Research On The Key Technology Of Fiber Bragg Grating Sensor Signal In Wavelength Demodulation

Fiber optic sensor technology is one of the core technologies in modern sensingtechnology. Due to the optical fiber sensor has a lot of advantages, such as terrainwithout restrictions, free from the impact of natural climate, high sensitivity and lowcost, so it is widely used in medical research, aerospace, military and other fields.According to the demodulation of FBG wavelength, the different physical parameters，such as temperature, stress, acceleration, was detected by Fiber Bragg Grating sensor(FBG). And the key technologies of the wavelength detection technology are de-noisingtechniques and peak detection techniques. Therefore, the research of key techniques inthe process of optical fiber Bragg grating sensor signal wavelength demodulation has acertain research significance and application value.In this thesis, the noise of signal acquisition system is analyzed in the FBGtemperature detection system. According to the characteristics of the noise introduced indifferent cases, two effective methods are proposed respectively. An improvedneighborhood wavelet coefficient de-noising method is proposed to remove the noise,which is generated by electronic equipment. According to the distribution properties ofthe signal and noise at different scales, the decomposition level is introduced to proposean adaptive threshold and threshold function. And a self-optimizing method was putforward to determine the size of neighborhood window for the uncertain issue ofneighborhood window. Meanwhile, an improved sparse representation de-noisingmethod was proposed to remove the noise, which is introduced by the equipment agingperformance degradation resulting in partial loss of information. The implementationsof the algorithm must be on the premise of the known signal the sparsity, but it isdifficult to determine in practice. So the signal sparsity is confirmed by introducing themulti-index with entropy weight method fusion and combine with the method ofsaturation point. And finally subspace pursuit algorithm is put forward by introducingthe thought of orthogonalization.Based on the spectral characteristics of the FBG, the asymmetric characteristics ofFBG reflection spectrum were able to affect the peak detecting precision. However, forthe peak detection of spectrum，the current peak detecting algorithms has a preconditionthat the spectrum is a standard Gaussian model. To solve the problem, the advantagesand disadvantages of the existing algorithm are analyzed; at the same time, the factors of the spectral asymmetric characteristics are considered, an Exponent ModifiedGaussian (EMG) Curve Fitting peak detecting algorithm is proposed.In the thesis, the FBG multi-point detection temperature calibration model andmulti-point temperature detection experiment system is established. And combined withthe proposed de-noising and peak detecting method to deal with the FBG sensing signal,the FBG temperature calibration and error analysis are completed; meanwhile it verifiedthe feasibility and validity of proposed methods.