Prediction of Gas Turbine Combustion Chamber Failure Evolution Trend Based on FA-LSTM
Gas turbine high-temperature components are prone to failures,which are often concealed and can cause significant damage.Moreover,the repair costs and difficulty are substantial after a failure occurs.Researching a method to predict the evolution trend of failures is of great importance for maintenance personnel to conduct timely repairs and make informed maintenance decisions.This paper introduces a failure evolution trend prediction method based on FA-LSTM,which combines the degradation mechanism of combustion chambers with Factor Analysis(FA)to construct a Health Index(HI)for assessing the health status of gas turbine combustion chambers.Utilizing the Long Short-Term Memory(LSTM)neural network's unique ability to process time-series data,the method predicts the evolution trend of combustion chamber failures.Focusing on a gas turbine as the research subject,this method is compared with six other traditional machine learning methods.The proposed method achieves the lowest Mean Absolute Error(MAE)and Root Mean Square Error(RMSE),enabling accurate degradation trend predictions and providing a reliable basis for short-term maintenance.
gas turbinefault evolution trendfactor analysislong short-term memory networks
NETL
NSTL
万方数据
