Information Decision Software Tools for Turbine Propulsion PHM Systems

The project will develop software tools to enable optimal estimation of engine fault state in embedded avionics system. The tools will address vulnerabilities of existing PHM systems and allow achieving low (the smallest theoretically possible) false alarm rates while simplifying system engineering and sustainment. The fault states will be estimated/detected from the residuals and BIT data using empirical and/or physical fault models. Existing PHM approaches combine several heuristics for characterization of engine health deterioration and design of engine fault alarms. This project will rely on recent breakthroughs in estimation and optimization that allow integrating all data (discrete and parametric) and all fault states (discrete and continuous) in one optimal solution. The tuning parameters of the solution would include sensor noise covariances, BIT/BITE mis-detection probabilities, fault probabilities, fault interdependencies, and model confidence. First, we will develop off-line tools for model-based system analysis and design. Second, we will develop embedded software functions for computationally efficient on-line implementation of the approach. The embedded computational functions could be employed similar to other packaged mathematical software, such as matrix arithmetics. Phase I will demonstrate viability and performance of the approach to design and analysis of PHM for an F-110 engine vane actuator subsystem.