A Novel Application of Rule Induction Algorithm for Parametric interpretation of Satellite On-orbit performance

摘要

This paper presents a novel method of applying Rule Induction mining algorithm for analysing the Telemetry Data received from a satellite in orbit. Health of a satellite in orbit is very important for it to deliver its intended space-based services. It is monitored by a set of house-keeping parameters which is received as telemetry by Satellite Ground Control Centre. This data is received continuously throughout the mission life of the satellite. Hence a huge repository of data is accumulated and archived. Analysis of this data by conventional means is tedious and time consuming. In addition, manual analysis can give information about related parameters. Data Mining techniques can be applied to explore the performance relationships by looking for unknown relationships between apparently unrelated telemetry parameters. The parameters of interest are collected in a multidimensional database. From the inputs provided, Rule-Induction Algorithm generates a set of rules which are found to be meaningful and giving new knowledge. The results obtained from the implemented algorithm show relationships between related and unrelated parameters that can lead to emergence of new knowledge consistent with satellite design philosophy.