Study On Engine Fault Diagnosis Based On Exhaust Airflow Dynamics

As the heart of vehicles, engine is fatal to the whole system's performance. Mon-itoring and fault diagnosis of engine on board without going to garage and be disas-sembled, will sure to improve vehicles performance, safety, particularly its reliabilityand fuel economy. Currently, investigation of the in-vehicle fault diagnosis techniqueis of great value and attracts many researchers worldwide.In the thesis, after analyzing of the source and propagation of engine exhaust air-flow, and studying the relationship between the engine running conditions and exhaustairflow dynamics, an engine fault diagnosis method is proposed based on exhaust noiseVQ (vector quantization) analysis, and a new system for engine misfire detection byexhaust wave is designed. The work is reported as follows:(1) By studying the working principle of the engine and exhaust system features,based on theoretical analysis and simulation, it is clear that exhaust airflow varies withengine working conditions, and causes the pressure wave in exhaust pipes, which thenproduces noise directly or indirectly via pipe wall vibration. Therefore, both exhaustpressure waves and exhaust noise can be used as signal sources of engine detection andfault diagnosis.(2) A three-cylinder four stroke gasoline engine model is built with the professionalengine simulation package. Exhaust wave dynamics is simulated and the link of engineexhaust pressure fluctuation and its working conditions is analyzed.(3) VQ algorithm is used for engine fault diagnosis based on exhaust noise forthe first time. Using the MFCC(Mel-frequency cepstrum coeflcients) of exhaust noisesignals as feature parameters, VQ approach used to analysis exhaust noise, identifydiflerences of exhaust noise under diflerent engine conditions, and diagnosis the enginefaults.(4) The system idea of fault diagnosis system of the engine is extracted, by collect-ing exhaust noise using PC sound card and LabVIEW, which breaks through the tradi-tional method of signal acquisition. Based on analyzing exhaust noise characteristics,the best measurement point is chosen, laying a good foundation for the experiments.And the signal acquisition platform of exhaust noise is established by LabVIEW.(5) The idea of engine misfire detection by exhaust wave is extracted, to lay thetechnological foundation for comprehensive fault diagnosis. After completing the ex-haust wave signals acquisition, a new method is proposed for engine misfire detection,using three pairs of peak-to-peak value within a working cycle as feature parameters.The experiments indicate that this method based on exhaust wave is accurate, intu-itional and convenient.