Real Time Prediction and Control of Spacecraft Health using Neural Networks
- Paper number
IAC-17,B6,IP,2,x41184
- Year
2017
- Abstract
Thermal control systems use temperature sensors and heaters to monitor and control spacecraft components' temperatures within operating limits. Heaters are usually implemented over components to maintain the package base temperatures. During mission, monitoring different components can be very complex especially when thermal control elements fail. In the event of off-planned manoeuvres, different strategies has to be planned beforehand. Post mission analysis can be very tedious given the volume of data available. This paper addresses on how to build an intelligent system using Neural Networks, capable of (1) identifying possible failures during mission & post mission (2) identify alternate control methods in case of failure (3) develop thermal control strategies for off-planned manoeuvres. Neural Networks is used to learn the thermal behaviour during testing and predict the temperatures of components in case of failure of any sensor. This is very helpful for future missions as it adapts to the environment and prior knowledge of the entire spacecraft thermal system is not mandatory. This neural network can be further expanded to include other mechanisms and mission operations.
- Abstract document
- Manuscript document
(absent)