Health monitoring systems (HMS) are crucial to monitor the performance deterioration of the critical components like compressors/turbines/fuel systems of the aircraft power plant with time. Such systems have a direct impact on reducing aircraft power plant maintenance costs, reducing fuel consumption, and increasing aircraft safety. Current practices rely on offline HMS to examine the performance deterioration of a gas turbine engine during overhaul. Parameters such as propeller speeds, vibration, oil pressure, oil temperature, Inter Turbine Temperature (ITT), and fuel flow of gas turbine engines are checked during maintenance.

In the current work, we intend to explore the modern diagnostics and prognostics monitoring methods that rely on the Gas Path Analysis (GPA) model to examine the changes in the component performance parameters such as efficiency, mass flow, etc. An advanced non-contact online machine vision (MV) fault diagnostics model will be developed to evaluate the performance penalties due to compressor blade foul and surface degradation in a turboprop engine. The algorithm (developed in MATLAB) uses image processing techniques to diagnose foul. After validating the algorithm on standard sandpapers, it has been extended to predict foul factors over the complex compressor blade surfaces. Subsequently, we intend to correlate the foul factor with the efficiency drop using machine learning-based data-driven models. In contrast to the offline diagnostics, such image processing-based strategies are useful to develop online health monitoring systems which can potentially inform the pilot in the event of fault in sub-system during ground and airborne operations.