Search

In the smoothing pretreatment for the quantitative analysis of hydrocarbon mixed gases by Fourier transform infrared analysis (FTIR), the Savitzky- Golay filter is usually used as one of the smoothing preprocessing methods in the Fourier transform infrared spectrum data smoothing pretreatment. However, the parameters of the Savitzky-Golay filter such as the polynomial order and frame size are not easy to decide. There is no one unified choice basis. Users usually adopt multiple sets in the special data set to try, and then select a set of relatively optimal data as the optimizing parameters of the Savitzky-Golay filter. The optimal selection method of the Savitzky-Golay filter parameters was explored, and the concrete calculation equations were deduced according to the relation among the normalized cut-off frequency, the normalized beginning frequency of the stopband, the normalized first side lobe peak frequency of the stopband, the normalized first side lobe peak amplitude with the polynomial order and frame size of the Savitzky-Golay filter parameters. Then when the polynomial order and frame size are set as 8 and 11 respectively according the above conclusion and the characteristics of the actual spectral data, the Savitzky - Golay filter smoothing effect is optimum. Through the acquisition the concentration of 0.1%, 0.2%, 0.5%, 1%, 2%, 5% for the actual CH4 spectra, the relative maximum and minimum error of the raw spectra converted absorbance were 17.230 5% and 0.243 0% respectively, and the relative maximum and minimum error of the smooth spectra converted absorbance were 0.088 0% and 0.088 0% respectively in the second absorption peak. The relative error of converted absorbance was basically stable through the Savitzky-Golay filter after the spectral data preprocessing and it was relatively low, so, it laid a foundation for the late spectral data accurate qualitative and quantitative analysis.

The generalized two-dimensional correlation spectroscopy and Fourier transform infrared were used to identify hydrocarbon isomers in the mixed gases for absorption spectra resolution enhancement. The Fourier transform infrared spectrum of n-butane and iso-butane and the two-dimensional correlation infrared spectrum of concentration perturbation were used for analysis as an example. The all band and the main absorption peak wavelengths of Fourier transform infrared spectrum for single component gas showed that the spectra are similar, and if they were mixed together, absorption peaks overlap and peak is difficult to identify. The synchronous and asynchronous spectrum of two-dimensional correlation spectrum can clearly identify the iso-butane and normal butane and their respective characteristic absorption peak intensity. Iso-butane has strong absorption characteristics spectrum lines at 2,893, 2,954 and 2,893 cm(-1), and n-butane at 2,895 and 2,965 cm(-1). The analysis result in this paper preliminary verified that the two-dimensional infrared correlation spectroscopy can be used for resolution enhancement in Fourier transform infrared spectrum quantitative analysis.

In the multicomponent mixture hydrocarbon gases Fourier transform infrared (FTIR) quantitative analysis, especially for light alkane gases, it is not easy to establish the quantitative analysis model because their IR spectra absorption peaks are seriously overlapped. Aiming at this problem, the Tikhonov regularization algorithm was used to select the characteristic wavelengths for seven kinds of light alkane mixture gases FTIR which are composed with methane, ethane, propane, iso-butane, n-butane, iso-pentane and n-pentane. And then the wavelength selection was used to establish the quantitative analysis model. By comparing the analysis characteristics wavelength selection and TR parameters optimization of the mixed gases in the infrared all wave band, the first absorption peak band and the second peak band, the characteristic wavelength of 7 kinds of gases were selected by Tikhonov algorithm. The wavelength selection and Tikhonov regularization parameters were used to test the actual measured methane spectral data, and then we got that with other gas components the max cross sensitivity was 11.153 7%, the minimum cross sensitivity was 1.239 7%, and the root mean square prediction error was 0.004 8. The Tikhonov regularization algorithm effectively enhanced the accuracy in the light alkane mixed gas quantitative analysis. The feasibility of alkane gases mixture Fourier transform infrared spectrum wavelength selection was preliminarily verified by using the Tikhonov regularization algorithm.

For the defects of requiring carrier gas and regular calibration, and low safety using chromatography to on line monitor transformer dissolved gases, it was attempted to establish a dissolved gas analysis system based on Fourier transform infrared spectroscopy. Taking into account the small amount of characteristic gases, many components, detection limit and safety requirements and the difficulty of degasser to put an end to the presence of interference gas, the quantitative analysis model was established based on sparse partial least squares, piecewise section correction and feature variable extraction algorithm using improvement TR regularization. With the characteristic gas of CH4, C2H6, C2H6, and CO2, the results show that using FTIR meets DGA requirements with the spectrum wave number resolution of 1 cm(-1) and optical path of 10 cm.

Aimed at the problem that baseline drift or distortion often appears in Fourier transform spectra after spectrometer has continuously worked for a long time, baseline variation caused by IR source temperature drift and fluctuation, tilt of moving mirror, performance parameter change of beam splitter and lateral shift of detector were studied by simulating using MATLAB. Simulation results show that spectral baseline drift is approximately linear. On this basis, a novel method named spectral baseline correction by piecewise dividing (SBCPD) is proposed to correct spectral baseline in the present work By comparing peak height of simulated spectra, it was found that performance of SBCPD is better than that of common methods of polynomial fitting, air-PLS (adaptive iteratively reweighted Penalized Least Squares). And the application in gas well-logging showed that prediction accuracy of SBCPD is also higher. Additionally, this method is reliable and of less calculation, and is suitable for engineering application.

In the present paper, an improved approach to the TR characteristic spectral selection is presented. For this approach, two ideas of TR1-norm and TR2-norm are used, two constraint items, spectral line distance and minimizing absolute value of coefficient are introduced, and a general formula of ill-posed optimization problem is established. The formula can reduce effectively the errors caused by experienced and experimental method when used in determining the regular matrix and parameter. Finally, the improved approach presented in the paper was used in the analysis of alkane gas mixture, with methane, ethane, propane, n-butane, iso-butane, n-pentane and iso-pentane included. The concentration range is 0.01%-20%. The experimental results show that the predicting error square is only 2.6 x 10(-4), and the coefficient of determination is 0. 959 2, which means that preceding accuracy is high, and that the practicability of TR regularization has been enhanced.

According to the baseline departure of multi-component alkane gas mixture spectra, direct orthogonal signal correction (DOSC) algorithm was proposed to pretreat the infrared spectra data. Fourier transform infrared (FTIR) spectrometer was used to sample 936 spectra data of seven components gas mixture, including methane, ethane, propane, iso-butane, n-butane, isopentane and n-pentane gases. The concentration of each component ranges from 0.01% to 0.1%, 0.01% to 0.1%, 0.01% to 0.15%, 0.0% to 0.1%, 0.0% to 0.1%, 0.0% to 0.05%, and 0.0% to 0.05%, respectively. For analyzing intuitively, partial least square regression (PLSR) was introduced to build the component gas quantitative analysis model. In experiment, DOSC method was compared with first derivative algorithm (FDA) and second derivative algorithm (SDA). In order to get the optimal model, ergodic optimization method was used to select the optimal parameters of the model, i. e. the step of the derivative algorithm, the number of the primary component of the PLSR and the number of orthogonal components of the DOSC algorithm. The experiment results show that DOSC algorithm has the better effect in the field of infrared spectra data pretreating. The average mean relative error (MRE) of the component gas analysis models is 16.58%, which declined by 66.80% from the average MRE before data pretreating 49.93%. Compared with DA, the average MRE declined by 51.51% from 34.19% after pretreated by FDA, and declined by 56.30% from 37.94% after pretreated by SDA. So DOSC method is feasible to pretreat the IR spectra data, and has definite practical and investigation value.

Aimed at the problem that Fourier transform infrared (FTIR) spectral analysis can't be used solely to analyze complex gas mxiture, in which there are many gas compositions, some compositions have same molecular group, and the concentrations of some compositions range greatly, in the present paper, a new spectral analysis method is proposed. For this method, the effects of gas temperature and gas pressure on the analysis result are taken into account, and feature variable extraction, recognition and correction of spectral disorder, robust modeling and multi-layer modeling with neural network (NN) are used to build gas mixture analysis models. At the end of this paper, on-line analysis of five alkane gases is taken as example to test the performance of the analysis method. CH4, C2 H6, Ca3 Ha8, iso-C4 H10 and n-C4 H10 are taken as object gases while iso-C5 H12 and n-C5 H12 are looked as disturbing gases. The comparison results that analysis result curves of FTIR overlap that of gas chromatograph indicate that FTIR can be solely used in on-line and in-situ analysis of multicomponent gas mixture with same molecular group if the analysis method presented in this paper is used.

Feature variable selection and modeling are two of the most principal research contents in spectral analysis. In the present paper, beginning from the introduction of feature spectrum selection based on Tikhonov regularization and discussion on it's application in multi-component mixed alkane gas analysis, 7 sets of feature spectra were abstracted from the absorption spectra of 7 kinds of alkane gas, including methane, ethane, propane, iso-butane, n-butane, iso-pentane and n-pentane. In order to overcome the problem of over-training of neural network, a method called optimal parameter selection of neural netework (NN) was presented to build analysis model of analyte. Optimal parameters were selected from many trained networks with same architecture based on error process. And analysis models of spectral analysis for 7 kinds of alkane gas were built. Finally, the testing analysis results done with standard gases are given. The results show that the method presented in this paper can be used to reduce the cross--sensitivity between any two kinds of gas. The cross-sensitivity is less than 0.5%. The resolving power is as high as 20 X 10(-6).

A method for online quantitative analysis system of alkane gaseous mixture with near infrared spectroscopy is described in the present paper. A single plane diffraction grating is used as the principal device in the monochromatic spectrum system. The key parameters of the monochromator were deduced and calculated in detail. A quantitative analysis system was designed and constructed according to the parameters. The narrow-band beam testing experiments and spectral scanning experiments of seven kinds of single-component alkane gases were accomplished on the above hardware system platform. The narrow-beam experiments show that a 10 nm narrow-band beam spectra was successfully obtained by the monochromatic system when the entrance slit width is 2 mm. And a step-scanning resolution of the outgoing beam's center wavelength with 0.1 nm can be realized within the spectra of 1.2-1.8 microm. The spectral scanning experiments indicate that there was some stronger characteristi absorption spectrum existing between the spectra of 1.6-1.8 microm, which is consistent with the HITRAN spectral database. And there is serious cross-aliasing phenomenon existing among the absorption spectra. These experiments demonstrate that this method has a successful application in mixed gas monitoring and on-line analysis with the characteristics of simple structure and low cost. And it also provides further technical reserves and opens a path way to spectral analysis in the follow-up studies.

According to the difficulty in selecting parameter of least square support vector machine (LS-SVM) when modeling on the gas mixture, and the high computational complexity of the infrared spectrum data, LS-SVM optimized by particle swarm optimization (PSO) algorithm was proposed to build an infrared spectrum quantitative analysis model with feature extracted by principal component analysis (PCA). Firstly, seven feature variables were extracted by PCA as the input of the model from 550 infrared spectrum data of the main absorption apex field, so the computational complexity was reduced. This model aimed at three components of gas mixture, in which methane, ethane and propane gases are included. The concentration of each component ranged from 0.1% to 1%, 0.1% to 1% and 0.1% to 1.5% respectively. Each component quantitative analysis model was built by LS-SVM and the parameters were optimized by PSO algorithm, then the regression model would be reconstructed according to the optimal parameters. This method replaced the traditional ergodic optimization. The experiment results show that the time of offline modeling by PSO was reduced to one fortieth of that of ergodic optimizing. The precision of the model was corresponsive. It can meet the requirement of the measure. PSO algorithm has more superior performance on global optimization and convergence speed. So it is feasible to combine PSO algorithm with LS-SVM to create the infrared spectrum quantitative analysis model. It has definite practice significance and application value.

Due to its many advantages, such as miniaturization, high accuracy, high resolution, fast scanning speed, increased robustness and good stability, acousto-optic tunable filter (AOTF)-near infrared (NIR) spectrometer has been successfully applied in many fields. However, up to now, the commercial AOTF-NIR spectrometers can only be used for liquid and solid detection, but not for the detection of gaseous samples. In the present paper, the feasibility of quantitative analysis of gaseous mixtures by using AOTF-NIR spectrometer was investigated. A homemade gas cell was assembled to an AOTF-NIR spectrometer with probe for liquid detection to obtain NIR spectra of detected gas samples. The gas samples were composed of two groups: single-component CH4 and ternary component gaseous mixture of CH4, C2 H6, and C3 H8. The detection ability of fitted AOTF-NIR spectrometer was tested firstly. Comparing the absorption spectra of various concentrations, the absorbance of CH4 in absorption bands obviously increased with concentration increasing when the concentration was over 0.1%. According to the detection results, the lower limit of detection (LLD) of the AOTF-NIR spectrometer with gas cell was estimated to be 898 microL x L(-1). Subsequently, the NIR spectra of ternary mixtures were collected. The kernel partial least squares (KPLS) regression was employed to create the quantitative analysis model of three components gases. To evaluate the analysis ability of KPLS model, the PLS model was also created. The prediction results of the identical testing set show that the root mean square error of prediction (RMSEP) of three components predicted by KPLS model was 1.08%, 0.87%, and 0.79%, respectively, less than the RMSEP by PLS model. The exploratory work indicates that accurate quantitative analysis of ternary component alkane gaseous mixtures can be achieved by fitted AOTF-NIR spectrometer despite of some limitations, and KPLS regression is an excellent approach to NIR spectra analysis.

An infrared spectrum quantitative analysis model was built based on particle swarm optimization algorithm (PSO) and backward propagation (BP) neural network This model aimed at three components of gas mixture, with methane, ethane and propane gases included. The concentration of each component ranged from 0.01% to 0.1%. Five features variables were abstracted from 1866 infrared spectrum data by principal component analysis as the input of the BP network The gas concentrations acted as the output. PSO was used to optimize the number of neural network hidden layer nodes. Then, the network was trained to construct models for quantitative analysis of these three kinds of gas. The experiment results show that the time taken for optimizing the prediction model by PSO, about 4600 second, reduced to one fifth of that of ergodic optimizing, which is about 24,500 second. The precision of the model is corresponsive and the structure of the network is approximately the same. So the PSO has definite practical significance and application potential.

In order to solve the difficulties that the spectrum training data samples of the massive mixed gas cannot be actually obtained, the analysis precision is low and it is not real time online analysis in the analysis of mixed gas component concentration, the support vector machine, a new information processing method, was used in the mixed gas infrared spectrum analysis, and the concept of mixed gas distribution pattern was proposed in the present paper. Based on the thought that the mixed gas distribution pattern recognition is carried out first, and then the analysis work of mixed gas component concentration is done, sixty kinds of mixed gas distribution pattern were determined after investigation and study, and 6000 mixed gas spectrum data samples were used for model training and testing. Sequential minimal optimization algorithm was applied to realize the decrement and the increase of online learning, and finally the model of infrared spectrum online pattern recognition of mixed gas distribution based on SVM was established. The model structure is composed of 2 levels, pattern recognition level and result output level. The pattern recognition level completes the task of mixed gas distribution pattern recognition; while the result output level is composed of 60 SVM calibration models, and it completes the task of mixed gas concentration analysis. Experimental results show that the correct recognition rate of mixture gas distribution pattern is not lower than 98.8%, and that the method can be used for online recognition of mixed gas distribution pattern under the conditions of small samples of a mixed gas, and can add new mixed gas online, and it has the practical application value.

In the present paper, the authors present a new quantitative analysis method of mid-infrared spectrum. The method combines the kernel principal component analysis (KPCA) technique with support vector regress machine (SVR) to createa quantitative analysis model of multi-component gas mixtures. Firstly, the spectra of multi-component gas mixtures samples were mapped nonlinearly into a high-dimensional feature space through the use of Gaussian kernels. And then, PCA technique was employed to compute efficiently the principal components in the high-dimensional feature spaces. After determining the optimal numbers of principal components, the extracted features (principal components) were used as the inputs of SVR to create the quantitative analysis model of seven-component gas mixtures. The prediction RMSE (phi x 10(-6))of seven-component gases of prediction set samples by use of KPCA-SVR model were respectively 124.37, 72.44, 136.51, 87.29, 153.01, 57.12, and 81.72, ten times less than that by use of SVR model. The elapsed time of modeling and prediction by using KPCA-SVR were respectively 46.59 (s) and 4.94 (s), which was consumedly less than 752.52 (s) and 26.21 (s) by using only SVR These results show that KPCA has an excellent ability of nonlinear feature extraction. It can make the most of the information of entire spectra range and effectively reduce noise and the dimension of the spectra. The KPCA combined with SVR can improve the model's analysis precision and cut the elapsed time of modeling and analysis. From our research and experiments, we conclude that KPCA-SVR is an effective new method for infrared spectroscopic quantitative analysis.

The hydrocarbon mixed gas was characterized by multi-component and varied density. In order to deal with the difficulties that can not be actually solved with mass mixture gas spectrum data samples, 15 kinds of subset patterns were determined on the basis of investigations and studies, which needed 5 500 spectrum data samples for training and testing. On the basis of this, a method of hydrocarbon mixed gas infrared spectrum analysis based on 2-levels and 15 SVM-subsets was proposed in the light of the idea of working pattern recognition --> mixture gas analysis --> the final result output. In order to solve the problem of new subset working pattern, the SVM online categorization algorithm based on spectrum data relational rule was used. The experimental results show that the component concentration maximal deviation is 0.41% and the maximal average deviation is 0.04%. The method can be used in other mixture gas infrared spectrum analyses, and has the theoretic and application value.

A new method of infrared spectrum analysis based on support vector machine (SVM) for mixture gas was proposed. The kernel function in SVM was used to map the seriously overlapping absorption spectrum into high-dimensional space, and after transformation, the high-dimensional data could be processed in the original space, so the regression calibration model was established, then the regression calibration model with was applied to analyze the concentration of component gas. Meanwhile it was proved that the regression calibration model with SVM also could be used for component recognition of mixture gas. The method was applied to the analysis of different data samples. Some factors such as scan interval, range of the wavelength, kernel function and penalty coefficient C that affect the model were discussed. Experimental results show that the component concentration maximal Mean AE is 0.132%, and the component recognition accuracy is higher than 94%. The problems of overlapping absorption spectrum, using the same method for qualitative and quantitative analysis, and limit number of training sample, were solved. The method could be used in other mixture gas infrared spectrum analyses, promising theoretic and application values.

According to support vector machine based on the regularization theory, a small scale machine study theory was proposed to solve the problem of multi-gas analysis, which is mainly restricted by the lack of experimental samples. With its well nonlinear mapping ability, the training error was decided to be zero and global optimal parameters were obtained, hence the cross-sensitivity of spectrum is preferably eliminated. In multi-component gas analysis, the results show that the cross-sensitivity decreased to 1/81. A method based on genetic algorithm and cross-validation was proposed to solve the parameters selection of support vector machine(SVM), which still lacks theory support. The optimal structure parameters were achieved by genetic random search algorithm, the mean square error(MSE) 0. 018 of the spectrometer was achieved in 20th generation, and MSE decreased by multi-times in the fore generations. This hints that the genetic algorithm SVM is more efficient and has better generalizing ability.

This paper analyzes the characteristics of the pyroelectric detector based on its working principle. The input andoutput mathematical model of DIGA (Dispersive Infrared Gas Analyzer) system with pyroelectric detector was established according to the design principle of DIGA. We have manufactured a novel multi-gas DIGA on the basis of this model, then pointed out several problems that should be taken into account in the design. Application indicates that this model is of considerable practical value for the design, study, performance analysis and further improvement of DIGA.