Dissertation > Excellent graduate degree dissertation topics show

Research on the Application of Bayes Neural Network and FTIR in Quantitative Analysis of Multi-gases

Author: WangZhiWen
Tutor: ZhangJiLong
School: University of North
Course: Signal and Information Processing
Keywords: Fourier transform infrared spectroscopy Quantitative analysis of multi-component gas BP neural network Principal component analysis Bayesian regularization
CLC: X831
Type: Master's thesis
Year: 2011
Downloads: 60
Quote: 0
Read: Download Dissertation

Abstract


Air pollution caused tremendous harm to human health and the environment will also destroy the ecological balance of nature . Therefore, to the sustainable development of the earth , it is necessary to take the means of monitoring of pollution sources for analysis and prediction , in order to expand the appropriate control and governance . Conventional gas detection technology in terms of scope, sensitivity , long service life and reliability have limitations and shortcomings , and Fourier transform infrared spectroscopy ( FTIR ) with its wide range of applications , high sensitivity, high accuracy and long service life characteristics become more preferably gas detection method . Fourier transform infrared spectroscopy technology application and development of the exposition and analysis of the research status of infrared quantitative analysis method , on this basis, to build the group assignment gas and spectral acquisition system , selected three common toxic gases CO , NO , NO2 carry out the analysis . The preparation of the mixed gas of different proportions in accordance with the dynamic flow of the gas distribution method , by controlling the precision flowmeter , and a large number of spectral data sample collected by the spectrometer . Choose to use artificial neural network modeling and analysis for the simultaneous determination of multi-component low concentration gas complex nonlinear relationship . On the advantages and disadvantages of various improved algorithm of neural network , according to the infrared spectra of large volumes of data and quantitative accuracy requirements , using principal component analysis and Bayesian regularization method to improve the neural network . First collected spectral region for a reasonable choice , and then principal component analysis of high-dimensional absorbance data dimensionality reduction , Bayesian regularization method error objective function of the neural network has been improved through the MATLAB neural network the development environment for modeling and optimization of the parameters of the model . The model was validated prediction sample set , results show that the the three gases quantitative fitting accuracy 0.974, forecast root mean square error of less than 20 ppm . Contrast to the conventional undescended dimensional neural network widely used LM neural network , the results show that the main ingredient - Bayesian neural network model in the infrared quantitative analysis applications can achieve better modeling speed and prediction accuracy .

Related Dissertations

  1. Research on Feature Extraction and Classification of Tongue Shape and Tooth-Marked Tongue in TCM Tongue Diagnosis,TP391.41
  2. Research on Visual Servo System of Mechanical ARM,TP242.6
  3. Analysis of Infrared Spectral Features and Identification of Al-responsive Genes in Small Black Soybean with Aluminum Stress,S529
  4. Municipal tourism land use planning environmental impact assessment,X820.3
  5. Application of Improved Principal Component Analysis Algorithm in Course Construction,G642.4
  6. Study on Taste Characteristic of Taste Peptide Enzymatic Production from Oyster Base on A Neural Network Method,TS254.4
  7. The Research on Evaluation of Living Status Systems of Expressway Relocated People,D523
  8. Research of Diagnosing Cucumber Diseases Based on Hyperspectral Imaging,S436.421
  9. The Impact of Tourism on Typical Vegetation in Luya Mountain Nature Reserve, Shanxi Province,S759.9
  10. Macaca mulatta palm morphological study of pattern ridge count,Q954
  11. Zhaoguan Lower Coal Group water inrush prediction and control techniques,TD745
  12. Mine Risk Information Integration and Intelligent Early Warning,X936
  13. Research of Orange Quality Classification Technology Based on Computer Vision,TP391.41
  14. Research on Cultural Industrial Competitiveness of Chong Qing,F224
  15. Optimization Study on Gating System and Molding Process Parameters of Injection Mold Based on Simulation,TQ320.662
  16. Study on Luohe Technical Supervision Bureau of Food Safety Early Warrning System Based on Neural Network,F203
  17. Research of Adaptive Active Noise Control Based on Neural Network,TP183
  18. The Research of Prairie Road Light Environment Effects on Physiological Indicators of Drivers,U491.254
  19. Research on Automatic Reading System for Digital Meters,TP391.41
  20. Research on Feature Extraction, Selection and Classification Algorithms for Pulmonary CAD,TP391.41
  21. The Research of Evaluation Method in Connect6 Based on BP-TD Learning,TP18

CLC: > Environmental science, safety science > Environmental Quality Assessment and Environmental Monitoring > Environmental monitoring > Atmospheric monitoring
© 2012 www.DissertationTopic.Net  Mobile