scope:

INTRODUCTION

Automated fault detection and diagnostics (FDD) can improve reliability, reduce operating costs, and increase equipment safety in many applications. Most of the previous research and development of FDD techniques has focused on critical processes, such as aircraft engines (Onken and Stuckenberg 1979, Patton et al. 1987), nuclear power plants (Kitamura 19801, and the space shuttle main engine (Cikanek 1986. Duyar and Merill 1992). However, the costs of applying FDD techniques are diminishing with reduced hardware and software costs, and FDD techniques are beginning to be applied to heating, ventilating, air conditioning, and refrigeration (HVAC&R] applications.

Many authors (Willsky 1976, Isermann 1984, Frank 1987. Basseville 1988. Gertler 1988, Frank 1990) have offered excellent review papers on fault detection and diagnostic techniques. isermann (1984) presents the application of FDD techniques as a series of four steps termed "process supervision": these are illustrated in Figure 1. The first step is fault detection, in which a fault is indicated when the performance of a monitored system has deviated from expectation. The second step, diagnosis, determines which malfunctioning component is causing the fault. Following diagnosis, fault evaluation assesses the impact of the fault on system performance. Finally, a decision is made on how to react to the fault. This is usually a choice between tolerating the fault, repairing it as soon as possible, adapting the control, and stopping operation until repair is complete.