Development Of Performance Testing And Failure Prediction System Of Turbocharger

Turbocharger applications greatly improve the performance of engine, so it has been a research hotspot. In order to test the reliability of the structure of turbocharger, and improve its structural design, it needs to check if all the parameters of turbocharger have reached design indicators and find other factors which affect the performance of turbocharger. Fault prediction of turbocharger can ensure the safe and efficient operation. It also reduces unnecessary economic loss and casualties. So it has a very important significance to develop performance testing and failure prediction system of turbocharger which has reliable performance and convenient operation, also meet the requirements of dynamic measurement precision. In order to meet the actual needs, this paper is dedicated to develop performance testing and failure prediction system of turbocharger.First, the basic structure and working principle of turbocharger are expatiated; the performance parameters and characteristic curve of turbocharger are also studied. On the basis of summarizing the common fault of turbocharger and its reason, the fault tree of turbocharger are established. In view of the structural features and common faults of turbocharger, the vibration analysis method is researched, such as time domain, frequency domain, wavelet analysis and shift orbit.Then, failure prediction method of turbocharger based on the MGM (1, n)-SVM are proposed. First of all, the multivariable grey model MGM (1, n) and support vector machine (SVM) theory are discussed respectively. Multivariable grey model is a multiparameter failure prediction. Through grey correlation analysis, important parametric are selected to model MGM (1, n) which is used for single step and multi-step prediction. SVM decision tree of turbocharger are constructed to distinguish failure by using multi-class classification method of support vector machine (SVM) based on binary tree. In order to qualitatively project faults for mechanical equipment, failure prediction method based on the MGM (1, n)-SVM are proposed. This method combines the advantages of MGM (1, n) and SVM. Its feasibility, effectiveness and accuracy are verified by example of turbocharger.Final, performance testing and failure prediction system of turbocharger are designed and developed. The structure, working principle and species of the turbocharger test bench are summarized and investigated. Testing solution of temperature, pressure, rotational speed, flow and vibration are designed, as well as the structure of database SQL Server2008. Performance testing and failure prediction system of turbocharger are developed by using LabVIEW and MATLAB. And the function modules of the system are introduced in detail, including data acquisition, data storage and management, data processing and analysis. The system can run effectively and steady.