You are here

Automated Performance Monitoring for Rotorcraft Turboshaft Engines Using a Multimodel Approach

Description:

OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Trusted AI and Autonomy

 

OBJECTIVE: Develop and integrate multiple models using machine learning and artificial intelligence to continuously and accurately estimate and predict power available for rotorcraft turboshaft engines across all aircraft operating conditions.

 

DESCRIPTION: Accurate estimates of engine health are critical for ensuring safe operation of helicopters supporting heavy-lift operations. Various approaches exist to assess engine health and available power in a rotorcraft context, and the rapid evolution of machine learning and artificial intelligence is further expanding the realm of possible solutions. The development and maturation of algorithms that utilize existing aircraft data parameters and that have the potential for real-time, or near real-time performance, are of considerable interest. In particular, significant operational efficiencies can be obtained if engine performance deterioration can be accurately determined and predicted over a wide range of operating conditions. Maintenance can be planned in advance, with necessary personnel and resources pre-positioned to minimize mission and readiness impacts. Specific aircraft operating conditions that lend themselves to accurate estimation of power available may not occur with regularity, thereby limiting the potential effectiveness of any individual approach. Optimal predictive performance can be achieved by combining multiple models and algorithms via decision-fusion, ensemble learning, and so forth. An ideal solution would also provide the means to monitor and evolve the models over time, support the incorporation of new models, provide interpretability and explainability, and be broadly applicable to different engines.

 

PHASE I: Design and demonstrate multiple approaches for engine health and/or power available estimation using Navy datasets and commercially available, open-source computing languages and packages (Python, etc.). Design and demonstrate technical feasibility for combining the models using machine learning and artificial intelligence approaches to improve model performance. The raw data may need to be filtered, manipulated, or normalized to enable implementation of the models. The Phase I effort will include prototype plans to be developed under Phase II.

 

PHASE II: Develop and demonstrate a multimodel approach for accurately estimating and predicting engine health and/or power available over a wide range of operating conditions. Demonstrate and validate the approach within a Navy data environment in an automated context.

 

PHASE III DUAL USE APPLICATIONS: Demonstrate scenarios involving model re-training, updating, and incorporation of new models, within the Navy data environment. Develop tools and processes to monitor model performance and assist with long-term management.

This software capability would be broadly applicable to aerospace, turboshaft engines, and could be commercialized as an engine management tool for commercial operators.

 

REFERENCES:

1.       Peddareddygari, L. M. (2020, August). Time to failure prognosis of a gas turbine engine using predictive analytics [Master’s thesis, Texas A&M University]. https://oaktrust.library.tamu.edu/bitstream/handle/1969.1/192563/PEDDAREDDYGARI-THESIS-2020.pdf?sequence=1&isAllowed=y

1.       Simon, D.L., & Litt, J.S. (2008). Automated power assessment for helicopter turboshaft engines. NASA/TM-2008-215270. https://ntrs.nasa.gov/citations/20080032562

2.       Li, Z., Goebel, K., & Wu, D. (2019). Degradation modeling and remaining useful life prediction of aircraft engines using ensemble learning. Journal of Engineering for Gas Turbines and Power, 141(4). https://c3.ndc.nasa.gov/dashlink/static/media/publication/2018_DegradationModelingRULEnsemble_Wu.pdf

3.       Li, Z, Wu, D., Hu, C., & Terpenny, J. (2019). An ensemble learning-based prognostic approach with degradation-dependent weights for remaining useful life prediction. Reliability Engineering & System Safety 184, 110-122. https://www.sciencedirect.com/science/article/pii/S0951832017308104

4.       Rigamonti, M., Baraldi, P., Zio, E., Roychoudhury, I., Goebel, K., & Poll, S. (2018). Ensemble of optimized echo state networks for remaining useful life prediction. Neurocomputing, 281, 121-138. https://c3.ndc.nasa.gov/dashlink/static/media/publication/2017_12_ESN_Ensemble_NEUCOM.pdf

 

KEYWORDS: Ensemble Learning; Artificial Intelligence; Machine Learning; Prognostic Health Management; Engine Health Monitoring; Turboshaft Engine

US Flag An Official Website of the United States Government