Cross-Validation-Based Multi-Fault Detection of Gas Turbine

Operating in harsh environments, gas turbines may encounter a variety of faults. Failure to promptly detect these faults can severely impact their safe and stable operation. During actual operations, accurately detecting the specific location of faults in gas turbines, such as sensors, actuators, or gas paths, to ensure prompt maintenance and safe, stable operation, is a crucial aspect of gas turbine maintenance. To address this issue, this study proposes a multi-fault detection method based on cross-validation. In situations with limited data, we have designed a classification method grounded in logical reasoning and constructed a cross-validation system. This system will be applied to layered analysis and discrimination of multiple faults, aiming to achieve effective detection of faults in gas turbine sensors, actuators, and air paths. Finally, simulation experiments have verified the effectiveness and feasibility of this method.