IFIR   05409
INSTITUTO DE FISICA DE ROSARIO
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Robust Adaptive Predictive Fault-Tolerant Control Linked with Fault
Autor/es:
ZUMOFFEN D., BASUALDO M., JORDAN M. AND CECCATTO A.
Lugar:
San Diego California-Estados Unidos de América
Reunión:
Conferencia; Conference on Decision and Control; 2006
Institución organizadora:
IEEE
Resumen:
Abstract—In this paper, a new fault-tolerant control (FTC) strategy is proposed to compensate for the effects of faults and to improve the overall performance for nonlinear stable plants. The FTC approach involves a commutation of a linear timevarying robust filter in the feedback path of the control loop in synchronization with an adaptive predictive controller. The decision of which mode has to work is based on specific conditions accounted in the control algorithm. In addition, the faulttolerance ability of the controller has been further improved by a fault-diagnosis system (FDS) designed for detecting two fault types: specific bias at the sensor and time delay at the actuator. The FDS is developed by using wavelet decomposition and identification techniques respectively. An implementation on a continuous stirred tank reactor (CSTR) with jacket is presented as a simulation example for analyzing the effectiveness of the FTC linked with FDS methodology proposed here.—In this paper, a new fault-tolerant control (FTC) strategy is proposed to compensate for the effects of faults and to improve the overall performance for nonlinear stable plants. The FTC approach involves a commutation of a linear timevarying robust filter in the feedback path of the control loop in synchronization with an adaptive predictive controller. The decision of which mode has to work is based on specific conditions accounted in the control algorithm. In addition, the faulttolerance ability of the controller has been further improved by a fault-diagnosis system (FDS) designed for detecting two fault types: specific bias at the sensor and time delay at the actuator. The FDS is developed by using wavelet decomposition and identification techniques respectively. An implementation on a continuous stirred tank reactor (CSTR) with jacket is presented as a simulation example for analyzing the effectiveness of the FTC linked with FDS methodology proposed here.