Research And Application About Spectroscopy And Fiber Optic Vibration Signal Preprocessing Algorithms

Signal preprocessing is very important in the field of engineering. It can be used widely in the field of military communication, chemical, food safety field and so on.So it has great theory value and operation significance to research on signal preprocessing. In this paper, researches based on different one-dimensional signal.That would show that the importance of signal perprocessing in different fields.Firstly, this thesis has a deep study on the time domain feature of pure R6 G Raman spectroscopy. Wavelet analysis, the least mean square error(LMS) adaptive filter and the Ensemble empirical mode decomposition(EEMD) theory and algorithms are given. When those methods were applied to pure R6 G Raman spectroscopy which signal-to-noise ratio(SNR) is lower than 10 dB, contrast and analysis the result with result of wavelet transform and EEMD. The three evaluation standards(SNR, root mean square error(RMSE) and the correlation coefficient( ?) fully proved denoising advantages of LMS adaptive filter in Raman spectroscopy.Secondly, his thesis has a deep study on the time domain feature of fiber optic vibration signal. Fast Fourier transform(FFT), wavelet packet frequency band energy spectrum, fast computation of the Kurtogram theory and algorithms are given.Analysis of the knock signal respectively using FFT, wavelet packet frequency band energy spectrum and fast computation of the Kurtogram. According to the results found that FFT algorithm can not reflect the characteristics of the knock signal in the frequency domain. Fast computation of the Kurtogram can analyze transient signal’s frequency spectrum and reflect the knock signal frequency domain feature. Using the wavelet packet frequency band energy spectrum can extract the characteristic of knock signal for pattern recognition.Finally, found in the denoising of Raman spectroscopy, after EEMD can screen out with high frequency noise of Raman spectroscopy. It was found that LMS algorithm is efficient, easy to implement. But This method has the contradiction which between the convergence rate and the steady-state error. Based on the above content, EEMD combined with VS-LMS algorithm is suggested in this paper. Experimental results show that the proposed algorithm is able to denoise more efficiently than the current algorithms.