| Monitoring and assessment of control systems have become an integral part of industrial process control applications due to their usefulness in meeting target objectives and increasing process productivity. In this work, we propose new approaches to controller monitoring by investigating the use of run lengths, Markov chains and ultimately, Bayesian analysis. Bayesian analysis is important for making decisions in the presence of uncertainty. Using model predictive controllers as a case study, we have addressed the issue of controller constraint tuning via a continuous-valued objective function within a Bayesian probabilistic framework. The benefits of this approach includes: a more generalized definition of quality variables; the development of a mathematically elegant formulation of the problem to address linear and quadratic objective functions, thereby obtaining closed form solutions; and maximum-likelihood location determination of the quality variables in the decision making process. The approaches are illustrated with simulations and pilot-scale experiments. |
