Spacecraft Fault Detection Isolation System Design using Decentralized Analytical Redundancy

摘要

Fault detection and isolation functionality constitutes a critical element of spacecraft fault protection system capabilities. The rule based FDI schemes currently implemented for FPS suffer from opacity of design and behavior due to their reliance on mappings from observed symptoms to probable diagnosis. While model based diagnosis (MBD) techniques can resolve a number of these issues, their applicability to spacecraft has been limited until now largely due to an unfavourable cost-benefit analysis. We examine and compare the net values addition with different techniques of MBD. To address the issues with conventional FPS design while ensuring a better cost benefit tradeoff, we present an approach integrating MBD into a conventional spacecraft FPS architecture. A systematic approach to designing decentralized model based diagnosers for spacecraft is discussed. The proposed diagnoser architecture is decentralized and hierarchically scalable. Analytical redundancy relation based error monitors and activation rules relying on fault signatures are derived during the design phase. The transfer of diagnoser design and development earlier in the systems engineering process is facilitated. We demonstrate the design of such a diagnoser for the Cassini attitude & articulation control system. A comparison with the FDI functionality as currently implemented in the Cassini FPS is presented in terms of design effort and system structure.