Research On Gas Turbine Gas Path Fault Diagnosis Method

Gas turbines are widely used in electric power,marine and aerospace industries.The demand for equipment maintenance and reliability of gas turbines is increasing.Gas turbine fault diagnosis plays a crucial role in equipment maintenance decision and system safety improvement.Firstly,the gas turbine gas path single fault and composite fault data are generated by the marine split shaft gas turbine simulation model.The gas turbine gas path single fault diagnosis model based on long short-term memory(LSTM)and the compound fault diagnosis model of gas turbine gas path based on long short-term memory and convolutional neural network(CNN)are designed.Then use the data test model’s recognition ability;use the performance index to evaluate the performance of the model;change the data signal-to-noise ratio and the data noise processing noise type detection model’s anti-noise performance and the generalization performance of the model.The recognition ability and anti-noise performance of the model are compared with other models.The comparison results show that the model has better recognition ability and anti-noise performance.The real data is adopted to further prove the validity of the designed models.Finally,the gas turbine health parameter prediction model based on the long short-term memory is constructed.The model reflects the change of gas turbine health parameter to a certain extent,and provides reference information for gas turbine gas path fault diagnosis.The method proposed in this paper can be effectively applied to gas turbine fault diagnosis.