LM2500+SAC Type Gas Turbine Health Evaluation And Fault Diagnosis

The LM2500+ gas turbine is an aeroderivative gas turbine developed by the General Electric Company(GE)in the 1960 s and is widely used in natural gas pipelines.The working environment of gas turbines is complex and severe,which is prone to failures and has a major impact on production operations.Therefore,it is of great significance to evaluate the health status of gas turbines and diagnose faults.This paper took the LM2500+SAC(Single Annual Combustor)gas turbine on the natural gas pipeline as the research object,based on the MATLAB platform,established the LM2500+SAC gas turbine health status evaluation model and fault diagnosis model.First,by analyzing the typical failure types of LM2500+SAC gas turbines,the failure criterion and characteristic parameters are determined;the gas turbine health status evaluation model uses the gray correlation analysis method to calculate the gray correlation degree,thereby judging the gas turbine health status level,and applying gas transmission.The field operation data of the pipeline gas turbine verified the state evaluation model,and the results showed that the state grades of this gas turbine were "fault" and "health",which were the same as the actual situation,indicating the accuracy of this model;the gas turbine fault diagnosis model used BP(Back Propagation)Neural network,support vector machine and extreme learning machine three artificial intelligence machine learning algorithms,and tested separately.The test results show that among the three models,the extreme learning machine model has the best recognition effect.Further use genetic algorithm to optimize the extreme learning machine model and improve its failure recognition rate,so the genetic algorithm optimized extreme learning machine model was finally selected as the gas turbine fault diagnosis model,and the gas pipeline on-site gas turbine The fault data verified the fault diagnosis model,and the predicted result was "power turbine fault",which was the same as the actual situation,indicating that the accuracy of this fault diagnosis model could meet the needs of engineering practice.