International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence (ICSMD2023)

120 / 2023-09-20 15:03:08

Efficient Planning in Spacecraft Telemetry Task Flow for Real time Fault Detection and Isolation Based on Directed Graph Topology Sorting

fault detection and isolution,topological sort,Pruning

Special Sessions > SS16 - Multi-Physical Sensors-Based Detection & Diagnosis Solution Empowered by AI Technology

To ensure that on-orbit spacecraft can conduct real-time fault detection and isolation timely facing high-frequency telemetry data with limited computing resource, it’s important to plan task flow efficiently. However, traditional methods like inferring relation simplifying and forward reasoning waste lots of calculation resource and restrict performance of real-time FDI, as the absence of middle result re-use and pruning for unnecessary calculation. This paper proposes an efficient planning in spacecraft telemetry task flow for real-time fault detection and isolation based on directed graph topology sorting to improve the performance. First, in this method, direct acyclic graph is used to construct comprehensible fault propagation and extract minimum correlative graph which prunes redundant computation based on traversal of successor nodes in constructed graph. Second, task flow planning method based on topology sorting can ensure both inferring order correctness and middle node re-use. Final, every task flow is saved as map to be searched and concatenate as task flow for one telemetry frame including arbitrary multiple parameters to prune more. An experiment is conducted in a power supply system of a spacecraft and the comparative advantage to traditional methods is proved.