Vibration Research Of F-T Dieselon Diesel Engine Based On Time-Frequence Domain And Wavelet Analysis

The rapid development of automobile industry has led to huge fuel demands. As a new alternative and excellent diesel fuel which can be directly used in the diesel engine, F-T diesel provides a new resource to ease the pressure of country’s energy shortages.Diesel engine has the advantage of big power、high efficiency and good economic, but it prone to failure because of complex operating conditions. Line fault diagnosis can be real-time monitoring of the diesel engine working conditions and detect faults in time,improve engine reliability and economy. Surface vibration signals of the engine contain abundant information related to engine condition. Extract the eigenvalues of surface vibration signal as a basis for fault diagnosis, can improve the accuracy and reliability of fault diagnosis.In this dissertation, without changing the engine structure and operations parameters, a fullcomparison of combustion performance is made between F-T and standard diesel s to identify any potential influences on key engine technical parameters including the starting point of combustion、ignition delay、cylinder pressure、pressure rise rate、heat release rate and pressure oscillations.Moreover, it has also investigated the dynamics of the diesel engine、main vibration excitation and their characteristics as well as transmission path for more detailed diagnosis of engines running F-T diesels. As the signal is non-stationary, wavelet methods is employed to reduce the noise signal in the surface vibration signal and to extract the eigenvalues of the surface vibration signal in the time、frequency domain and time-frequency domain. Finally it has established the time、frequency eigenvalues prediction model and the sensitive parameters to the time、frequency eigenvalues were predictive analysis.The results show that F-T diesel has softer combustion and smaller vibrationresponses, compared with that of the0#diesel.Brige-Massart threshold allow better noise canceling performance.The time-frequency analysis method is superior to the traditional Fourier analysis, revealing the time、intensity and frequency characteristics of the various stimulus-response signal.In addition, BP neural network has good generalization capability in that it is able to predict various responses more accurately for implementing a model based online detection and diagnosis toevaluate the influences of different fuels.