Fault Detection and Isolation Based on Structural Analysis: Application to a Multi-Engine Propulsion Cluster

Area of Science:

• Aerospace Engineering
• Control Systems Engineering
• System Reliability

Background:

• Multi-engine rocket clusters operate under extreme conditions, complicating fault detection and isolation (FDI).
• Accurate detection of small-magnitude faults is critical for mission success and safety.

Purpose of the Study:

• To develop a model-based FDI system for multi-engine clusters using Structural Analysis (SA).
• To create algorithms for minimal residual subset selection and fault isolation.
• To enhance fault diagnosis accuracy through novel sensitivity metrics.

Main Methods:

• Structural Analysis (SA) applied to a three-engine cluster model.
• Generation and selection of over 16,000 residual generator candidates.
• Development of algorithms for minimal cardinality residual subset identification.
• Introduction of Subset Sensitivity Index (SSI) and Residual Sensitivity Index (RSI) for optimal subset selection.
• Monte Carlo simulations for performance evaluation under ten fault scenarios.

Main Results:

• A methodology for generating unique fault signatures for each fault using minimal residual subsets.
• Identification of optimal residual subsets based on the proposed SSI metric.
• A new fault isolation algorithm leveraging RSI for predicting active faults.
• Demonstrated effectiveness of the FDI system in detecting and isolating sensor and actuator faults.

Conclusions:

• The proposed model-based FDI system effectively addresses the complexity of fault diagnosis in multi-engine rocket systems.
• Novel sensitivity indices (SSI, RSI) improve the selection of residuals for robust fault isolation.
• The developed algorithms provide a reliable method for identifying and isolating faults, enhancing operational safety.