An Intelligent Fault Diagnosis and Accommodation Approach for Nonlinear Aircraft

Here Intelligent Automation, Inc. (IAI) and its subcontractor, Prof. Marios M. Polycarpou of University of Cincinnati, propose a new approach for designing intelligent health monitoring supervisor and fault-tolerant controllers for nonlinear air vehicles operating in multiple regimes. First, a novel on-line monitoring system performs model validation and fault diagnosis. The monitor automatically distinguishes between fault occurrence and normal operating regime switching and determines the particular fault type after fault detection. It can directly deal with nonlinear fault models, handle unstructured modeling uncertainty and unanticipated failures, and is suitable for real-time operations. Second, a new reconfiguration supervisor makes decision regarding controller reconfiguration and resource management. The supervisor provides a unified framework for fault detection, fault isolation and fault-tolerant control, which is still lacking in literature according to a recent survey paper. Another new feature is that the effect of fault detection time on system stability has been explicitly taken into account. Third, a controller suite with novel robust adaptive controllers to deal with different failure modes in each regime is proposed. The fault-tolerant controllers are designed based on recently developed nonlinear adaptive/neural control techniques. The controller guarantees closed-loop stability for a general class of nonlinear systems. Moreover, it has an on-line tuning scheme that eliminates off-line training of the neural net, which is extremely important in rapidly varying environment.