AMPERES: A REAL-TIME FAULT MONITORING AND DIAGNOSIS KNOWLEDGE-BASED SYSTEM FOR SPACE POWER SYSTEMS

Abstract

A real-time fault monitoring and diagnosis knowledge-based system (KBS) has been developed for space power systems to save costly operational manpower and to achieve reliable space power system operation. The developed KBS, called AMPERES (autonomously managed power-system extendible real-time expert system), was developed using the autonomously managed power system (AMPS) test facility currently installed at NASA Marshall Space Flight Center (MSFC). The approaches taken for AMPERES are generic and should be applicable for other space power systems. A sensor value validation method based on causal relations (CRs) and their interrelations within the sensor redundancy graphs (SRG) has been developed and applied in AMPERES to increase the confidence in the validity of the sensor values and consequently in the results of the fault diagnosis. The validation results also yield essential information for fault diagnosis identifying the existence of a potential fault and the location or the extent of the fault. A disturbance interrelation analysis graph (DIAG) scheme has been developed to characterize the fault diagnosis problems clearly and to perform the fault monitoring and diagnosis tasks effectively by explicitly showing the diagnosis reasoning paths. The DIAG approach also enables the fault diagnosis KBS to describe a fault in several different abstraction levels and in different component granularities.