A qualitative shape analysis formalism for monitoring, control loop performance

Optimal performance of controllers and control loops is crucial for process economy, quality and safety in chemical plants. Industrial statistics show that often a significant percentage of them are performing sub-optimally at any given time. Effective real-time monitoring of control loops is a difficult task as there may be dozens of loops to monitor in a typical process. In addition, abnormal or suboptimal performance is often not apparent under cursory inspection. Hence, automated approaches for the real-time monitoring of control loop performance is of considerable insterest. In this paper, we propose an automated qualitative shape analysis (QSA) formalism for detecting and diagnosing different kinds of oscillations in control loops. We extend our earlier QSA methodology to make it more robust by developing an algorithm for automatic identification of the appropriate global time-scales. We demonstrate this formalism on three case studies to detect and diagnose control loop oscillations.