| In the21st century, with the problem of Energy shortage growing seriously, dieselengine becomes more and more people pro-gaze with its high efficiency, low energyconsumption, but the attendant emissions cause great harm to the environment. In order tomeet the increasingly stringent emissions regulations, the selective catalytic reduction (SCR)technology for diesel engine has been used widely. The sensor signal plays a crucial role inthe detection process of the operating status of the SCR system, the authenticity andreliability of the sensor largely determines whether the SCR system can run safely andsteadily. Compared with the other components, the sensors which have complex workingenvironment are more vulnerable to failure. If the sensor faults are not detected and handledtimely, it may cause the SCR system stopping work or even more serious security incidents.Therefore, developing a method to diagnose the sensor fault quickly and accurately is ofgreat significance.In recent years, wavelet analysis theory stands out in many fault diagnosis methodswith its good frequency localization features. Applying wavelet analysis theory on the sensorfault diagnosis of the SCR system has broad prospects. In this context, the following studiesare done:Analyze on the advantages and disadvantages of the traditional fault diagnosis method,the faults classification in accordance with the causes of the faults, explanation and modelanalysis of several major faults, such as the bias fault, drifting fault, impact fault andcomplete fault.A comprehensive study of Wavelet transform theory, including the process of wavelettransform, the similarities and differences of wavelet transform and the Fourier transform,the characteristics of wavelet transform and the application of wavelet transform in signal processing. And then the paper focuses on the use of extreme point theory of continuouswavelet transform for fault diagnosis and conducted a simple simulation with MATLABmodel. Finally, based on the fault classification, the aforementioned theory is adopted tosimulate and diagnose the six types of the SCR system sensor faults.In this paper, wavelet analysis application effect on the sensor fault diagnosis is tested,through a large number of simulation examples. The results show that this method candiagnose various types of sensor faults rapidly and accurately. The simulation also provedthe wavelet analysis for fault diagnosis has the characteristics of high efficiency, highsensitivity and a strong ability to overcome the noise. This provides a guideline and basis ofcomparison for follow-up additional sensor faults diagnosis method. |
