A dissipativity-based robust fault diagnosis approach for clusters of small satellites

Abstract

This study explores distributed fault diagnosis strategies for clusters of diverse small satellites, encompassing variations in Lipschitz-based non-linear dynamic systems and environmental factors. The investigation focuses on identifying rapid time-varying fault magnitudes and profiles across cluster agents interconnected by a directed graph. The analysis acknowledges uncertainties linked with state-coupled disturbances in the system model. Utilizing satellite communication topologies, a comprehensive augmented vector is formulated for individual satellites to diagnose both their states and faults, along with the states and faults of entire satellite clusters. Integrating an Unknown Input Observer (UIO) effectively untangles error dynamics from state-coupled disturbances. The UIO design is approached under two distinct conditions: one satisfying the case-matching criterion and the other that does not. The computation of observer gains employs the application of Linear Matrix Inequalities (LMIs). Furthermore, an adaptive robust fault estimator is proposed for scenarios where the matching condition is not met. This approach integrates dissipativity principles and is benchmarked against the H∞ framework. Simulation outcomes demonstrate the effectiveness of the proposed methodology. Specifically, the results show fault estimation across three clusters comprising seven small satellites. Notably, observers can be strategically positioned within select satellites rather than all, thereby mitigating the computational load for clusters with numerous satellites. This approach retains diagnostic efficacy while enhancing computational efficiency.