Advanced Fault Diagnosis and Performance Monitoring Techniques for Gas Turbines in Gas and Refinery Industries

Gas turbines are vital in power generation and chemical processing within gas and refinery industries, ensuring efficient and reliable operations. This review explores recent advancements in both traditional and AI-based fault diagnosis and performance monitoring methods. Studies emphasize the use of artificial neural networks (ANNs), convolutional neural networks (CNNs), and hybrid models combining ANNs with support vector machines (SVMs) and other machine learning techniques. Trained on extensive experimental and simulated datasets, these models enable accurate fault prediction and diagnosis. Key developments include high-precision thermodynamic models that account for variable geometry compressors and environmental conditions, crucial for fault detection. The integration of physics-based and data-driven approaches, particularly deep learning, has demonstrated significant potential for real-time diagnostics and monitoring. Challenges such as transient performance in micro gas turbines (MGTs) and the impact of alternative fuels like hydrogen are also addressed. Additionally, the review highlights the importance of data preprocessing and AI model interpretability. Future research should focus on validating models in real-world conditions, advancing hybrid methods, and developing novel control strategies for MGTs operating on alternative fuels.

Keywords:

Gas Turbines , Fault Diagnosis , Performance Monitoring , Artificial Neural Networks (ANNs) , Convolutional Neural Networks (CNNs) , Support Vector Machines (SVMs) , Hybrid Models , Thermodynamic Models , Data Preprocessing , Alternative Fuels