monitoring and fault diagnosis.