577c Data-Based Fault Detection and Isolation Using Feedback Control: Output Feedback, Optimality and Asynchronous Measurements

Panagiotis D. Christofides¹, Ben Ohran², David Muñoz de la Peña², and James F. Davis³. (1) Department of Chemical and Biomolecular Engineering, Department of Electrical Engineering, University of California, Los Angeles, Los Angeles, CA 90095, (2) Department of Chemical and Biomolecular Engineering, University of California, Los Angeles, Los Angeles, CA 90095, (3) University of California - Los Angeles, Los Angeles, CA 90095

With billions in lost revenue each year in U.S. industries [1] due to abnormal events, the need for better methods for handling abnormal situations in chemical processes is critical. Fault-tolerant control (FTC) minimizes the economic, environmental, health and safety impacts in abnormal situations encountered in chemical processes by maintaining system stability in the presence of a fault. Fault detection and isolation (FDI) is a key element of successful fault-tolerant control structures. In FTC, information obtained from an FDI scheme allows systems with redundant control and/or measurement devices to be reconfigured in order to maintain stability, thereby averting potential disaster (see, for example, [2]). As one of the most complex elements in fault-tolerant control, fault detection and isolation is an area of great importance that has received a lot of attention over the past years (see [3, 4] for reviews).

Recently, we introduced a method of fault detection and isolation that combines aspects of both model-based FDI and data-based FDI by using model-based control laws to induce an appropriate closed-loop system structure that allows fault detection and isolation based on available measured process data [5]. The present work focuses on extending these results to consider three new aspects. First, the case of output feedback control is considered. A continuously stirred tank reactor example is used to demonstrate how controller-enhanced FDI can be performed without full state information through the use of an appropriate state estimator. Second, the same model is used to demonstrate the use of model predictive control (MPC) as the external control law to optimize system performance. Third, the case where some of the state measurements are received asynchronously is considered and the cases in which it is possible or not possible to perform fault detection and isolation is explored. These results relax the requirement of full-state feedback requirement and demonstrate the general applicability of the controller-enhanced FDI method.

References

[1] Nimmo I.. Adequately Address Abnormal Operations /Chem. Eng. Prog.. /1995;91:36-45.

[2] Mhaskar P., Gani A., El-Farra N. H., McFall C., Christofides P. D., Davis J. F.. Integrated Fault Detection and Fault-Tolerant Control of Process Systems /AIChE Journal. /2006;52:2129-2148.

[3] Venkatasubramanian V., Rengaswamy R., Kavuri S.N., Yin K.. A review of process fault detection and diagnosis Part III: Process history based methods /Computers and Chemical Engineering. /2003;27:327-346.

[4] Venkatasubramanian V., Rengaswamy R., Yin K., Kavuri S.N.. A review of process fault detection and diagnosis Part I: Quantitative model-based methods /Computers and Chemical Engineering. /2003;27:293-311.

[5] Ohran B. J., Muñoz de la Peña D., Christofides P. D., Davis J. F.. Enhancing Data-based Fault Isolation Through Nonlinear Control /AIChE Journal. /2008;54:223-241.