Optimal variable selection for effective statistical process monitoring

• Does the dimensionality reduction offered by PCA eliminate the need to select variables for process monitoring?
• This paper establishes that variable selection is indeed necessary even with PCA models.
• It derives the theoretical conditions under which reduced PCA model based on a subset of measurements is beneficial compared to a full model that uses all available ones.
• A genetic algorithm based search strategy is proposed for identifying the optimal set of variables to be included in the reduced PCA model.
• For the Tennessee Eastman challenge problem, the reduced PCA leads to a 7% reduction in error rate and 60 samples reduction in detection delay.

In a typical large-scale chemical process, hundreds of variables are measured. Since statistical process monitoring techniques typically involve dimensionality reduction, all measured variables are often provided as input without weeding out variables. Here, we demonstrate that incorporating measured variables that do not provide any additional information about faults degrades monitoring performance. We propose a stochastic optimization-based method to identify an optimal subset of measured variables for process monitoring. The benefits of the reduced monitoring model in terms of improved false alarm rate, missed detection rate, and detection delay is demonstrated through PCA based monitoring of the benchmark Tennessee Eastman Challenge problem.