(546c) Performance-Based Accommodation of Sensor Faults in Output Feedback Sampled-Data Control Systems with Measurement Delays | AIChE

(546c) Performance-Based Accommodation of Sensor Faults in Output Feedback Sampled-Data Control Systems with Measurement Delays

Authors 

Allen, J. - Presenter, University of California, Davis
El-Farra, N., University of California, Davis
The ability to compensate for control system faults and malfunctions is an important capability for modern-day process control systems. When left unaddressed, faults in the control system components, such as the measurement sensors and the control actuators, can lead to substantial degradation in the achievable closed-loop control quality, which can translate into economic losses through product quality deterioration and, in some cases, safety hazards due to process instabilities and loss of controllability. The increased emphasis placed on process safety and meeting stringent product quality requirements in industrial applications have motivated a significant and growing body of research work on the problem of fault-tolerant control (e.g. [1]-[6]).

Compared with the extensive body of research work on the problem of handling control actuator faults, the problem of sensor faults has received less attention even though sensor faults are commonplace in practice and are critical for the overall process performance, especially with the increased reliance on dense sensor deployment and sensor networks in many industrial applications. The measurement errors induced by sensor faults may deteriorate the overall monitoring and control quality and need to be accounted for explicitly in the design and operation of the fault-tolerant control system.

The problem of designing fault-tolerant control systems that account explicitly for sensor faults has been the focus of prior works. For example, in [7] an approach was developed to assess the fault accommodation logistics of a multi-rate sampled-data control system with an observer-based output feedback controller when implemented subject to measurement sensor faults. The sensor faults were introduced as deviations in the measurement sampling rates and were accommodated by means of control actuator configuration switching (thus changing the manipulated input), with emphasis on obtaining the most robustly stable post-accommodation operational configuration with respect to future deviations in sampling rates. In order to assess the stability of the pre- and post-fault systems, an explicit characterization of the stability region of the sampled-data closed-loop system was obtained in terms of the process parameters such as the sampling rates, the fault magnitudes, and the choice of manipulated input.

The developed methodology was subsequently extended in [8] to develop performance-based fault accommodation strategies that account not only for closed-loop stability but also for closed-loop performance considerations in the selection of the fault accommodation measure. In this study, sensor faults were introduced both as errors in the sensor readings as well as variations in the sampling rates. A performance metric, based on the notion of the extended H2-norm, was introduced and characterized explicitly by the same process parameters used to characterize the closed-loop stability region, thus allowing direct comparison and visualization of the interplay between the various process parameters and their effects on both the stability and performance of the closed-loop system. Under the prior stability-based accommodation approach there was only a binary metric to determine the "fitness" of a candidate accommodation measure, and through the introduction of a performance metric this binary assessment of the accommodation measure expanded into a continuous spectrum of values characterizing the closed-loop system performance.

A key aspect of these prior works is the assumption that all measurements of the process states were instantaneously made available for use in determining the appropriate control actions. The nuances associated with measurement delays were therefore not considered in the development or implementation of the proposed fault accommodation strategies. Specifically, the errors introduced in the feedback loop due to the lack of timely state measurements can potentially erode the achievable stability and performance capabilities of the fault-tolerant control system, and must therefore be explicitly addressed. Furthermore, the inaccessibility of the full-state for measurement imposes limitations on the practical implementation of the fault-tolerant control system and must be addressed as well.

At this stage, a unified framework for performance-based fault accommodation in sampled-data process systems with sensor faults, delayed and incomplete state measurements remains lacking. This is largely due to the fact that the explicit characterization of the closed-loop performance of such systems has not been addressed. While previous work has looked into the performance characterization of systems with sampled and delayed measurements [9], the resulting methods do not explicitly account for sensor faults or the lack of full-state measurements. This contribution aims, in part, to bridge this gap.

In this work, we focus on the problem of fault accommodation in sampled-data process systems subject to measurement sensor faults, delayed and incomplete state measurements. Initially, an observer-based output feedback controller that includes an inter-sample model predictor, a state observer and a propagation unit, is designed. The inter-sample model predictor aims to compensate for the unavailability of continuous output measurements between sampling times. The output of the model is used by the observer to generate an estimate of the state which is then used to compute the control action when the measured output is unavailable from the sensors. To compensate for the effects of measurement delays, the propagation unit is used to provide, based on the delayed output, an estimate of the current output which is then used to reset the model output at discrete times.

The stability and performance properties of the sampled-data closed-loop system are analyzed, and conditions that explicitly characterize those properties in terms of the sensor fault size, the sampling rate, the delay size, as well as the controller and observer design parameters are obtained. Based on these characterizations, stability-based and performance-based fault accommodation strategies that explicitly account for measurement delays are devised. The results are illustrated through an application to a chemical process example, where the impacts of discrete measurement sampling and measurement delays on the fault-tolerance capabilities of the control system are explored, and the efficacy of different delay-compensation schemes is assessed.

References:

[1] S. Simani, C. Fantuzzi, R. Patton, Model-based Fault Diagnosis in Dynamic Systems Using Identification Techniques, Springer, London, 2003.

[2] M. Blanke, M. Kinnaert, J. Lunze, M. Staroswiecki, Diagnosis and Fault-Tolerant Control, Springer, Berlin-Heidelberg, 2003.

[3] M. Isermann, Fault-Diagnosis Systems: An Introduction from Fault Detection to Fault Tolerance, Springer, Berlin, 2005.

[4] P. Mhaskar, J. Liu, P. D. Christofides, Fault-Tolerant Process Control: Methods and Applications, Springer-Verlag, London, 2013.

[5] S. Ghantasala, N. H. El-Farra, "Active Fault-Tolerant Control of Sampled-Data Nonlinear Distributed Parameter Systems," International Journal of Robust and Nonlinear Control, 22, 24-42, 2012.

[6] T. Napasindayao, N. H. El-Farra, "Fault Detection and Accommodation in Particulate Processes with Sampled and Delayed Measurements,” Industrial Engineering and Chemistry Research, 52, 12490-12499, 2013.

[7] T. Napasindayao, N. H. El-Farra, “Sensor Fault Accommodation Strategies in Multi-Rate Sampled-Data Control of Particulate Processes," Proceedings of 10th IFAC International Symposium on Dynamics and Control of Process Systems, pp. 379-384, 2013.

[8] J. Allen, S. Chen, N. H. El-Farra, “Model-Based Strategies for Sensor Fault Accommodation in Uncertain Dynamic Processes with Multi-Rate Sampled Measurements," Chemical Engineering Research and Design, 142, 204-213, 2019.

[9] J. Allen, N. H. El-Farra, “Performance-based Actuator Fault Accommodation in Sampled-Data Systems with Measurement Delays,” Proceedings of IEEE Conference on Decision and Control, submitted, 2019.