22 / 2025-05-02 14:49:47

Early Warning of Axial Compressor Performance Degradation under Variable Operating Conditions Using a Hybrid Parameter-Based Diagnostic Framework

axial compressor, simulation model, performance calculation model, hybrid parameter-based early warning

Special Sessions > SSL 05 Condition monitoring, fault diagnosis and intelligent maintenance of mechanical equipment

The axial compressor, a pivotal component in gas turbines, is often considered essential in critical industries such as oil and gas, maritime transport, and aerospace. Reliable fault detection and early warning mechanisms for axial compressor failures are essential to ensure the operational safety of gas turbines. Existing fault detection systems, which typically rely on single-parameter monitoring, frequently suffer from false alarms and missed detections, while some multi-parameter methods do not adequately consider the impact of varying operating conditions on alarm thresholds. To address these limitations and improve fault warning accuracy, we propose a hybrid parameter-based gas path fault warning method for axial compressors. This method integrates a gas turbine performance simulation model and a performance calculation model, developed based on thermodynamic principles, to simulate measurable parameters and assess the performance of key components. Non-steady-state data are filtered using interval estimation to enhance data reliability. Deviation features for both measurable and performance parameters are constructed, with corresponding thresholds determined for various operating conditions. To improve robustness, an impulse anomaly checking strategy and a hierarchical warning mechanism are incorporated. The proposed method is validated through a real-world case study involving compressor blade fouling, based on 1,150 field data sets. Performance parameter deviation features exceeded the threshold at the 795th data point, while measurable parameter deviation features surpassed the limit at the 1004th point, demonstrating the method’s ability to provide early, reliable warnings. This approach offers a valuable tool for the monitoring, operation, and maintenance of axial flow compressors, with significant implications for engineering applications.