Novel approach of satellite health monitoring, diagnosis and prediction via PLS batch modelling

Satellite health analysis executed to guarantee that service systems would keep up satellite's functionality all over the mission time. Monitoring, early anomaly detection and diagnosis are the main concerns of space operations. Service systems analysis specialist deals with massive amount of telemetry data that downloaded from satellite. The ground segment has multiple means to perform the task of telemetry monitoring and analysis. Multivariate analysis introduced as a powerful method in this field. In this paper, a survey on multivariate techniques (PCA-PLS) explained. Then the technique of batch modeling applied for first time to the telemetry of the thermal control system of Nickel-Hydrogen batteries in the power supply system of real satellite during test of batteries. Batch modeling is a partial least squares projections to latent structure (PLS) technique. This technique based on developing a model of evolution for normal batches; the model used for monitoring the new batches, and classifies them as normal or faulty ones. In addition, control charts such as CUSUM and Shewhart used for monitoring and early detection of anomalies. The discussion shows that using these techniques could avoid the failures that had happened if it used. In addition, contribution and variable patch plots used in diagnosis and physical interpretation.