Diagnosis, prognosis and fault accommodation for dynamical systems

Abstract

The prevention and prompt correction of faults are of paramount importance in the operation of critical systems such as medical equipments, aircrafts, chemical plants and nuclear reactors. Apart from possible environmental impacts and safety of the operators and users, the improvements on reliability and ready management of possible system faults can yield reduced operational costs, limits losses due to material damages and augments the availability of eventually highly valuable equipments. In this context, the present project proposes to investigate new methods for the diagnosis, prognosis and accommodation of faults that are applicable to dynamical systems. For instance, the automated detection of faults allows one to immediately alter the operation mode to keep a certain degree of functionality of the system, or, otherwise, trigger an alarm. Even if the equipment stops operating, the availability of an efficient diagnosis mechanism can be of value to promote easy location of the faulty component, so as to reduce the maintenance time-lag. It is also possible to design the equipment in such a manner that it becomes robust to some types of faults or, alternatively, be endowed with a capability to reconfigure itself to tolerate faults. In other applications, methods for fault prognosis can be exploited in the scheduling of preventive or condition based maintenance procedures, via close monitoring of trends related to degeneration of some components and devices. The multidisciplinary perspective required to study the subject leads, in a natural way, to a strong and synergic action of several research groups with mixed and complementary backgrounds. More specifically, the assembled research team has a wide variety of competence that spans from signal processing to artificial intelligence and from system theory to statistics. (AU)