Fault Prediction And Health Management Of Diesel Turbocharger

With the development of trains in China,turbochargers,as one of the key components of train engines,play an important role in the dynamics,economy and reliability of locomotives.Therefore,the prediction of turbocharger malfunctions is very significant.This paper focuses on the method of locomotive turbocharger fault prediction,establishes a fault prediction model,and designs and develops a locomotive turbocharger fault prediction system.Firstly,according to the literature data and the fault data of the supercharger,summarize the common faults of the turbocharger,analyze the main causes of each type of fault and give solutions to provide technical support for the subsequent forecasting system.Secondly,the GA-SGNN locomotive turbocharger fault prediction method is proposed.Firstly,find the state parameters that characterize the working performance of the supercharger,calculate the degree of correlation between the multivariables,and then establish the MGM(1,n)model to predict the variables and test the accuracy of the model with residuals and relative errors.Calculate the relative error less than 5%,indicating that the prediction accuracy of MGM(1,6)meets the requirements.Based on the MGM(1,n)model,the BP neural network is combined with the neural network to establish SGNN model.The results show that the prediction accuracy of the SGNN model is higher than that of the MGM(1,6)model.Based on the established SGNN model,the genetic algorithm is introduced to improve the local optimality and convergence of the grey neural network model.The accuracy of the GA-SGNN model shows that the relative error is within 4%,which can guarantee the reliability and accuracy of the model.Then,a turbocharger fault diagnosis method based on RBF neural network is proposed.According to the characteristics of locomotive turbocharger operating parameters and the advantages of RBF neural network in fault diagnosis,the RBF neural network model is established,and then the fault data is studied by the model.Finally,the predicted value of GA-SGNN model is faulty.The prediction shows that the RBF neural network can accurately diagnose the fault of the supercharger,and the fault is consistent with the actual situation.Finally,a locomotive turbocharger fault prediction system was developed.The common causes and solutions,GA-SGNN and RBF neural network were applied to the prediction system.The prediction system was completed by MATLAB GUI.It implements data import,data and function data to predict fault diagnosis,and established a technical library provides a common cause turbocharger and solutions.