Autotunning and robustness of multi-loop pid-controlled mulltivariable processes

Abstract

In this study, decentralized PID controllers which are commonly used in process industry are applied to multi-input multi-output systems. A short literature survey is done and basic characteristics of these systems with their control procedure options are presented. A MATLAB based software to automate the previously developed multivariable decentralized PID tuning is developed. Autotuning is based on the multivariable version of relay feedback method. Structured singular value m, m-optimal tuning, m-interaction measure and robustness properties of closed loop multivariable systems are also investigated. Autotuning of multivariable systems is done, and then PID controller parameters are found using different tuning methods. Five systems are simulated with three different tuning methods. Their time responses are compared with each other. Robustness of the controllers is investigated. Results show that Biggest Log Modulus Tuning method does not give satisfactory responses, requires a model and is sensitive to uncertainties. Shen and Yu's tuning method does not require a model, is easy to apply to process systems, gives reasonable time responses and is robust against small uncertainties and perturbations. m-optimal tuning gives best results, but requires extensive computation.