In this paper, a new monitoring system is proposed by connecting different research areas, such as statistical
monitoring, as well as knowledge-based and history-based systems. Tools such as adaptive principal components
analysis (APCA), fuzzy-logic (FL) methods, and artificial neural network (ANN) methods are integrated to
develop an efficient fault detection, isolation, and estimation (FDIE) system, especially for large chemical
plants. It is capable of detecting, classifying, and estimating several faulty process elements. The information
given by this new monitoring system is able to support the proper decisions for connecting and transforming
an existing decentralized control strategy to a fault-tolerant method, based on an on-line reconfiguration.
Thus, the obtained FDIE system is a valuable tool that is able to improve the overall performance of large
and complex nonlinear controlled plants. In this case, inherent faults in sensors and actuators are analyzed.
The FDIE system is tested for single as well as sequential abnormal events on a pulp mill benchmark, which
is one of the biggest processes in the fault-tolerant control (FTC) that is integrated into the FDIE areas analyzed
in the literature. A complete set of simulation results, evaluated by different indexes, together with cost analysis
about the process operational profits with and without an FDIE system, are used here, to demonstrate the
effectiveness of the proposed methodology.
A reconfigurable fault-tolerant control system typically includes a nominal controller (NC), a fault detection/ diagnosis (FDD) and decision subsystem, a reconfiguration mechanism, and a reconfigurable controller (RC). Here, a systematic methodology for designing a fully decentralized NC of reduced dimension is presented, providing (i) fault-tolerant capability due to the structural flexibility and (ii) availability of redundancies for RC design. The fulfillment of a sufficient condition for decentralized integral controllability is searched to guarantee the stability while the FDD scheme is identifying the faults. A novel framework based on a genetic algorithm is developed for obtaining alternative NCs. They are screened considering quantitative measures derived from stability and performance considerations. The procedure features complexity reduction because (i) it only utilizes steady-state process information and (ii) it is independent from the controller design. The methodology is tested in the Tennessee Eastman process to demonstrate its potential against set point/disturbance changes and stuck actuator faults.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.