Diagnosing the health of engineering systems

Abstract

System reliability evaluation can be performed using the two broad categories of deterministic and probabilistic techniques. The basic weakness associated with deterministic methods is that they do not specifically take into account the likelihood of component failures in the evaluation. There is, however, considerable reluctance to using probabilistic techniques in many engineering areas owing to the difficulty in interpreting the resulting numerical indices. An approach is presented in this paper to alleviate this difficulty by combining deterministic considerations with probabilistic indices to monitor the system well-being in the form of system health and margin states in addition to a conventional risk index. The concept of system well-being is illustrated using relatively simple network configurations. The proposed technique is also applied to a power system network to determine the degree of well-being of the various load points and the overall system. © 1997 John Wiley & Sons, Ltd.