Study On Fault Diagnosis Technology For Valve-train System Of Diesel Engines Based On Hilbert-Huang Transform And Support Vector Machine

As one of the main moving parts of diesel engines,the working status of the valve-train system directly affects the quality of gas exchanging of the diesel engine.Due to frequent movement as well as extreme work environment,the fault of valve-train systems often occurs resulting in low working efficiency of the diesel engine,which even leads to failures of other parts of the diesel engine and endangers the whole machine and working staff.Therefore,it is of great significance to correctly predict and diagnose the fault of valve-train system for working efficiency and security of diesel engines.At present,vibration fault diagnosis method is wildly applied in the fault diagnosis of valve-train systems with many advantages such as simple signal acquisition and convenient monitoring.However,since the diesel engine is always composed by many complicated structures,there are many ways for vibration dissemination.The vibration signal,as a result,shows strongly non-stationary behaviors in engineering practice.Consequently,effective fault feature extracting and correct identification of fault types and locations are two keys for fault diagnosis of the diesel engine based on vibration signal analysis.From the views of fault feature extracting and fault identification,the thesis presents a fault diagnosis method based on Hilbert-Huang transform(HHT)and Support Vector Machine(SVM)for the valve-train system of diesel engines.Firstly,a signal sampling system is established to collect the vibration signal of the cylinder head.The signal pretreatment will be used to remove trend terms as well as outliers in the signal.After,vibration signals collected with respect to time will be transformed to those related to crank angle by using order tracking arithmetic.Secondly,HHT arithmetic is investigated and also improved aiming to retrain the end effect as well as false patterns caused by HHT itself.The improved HHT arithmetic can be used to analyze the vibration signal with respect to crank angle and extract value fault feature information.In the procedure,the vibration signal is decomposed to a series of intrinsic mode functions with different frequencies,which,after Hilbert transform,can show the time-frequency characteristics of the vibration signal in Hilbert spectrum.Thirdly,the SVM pattern recognition method is studied for fault identification.The SVM classifier is established according to different working status of diesel engines,through whichthe working status and fault types can be identified correctly by inputting fault features extracted by HHT.Finally,online fault diagnosis system based on the HHT and SVM for valve-train system of diesel engines is built.With using the online system,the valve clearance fault and the air leakage fault are diagnosed in real time according to practical vibration signal of the cylinder head in the experiment.Accurate results show that the vibration signal analysis method based on HHT and SVM is effective in the field of fault diagnosis of the valve-train system of diesel engines.