Dissertation > Excellent graduate degree dissertation topics show

Research on Content Measurement of Textile Mixture by Fourier Transform Near Infrared Spectroscopy

Author: YanLi
Tutor: LiuLi
School: Wuhan Textile University
Course: Physical Electronics
Keywords: Near Infrared Spectroscopy Fiber content Wavelet Transform Principal component analysis (PCA) Multiple linear regression model (MLR) Principal component regression model (PCR) BP neural network model (PCA-BP, WT-ca3-BP, WT-ca4-BP, WT-ca5-BP)
CLC: TS107
Type: Master's thesis
Year: 2010
Downloads: 91
Quote: 0
Read: Download Dissertation

Abstract


Near Infrared Spectroscopy faster its speed of analysis, the effect is good, low-cost, pollution, etc., are widely used in many fields. The work using near infrared spectroscopy combined with chemometric methods, blended fabrics component testing conducted in-depth research in cotton, polyester and cotton wool sample preparation, the near-infrared spectral data collection and collation of data, pre-processing and a variety of model the established model upgrade a lot of work, and ultimately achieve the intended purpose, and achieved satisfactory results. (1) Details of the near infrared spectroscopy of the basic theory, study the near-infrared spectral detection method, and the process of analysis, spectral data pretreatment technology and quantitative analysis methods; (2) design the experiment of of blended fabrics Fourier Near Infrared Spectroscopy programs created from the sample preparation, the spectral data acquisition, data preprocessing and correction model study of the entire experimental process, initially identified experiments particular embodiment the various segments; (3) was prepared according to the experimental scheme of the design, the CVC and Mianmao kinds of fiber samples, each sample was divided into a calibration set (41), validation set (10) and the prediction (5) three categories, and collecting the near-infrared spectral data, the establishment is behind the data preprocessing, model ready; (4) data preprocessing techniques based on near-infrared spectroscopy, the collected spectral data according to its own characteristics, data collation, analysis spectral regions, spectral data using wavelet transform method to analyze the spectral region denoising and compression processing, the principal component analysis method to reduce data dimension, and optimize the data obtained pretreatment greatly reduce the amount of data and improve data quality; (5) establishment of linear and nonlinear models. Followed by the establishment of the model are: multiple linear regression model (MLR), principal component regression model (PCR), principal component analysis and BP neural network combined model (PCA-BP), the combination of wavelet transform and neural network model ( WT-ca3-BP, WT-ca4-BP, WT-ca5-BP), wherein the former two kinds normalized for the linear model, the after four kinds normalized for the nonlinear model. Create a model evaluation system (absolute error AE, absolutely average error MAE, standard deviation, RMSE of) provide theoretical support for the analysis and evaluation of the calibration model. (6) compare the model analysis reveals that: simple linear model principle, help to understand the basic principles of near-infrared analysis; nonlinear model prediction accuracy is significantly better than the linear model and high flexibility, especially combination of wavelet transform and neural network model WT-ca3-BP (41-17-2) predicted the highest accuracy, Miandi MAE of less than 1.8%, RMSE less than 2.4%, the interlock of less than 2% MAE, RMSE in less than 2.5%, this model is the most suitable to be used to predict the unknown sample. 41 calibration set and 10 validation set samples in the case does not change the structure of WT-ca3-BP model, incorporated as a calibration set of the new model, the model upgrade, upgrade the model prediction accuracy is higher, the effect is more ideal. Finally, upgraded model to predict the five unknown samples; (7) analysis of the error is the source of generation, establish the component of blended fabrics Near Infrared detection system.

Related Dissertations

  1. Feature Extraction, Selection and Combination in Lipreading,TP391.41
  2. Research on Identification System of Cashmere and Wool Fiber,TS101.921
  3. Monitoring Leaf Nitrogen Status in Rice with Near Infrared Spectroscopy,S511
  4. Characteristics of sensory stimulation evoked,R318.0
  5. Network transmission ROI image coding algorithm,TN919.81
  6. Image Fusion Algorithms Based on Multi-scale Analysis,TP391.41
  7. Feature Extraction Technologies Research and Implementation of 3D Models Based on Wavelet Transform,TP391.41
  8. An Algorithm on Clustering and Anomaly Detection for Multiple Data Streams,TP311.13
  9. Research on Information Hiding Based on Images,TP309.7
  10. Design and Implication of Image Compression System Based on Wavelet and DM6446,TP391.41
  11. SAW gas sensor array pattern recognition technology research,TP212
  12. The Study of Vibration Signal Analysis System Based on Wavelet Fractal Theory,TN911.6
  13. Combined DWT Dynamic Data Reconciliation Research and Application,TP274
  14. Traffic Prediction Model Based on Wavelet and Markov,TP393.06
  15. The Study of Image Watermark Base on Discrete Wavelet Transformation,TP309.7
  16. Research on Automatic Picking of First Arrival Based on Wavelet Transform,TP311.52
  17. The Compression and Application of the Brain CT Image,TP391.41
  18. Inspection and Recognition on a Class of Tool Processing Trace,TP391.41
  19. The Research and Design of a Fire Detection System Based on Combustion Audio Recognition,X924.4
  20. The Realization and Study of a Scalable Data Encoder Based on SPIHT,TN762
  21. Research and Implementation of Dwt-based Video Codec System at Low Bit Rates,TN919.81

CLC: > Industrial Technology > Light industry,handicrafts > Textile industry,dyeing and finishing industry > General issues > Standards and testing of textiles
© 2012 www.DissertationTopic.Net  Mobile