Vibration Feature Extraction Of The Automobile Rear Bridge And Virtual Instrument Design

The Rear Bridge is one of the key parts of the automobile, according as statistic, 20% accidents result from Rear Bridge. So it is necessary to inspect it strictly before Rear Bridge leave factory.In this thesis vibration diagnosis method is adopted to realize the quality inspection of the Rear Bridge. The vibration feature of the rear bridge was analyzed and measure position was ascertained according to the actual operating mode and the surveying result. Then using the Virtual Instrument which I designed completed the vibration signal sampling and processing. The emphases of this thesis are signal denoising and feature parameters extraction of the Rear Bridge in running condition.Signal denoising is the foundation of the following signal processing, which can affect the validity of the analyzed signal results. Wavelet and wavelet packet were utilized to denoise, and denoising results show the denoising effects of the wavelet packet are better than wavelet's.Whether the feature parameter which can right reflect the running condition of the Rear Bridge can be extracted is most important in product inspection, and can affect the reliability and validity of the inspection results. Three methods are adopted to extract the feature parameter: Time Averaged Wavelet Spectrum method, Wavelet Packet Energy method and frequency-domain indexes method.The methods of Time Averaged Wavelet Spectrum and Wavelet Packet Energy can respectively determine its own Euclidean distance as the only quality inspection threshold of the Rear Bridge. Spectrum Entropy Arm or CG frequency can be used as feature parameter to judge the running quality of the Rear Bridge.All these three methods can effectively ascertain the threshold which can reflect the running quality of the Rear Bridge through actual measuring, butthe calculation of the Time Averaged Wavelet Spectrum method and Wavelet Packet Energy method are very complicated ,and their calculating amount are very great. However, the calculation of the frequency-domain index is simpler and faster, so it especially suits to automation diagnosis and be transplanted to the holding type portable inspection instruments.