(601h) Performance Characterization and Fault-Tolerant Control of Multi-Rate Sampled-Data Process Systems with Unknown Measurement Delays

A key capability of modern-day process control systems is that of fault handling. Faults in process elements such as measurement sensors and control actuators if unaccounted for can result in process instabilities and more commonly loss of product quality. Naturally the importance of this problem has led to a vast and growing body of research on the topic of fault-tolerant control (e.g., see [1],[2],[3]). Conventional methods of addressing this problem include control formulations with a controller driving the process to a desirable operating point through use of continuous state measurements. However, with real-time operation practices the availability of state measurements is governed by the sampling rate of the measurement sensor. Sampled-data systems are found in industrial processes in places where continuous measurements of system data prove untenable due to physical or logistical constraints.

Fault-tolerant control of sampled-data systems has been the focus of prior works, where stability-based reconfiguration [4] and accommodation [5] approaches were developed to address the problem of control actuator fault handling in processes with known measurement delays. These approaches where extended in [6] to include fault estimation capabilities through use of an optimization-based scheme to aid in identifying the fault size and location, and facilitate the implementation of the fault accommodation logic. A framework for the integration of data-based fault detection and performance-based fault accommodation was subsequently developed in [7] and applied to a distributed energy system.

A key underlying consideration of prior works is that all process states are sampled at the same rate. In many practical settings, limitations on the sensing and processing capabilities of different sensors may result in different sampling rates. The use of multi-rate sampling can also be triggered by the relative importance of the measurements collected, where a fast sampling rate is applied to sensors placed at critical locations in the process, while reduced sampling rates are applied to less critical sensors to optimize energy resource consumption and reduce cost. A more robust framework for fault-tolerant control should therefore allow for the possibility of sampling each state at a different rate. Furthermore, the assumption that the measurement delays are fixed and known a priori needs to be re-examined as the controller would not realistically have access to the exact size of the delay which can also be time varying in general.

The objective of this contribution is to address the problem of performance-based actuator fault accommodation in multi-rate sampled-data process systems with unknown measurement delays. A key feature of this contribution is the explicit integration of online delay and fault estimation into the performance-based fault accommodation framework that would allow for the measurement delay and fault magnitudes to be estimated and used as parameters in the fault accommodation strategy. Based on the estimated delay and fault sizes, the potential fault accommodation measure can then be evaluated with reference to the baseline to determine if more drastic measures are needed or if the fault can be sufficiently mitigated through minor adjustments in the controller design parameters. The approach is implemented through leveraging the available sampled-data from the process to estimate the severity of the fault and the magnitude of the measurement delay through use of a data-based moving-horizon parameter estimation scheme. The accommodation scheme will then determine an appropriate response while using the estimated fault and delay values to meet performance baselines needed by the process and maintaining closed-loop stability. This will be achieved by characterizing the closed-loop stability region and the chosen performance metric of the system as functions of the fault size, the delay size, the frequencies of measurement sampling and other controller design parameters. These characterizations, which evolve in time as the estimated delay and fault values vary, provide the basis for selecting the appropriate accommodation strategy that accounts explicitly for both the faults and the delays in the process. The implementation of the developed fault-tolerant control approach will be demonstrated using a chemical process example.

References:

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

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

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

[4] Sun, Y. and El-Farra, N.H. (2011). Model-based fault detection and fault-tolerant control of process systems with sampled and delayed measurements. In Proceedings of 18th IFAC World Congress, 2749-2754. Milan, Italy.

[5] Napasindayao, T. and El-Farra, N.H. (2013). Fault detection and accommodation in particulate processes with sampled and delayed measurements. Ind. & Eng. Chem. Res., 52(35), 12490-12499.

[6] Napasindayao, T. and El-Farra, N.H. (2015). Model-based fault-tolerant control of uncertain particulate processes: Integrating fault detection, estimation and accommodation. In Proceedings of 9th IFAC Symposium on Advanced Control of Chemical Processes, 872-877. Whistler, Canada.

[7] Allen, J.T. and El-Farra, N.H. (2017). A model-based framework for fault estimation and accommodation applied to distributed energy resources. Renewable Energy, 100, 35-43.