Attainment of PI Achievable Performance for Linear SISO Processes with Deadtime by Iterative Tuning

Abstract

industry can reap great economic benefi ts. It must also be pointed out that most of the industrial control loops are of PID type (about 90%) and subject to stochastic disturbances. It will be of practical value and interest to fi nd the maximum CLPI that can be achieved by a PI(D) controller in the control loop, which is referred to as PI(D) achievable performance, and attain it through proper tuning. Ko and Edgar (1998) proposed the Approximate Stochastic Disturbance Rejection (ASDR) technique to determine PI achievable performance. Their technique assumes the availability of a process model and uses routine operating data to approximately determine the noise model followed by estimation of the PI achievable performance. However, in many cases, the process model is not known. In such cases, it is diffi cult to fi nd the PI achievable performance using ASDR technique. Agrawal and Lakshminarayanan (2003) in their work do not assume a This work presents a practical scheme to attain PI achievable performance (the performance achievable with a proportional integral (PI) controller) for linear SISO processes with dead time driven by stochastic disturbances. It uses a combination of a quantitative measure called Control Loop Performance Index (CLPI) and signature plots of the closed loop disturbance impulse response (IR) to arrive at PI achievable performance in an iterative manner. The proposed methodology requires only routine operating data and knowledge of the process delay, and its effi cacy is illustrated through several case studies.On présente dans ce travail un schéma pratique pour atteindre la performance PI réalisable (performance réalisable avec un contrôleur à intégrale proportionnelle) pour des procédés SISO linéaires avec un temps mort induit par des perturbations stochastiques. Ce schéma utilise une combinaison d'une mesure quantitative appelée "Indice de performance à boucle de contrôle" (CLPI) et des graphes de la signature de la réponse à des pulses de perturbation en boucle fermée (IR) pour atteindre la performance PI réalisable de manière itérative. La méthodologie proposée requiert seulement des données opératoires de routine et la connaissance du retard dans le procédé. Son effi cacité est illustrée par plusieurs études de cas.

A novel pattern-based approach for diagnostic controller performance monitoring

“…Since the ACF is providing visual information about the controller response pattern, it is possible to build a visual guide for controller retuning. Iterative approaches to controller retuning based on controller performance metrics have been established (Goradia, Lakshminarayanan, & Rangaiah, 2005;Tan, Zhao, Chua, Ho, & Tan, 2006) and here a similar approach using a tuning map is suggested. Fig.…”

Running discrete Fourier transform and its applications in control loop performance assessment

“…This challenge was addressed earlier e.g. in [17], [18], [19], [20]. However, only the low order plant models are used there and the maximum achievable performance is computed numerically.…”

Networked Control systems based on novel Smith predictor and GPC

“…Considering both known and constant process delays with noise, a minimum variance control law [9] and a step-by-step tuning procedure [10] were developed separately to attain PI achievable performance for linear SISO time-delayed processes. Furthermore, extension of [9] was then developed to MIMO system [11].…”