(721a) Handling Preventive Sensor Maintenance in Economic Model Predictive Control of Nonlinear Chemical Processes
AIChE Annual Meeting
2014
2014 AIChE Annual Meeting
Computing and Systems Technology Division
Economics and Process Control
Thursday, November 20, 2014 - 3:15pm to 3:37pm
Traditionally, fault detection schemes are used to identify faulty sensors [1]-[2]. Essential to the objectives of next-generation advanced manufacturing is the development of tightly integrated components of process systems (i.e., process control, process economics, scheduling) which are proactive instead of reactive. One potential area of interest is to integrate process economics, control, and preventive maintenance. Preventive maintenance plays an important role in maintaining a high degree of reliability of a process system and thus, reducing the changes of a failure or fault. For example, as process sensors age and becomes sluggish (e.g., due to gradual damage), it becomes necessary to conduct preventive maintenance (e.g., recalibration, verification, or replacement) on sensors to avoid process faults or failures. The driving force behind integrating control and maintenance is to avoid the need for a process shutdown required to complete a preventive maintenance task. To accomplish this goal, the process control system should explicitly account for a preventive maintenance task which may take process control components (e.g., actuator or sensor) off-line for a period of time. In a previous work, we considered integrating preventive maintenance of control actuators, process control, and process economics through the design of an economic model predictive control (EMPC) scheme [3]. However, we did not consider handling sensor maintenance and thus, the consideration of sensor maintenance is an open problem.
This work focuses on developing a state-estimation-based Lyapunov-based economic model predictive control (LEMPC) method (e.g., [3]-[4]) that can handle preventive sensor maintenance. To complete the sensor maintenance task, the sensor is taken off-line until the maintenance task is completed. Since a measurement from the sensor is not available during this time, a certain observability assumption is imposed on the process with the remaining sensors. A state estimator is used to estimate the current state when the sensor is taken off-line and provide the state estimation to the LEMPC. For guaranteed closed-loop stability as well as to provide an accurate state estimate to the EMPC, a robust moving horizon estimator (RMHE) is used [4]. The approach is applied to a chemical process example under the proposed LEMPC to demonstrate that the LEMPC is able to maintain stability of the process when a sensor is taken off-line for preventive maintenance.
[1] Qin SJ, Li W. Detection and identification of faulty sensors in dynamic processes. AIChE Journal. 2001;47:1581-1593.
[2] Mehranbod N and Soroush M and Panjapornpon C. A method of sensor fault detection and identification. Journal of Process Control. 2005;15:321-339.
[3] Lao L, Ellis M, Christofides PD. Smart Manufacturing: Handling preventive actuator maintenance and economics using model predictive control. AIChE Journal. 2014;60:2179-2196.
[4] Ellis M, Zhang J, Liu J, Christofides PD. Robust moving horizon estimation based output feedback economic model predictive control. Systems \& Control Letters. 2014;68:101-109.