Scholarlink

<p>Engine health management systems have progressed greatly due to the fast development of machine learning methods. Object identification and computer vision are only two of the numerous application fields that have given deep learning techniques a lot of attention in the last decade. Convolutional neural networks’ strong feature learning and classification capabilities have led to a recent explosion in their use for diagnostics of rotating equipment. Having said that, gas turbine diagnosis is still an area where it has limited use. In this paper, we first describe a physics-driven efficiency trend-monitoring system. Then, we show how to use convolutional neural networks to find and isolate gas turbine defects.Then, we present an RBF neural network-based framework to detect and diagnose gas turbine engine performance degradation and fault conditions using extracted sensor-based health indicators. In order to record degradation-related performance changes, create a new baseline as required, and produce fault signatures, the trend-monitoring method was used. The physics-based model generated fault signatures that were used to train the fault detection and isolation system for identifying and categorizing gas path problems at the component level. Under conditions of substantial measurement noise to guarantee model robustness, the suggested method's performance was assessed using several failure situations for a three-shaft turbofan engine. A fault classification approach based on convolutional neural network architecture and a fault detection and isolation method supported by deep long short-term memory were evaluated and compared. In order to decrease fuel consumption, improve aircraft safety, and lower maintenance costs, modern condition monitoring-based approaches are used. Gas turbine engine performance degradation is defined in the literature as a function of vibration, oil pressure,&#xa0;motor fan velocity, oil temperatures, EGT, and fuel flow. A novel model for assessing the health of gas turbine engines was created for this research. An ANN method was used to identify the EGT parameter in this model, and multiple regression analysis was used to quantify the impact of the flight factors on the EGT parameter. By the conclusion of the research, a network had been constructed that could forecast the EGT variable with the least amount of inaccuracy. The aircraft engine's state may be instantly monitored using an interface created in MATLAB Simulink. The aircraft’s gas turbine engine's performance deterioration value or graphical representation of any such degradation makes it easy to see. On top of that, rumors have it that it may be a brand-new sign that alerts pilots when the EGT parameter sensor they use while in the air has a problem.</p><p>The proposed framework integrates sensor-based degradation indicators, normalization, sliding window feature extraction, and outlier detection to improve prediction reliability. Experimental evaluation was conducted using the NASA C-MAPSS turbofan engine simulation dataset. The performance of the proposed RBF neural network model was evaluated using RMSE, score, and average score (AS) metrics and compared with conventional approaches such as multilayer perceptron (MLP) and support vector machines (SVM). The experimental results demonstrate that the proposed model achieves improved prediction performance and provides a reliable approach for aircraft engine performance monitoring and remaining useful life estimation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance Monitoring and Fault Diagnosis of Aircraft Engines via RBF Neural Network Models

  • 期刊论文

  • 发表日期：

    2026年7月2日

查看全文
  • Anuj Mangal,
  • Sahunthala Sanmugam,
  • Manjunathan Alagarsamy,
  • J. Senthil Murugan,
  • Sangeetha Kuppusamy,
  • A. Rajaram

摘要

Engine health management systems have progressed greatly due to the fast development of machine learning methods. Object identification and computer vision are only two of the numerous application fields that have given deep learning techniques a lot of attention in the last decade. Convolutional neural networks’ strong feature learning and classification capabilities have led to a recent explosion in their use for diagnostics of rotating equipment. Having said that, gas turbine diagnosis is still an area where it has limited use. In this paper, we first describe a physics-driven efficiency trend-monitoring system. Then, we show how to use convolutional neural networks to find and isolate gas turbine defects.Then, we present an RBF neural network-based framework to detect and diagnose gas turbine engine performance degradation and fault conditions using extracted sensor-based health indicators. In order to record degradation-related performance changes, create a new baseline as required, and produce fault signatures, the trend-monitoring method was used. The physics-based model generated fault signatures that were used to train the fault detection and isolation system for identifying and categorizing gas path problems at the component level. Under conditions of substantial measurement noise to guarantee model robustness, the suggested method's performance was assessed using several failure situations for a three-shaft turbofan engine. A fault classification approach based on convolutional neural network architecture and a fault detection and isolation method supported by deep long short-term memory were evaluated and compared. In order to decrease fuel consumption, improve aircraft safety, and lower maintenance costs, modern condition monitoring-based approaches are used. Gas turbine engine performance degradation is defined in the literature as a function of vibration, oil pressure, motor fan velocity, oil temperatures, EGT, and fuel flow. A novel model for assessing the health of gas turbine engines was created for this research. An ANN method was used to identify the EGT parameter in this model, and multiple regression analysis was used to quantify the impact of the flight factors on the EGT parameter. By the conclusion of the research, a network had been constructed that could forecast the EGT variable with the least amount of inaccuracy. The aircraft engine's state may be instantly monitored using an interface created in MATLAB Simulink. The aircraft’s gas turbine engine's performance deterioration value or graphical representation of any such degradation makes it easy to see. On top of that, rumors have it that it may be a brand-new sign that alerts pilots when the EGT parameter sensor they use while in the air has a problem.

The proposed framework integrates sensor-based degradation indicators, normalization, sliding window feature extraction, and outlier detection to improve prediction reliability. Experimental evaluation was conducted using the NASA C-MAPSS turbofan engine simulation dataset. The performance of the proposed RBF neural network model was evaluated using RMSE, score, and average score (AS) metrics and compared with conventional approaches such as multilayer perceptron (MLP) and support vector machines (SVM). The experimental results demonstrate that the proposed model achieves improved prediction performance and provides a reliable approach for aircraft engine performance monitoring and remaining useful life estimation.