Posted in | News | Aerospace Robotics

AI Boosts Fault Detection in Turbofan Engines

A recent study published in the journal Aerospace introduced a novel diagnostic technique for detecting gas-path faults in turbofan engines. This approach combines a probabilistic neural network (PNN) with an engine performance model (EPM) to enable efficient and reliable fault detection and diagnosis, even under varying operating conditions. The study aimed to improve diagnostic accuracy and enhance aircraft safety.

AI Boosts Fault Detection in Turbofan Engines
Study: Model-Assisted Probabilistic Neural Networks for Effective Turbofan Fault Diagnosis. Image Credit: Omelchenko/Shutterstock.com

Advancement in Gas Turbine Engine Fault Diagnosis

The aerospace industry relies on advanced diagnostic methods to ensure the safety and efficiency of gas turbine engines. Gas-path faults in turbofan engines can significantly reduce performance, increase fuel consumption, reduce efficiency, and pose safety risks.

Traditional methods often require extensive data collection and analysis, which can be time-consuming and may not fully address the dynamic conditions of modern aircraft operations.

Integrating artificial intelligence (AI) techniques, such as neural networks and machine learning, into diagnostic processes offers a more efficient and accurate approach to fault detection, which helps overcome the limitations of conventional methods.

Novel Diagnostic Approach: Model-Assisted PNN

In this paper, the authors introduced a two-step diagnostic approach that combines a PNN with an EPM. The method dynamically generates training data to reflect the engine’s operating conditions during measurement. This approach is especially useful for real-time fault detection without needing large pre-collected datasets, which are often difficult to obtain due to variability in aerospace engine operations.

The study includes key components such as preprocessing measurement data, health assessment using the PNN, training data generation via the EPM, and post-processing to identify fault characteristics when detected.

Using an engine fault knowledge base, the PNN classifies the engine’s condition into predefined states, covering both healthy and faulty conditions. The researchers designed PNN as a probabilistic classifier that evaluates engine health based on aerodynamics-based measurements.

Training data is generated in real-time, with EPM simulating various operational and fault scenarios. This method addresses data scarcity in traditional machine learning techniques, ensuring the PNN is trained with data relevant to the engine’s current state.

Key Findings and Insights

The proposed diagnostic method showed high accuracy in detecting and identifying common gas-path faults in compressors and turbines of a mixed-flow turbofan engine. These faults included fouling, foreign-object damage (FOD), erosion, and tip clearance issues, with data covering various operating conditions within a typical flight envelope.

The results were presented using confusion matrices, showing the number of correctly and incorrectly classified cases for each fault type. For instance, 85.9% of tip clearance faults at the High-Pressure Turbine (HPT) were correctly identified, while 14.1% were misclassified as Low-Pressure Turbine (LPT) faults due to limited inter-turbine measurements.

Overall, diagnostic accuracy exceeded 90% across fault types, reaching over 95% for healthy engine states and 98% for faults in the fan and high-pressure compressor. These outcomes highlight the method’s effectiveness in aerodynamics and gas flow analysis.

The novel technique performed consistently across different operating points, showcasing its adaptability. By generating training data that matches the engine’s real-time conditions, it overcame the limitations of static datasets, enhancing the robustness of the PNN and maintaining high diagnostic accuracy despite potential noise in the data.

Dynamic data generation also mitigated data scarcity and reduced computational costs by minimizing the need for extensive datasets. The successful application of this method to a mixed-flow turbofan engine underscores its potential for practical use in aerospace maintenance.

Furthermore, the authors highlighted the method’s efficiency, with real-time operation enabling rapid fault detection to sustain engine integrity during flight.

Practical Applications

This research has significant implications for the aviation industry. The proposed diagnostic technique can be integrated into engine monitoring systems to enhance predictive maintenance strategies, reduce unscheduled maintenance, and improve aircraft reliability.

Timely interventions can contribute to safer flight operations and cost savings for airlines through optimized maintenance schedules. Integrating this method into aerospace systems can improve operational efficiency and safety and provide accurate, real-time assessments of engine performance.

Conclusion and Future Directions

In summary, the PNN and EPM-based novel diagnostic framework proved effective for gas-path fault detection in turbofan engines. The dynamic generation of training patterns tailored to specific operational conditions significantly improved diagnostic accuracy and efficiency.

As the aviation industry continues to evolve, such advancements in diagnostic technology are essential for ensuring the safety and reliability of aircraft operations.

Future work should expand the scope of the PNN to handle multiple simultaneous faults and refine the EPM to include more complex engine dynamics. Additionally, integrating this diagnostic method with other machine learning techniques could enhance fault prognosis capabilities, potentially leading to more advanced autonomous monitoring systems in the aerospace industry.

Journal Reference

Romesis, C.; Aretakis, N.; Mathioudakis, K. Model-Assisted Probabilistic Neural Networks for Effective Turbofan Fault Diagnosis. Aerospace. doi: 10.3390/aerospace11110913, https://www.mdpi.com/2226-4310/11/11/913

Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Muhammad Osama

Written by

Muhammad Osama

Muhammad Osama is a full-time data analytics consultant and freelance technical writer based in Delhi, India. He specializes in transforming complex technical concepts into accessible content. He has a Bachelor of Technology in Mechanical Engineering with specialization in AI & Robotics from Galgotias University, India, and he has extensive experience in technical content writing, data science and analytics, and artificial intelligence.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Osama, Muhammad. (2024, November 15). AI Boosts Fault Detection in Turbofan Engines. AZoRobotics. Retrieved on December 13, 2024 from https://www.azorobotics.com/News.aspx?newsID=15476.

  • MLA

    Osama, Muhammad. "AI Boosts Fault Detection in Turbofan Engines". AZoRobotics. 13 December 2024. <https://www.azorobotics.com/News.aspx?newsID=15476>.

  • Chicago

    Osama, Muhammad. "AI Boosts Fault Detection in Turbofan Engines". AZoRobotics. https://www.azorobotics.com/News.aspx?newsID=15476. (accessed December 13, 2024).

  • Harvard

    Osama, Muhammad. 2024. AI Boosts Fault Detection in Turbofan Engines. AZoRobotics, viewed 13 December 2024, https://www.azorobotics.com/News.aspx?newsID=15476.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type
Submit

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.