Multivariable model validation in the presence of time-variant disturbance dynamics

This paper is concerned with dynamic model validation through detection of parameter changes using the local detection approach. The local approach has ability to detect small changes very effectively. To enhance its robustness in the presence of time-variant disturbance dynamics, the local approach based on the output error identification algorithm is proposed. The effectiveness and robustness of the proposed method are illustrated through Monte-Carlo simulations. The result is also extended to multivariable model validation and verified on a multivariable pilot scale process. (C) 2000 Elsevier Science Ltd. All rights reserved.