基于近紅外光譜的發(fā)酵過(guò)程pH值軟測(cè)量技術(shù)研究

1.89萬(wàn)字
我自己原創(chuàng)的畢業(yè)論文，僅在本站獨(dú)家提交，大家放心使用


摘要 固態(tài)發(fā)酵是一個(gè)多相多變量、強(qiáng)耦合的非線(xiàn)性系統(tǒng)。在固態(tài)發(fā)酵生產(chǎn)過(guò)程中，一些關(guān)鍵參數(shù)如pH值，只能通過(guò)離線(xiàn)檢測(cè)來(lái)獲得，往往造成信息滯后，這嚴(yán)重制約了固態(tài)發(fā)酵系統(tǒng)控制性能的提高。近紅外光譜分析技術(shù)具有快速、無(wú)損、準(zhǔn)確，多組分同時(shí)檢測(cè)等優(yōu)點(diǎn)，將其與軟測(cè)量方結(jié)合是解決上述問(wèn)題的有效途徑之一。
本文首先介紹了近紅外光譜技術(shù)的特點(diǎn)和應(yīng)用，及其研究現(xiàn)狀和發(fā)展前景，并以小麥秸稈蛋白發(fā)酵過(guò)程為主要研究對(duì)象，對(duì)獲取的固態(tài)發(fā)酵物樣本的原始近紅外光譜，采用離散小波變換結(jié)合主成分分析對(duì)其進(jìn)行濾噪和特征提??；然后利用提取的特征變量建立基于支持向量回歸的參數(shù)模型進(jìn)行回歸預(yù)測(cè)，并采用網(wǎng)格搜索算法對(duì)模型進(jìn)行參數(shù)尋優(yōu)。本實(shí)驗(yàn)的數(shù)據(jù)處理工作是在Matlab平臺(tái)下完成，140個(gè)樣本分成訓(xùn)練集105個(gè)和測(cè)試集35個(gè)，利用K-CV進(jìn)行交叉檢驗(yàn)，SVR為ε-SVR類(lèi)型，核函數(shù)選取RBF。研究結(jié)果表明，利用近紅外光譜技術(shù)結(jié)合支持向量回歸來(lái)進(jìn)行固態(tài)發(fā)酵的pH值軟測(cè)量是可行的，并且具有較理想的結(jié)果。
關(guān)鍵詞：固態(tài)發(fā)酵，近紅外光譜技術(shù)，支持向量回歸，網(wǎng)格搜索

Research on Soft-sensing Technique of pH Value of Fermentation Processes Based on Near-infrared Spectroscopy
Abstract Solid-state fermentation is a multi-phase and multi-variable nonlinear system with high coupling. During the production process of solid-state fermentation, some key variables like pH value can only be obtained through off-line testing, resulting in information lag, this has seriously hampered the control performance of solid-state fermentation system from being improved. Near Infrared Spectroscopy has the advantages of celerity, accuracy, non-destruction, multi-component detection, etc. Therefore, to combine the near-infrared spectroscopy with the soft sensor is one of the effective ways that can solve the problem mentioned above.
Firstly, the structure, present situation of near-infrared spectroscopy and its application prospect are introduced, and wheat straw feed protein fermentation process has been selected as the main research object, the raw spectra of all fermented samples obtained were denoised by use of the discrete wavelet transform (DWT), and the feature vectors were extracted by use of principal component analysis (PCA) from the spectral data preprocessed. Then, the parametric model was developed by use of Support Vector Regression（SVR）,and using Grid search algorithm for model parameter optimization. The experimental data processing is completed in Matlab, 140 samples are divided into training set of 105 and test set of 35, using K-CV, ε-SVR type and RBF kernel function. The overall results sufficiently demonstrate that near-infrared spectroscopy technology coupled with that SVR could be successfully used in soft-sensing of pH value during solid-state fermentation, and have an ideal result.
Key words：Solid-state fermentation，Near-infrared spectroscopy technology，Support Vector Regression，Grid search

目 錄
第一章 緒論 1
1.1 研究背景與意義 1
1.2 近紅外光譜技術(shù)的發(fā)展和應(yīng)用 2
1.2.1 近紅外光譜概述 2
1.2.2 近紅外光譜技術(shù)的發(fā)展與應(yīng)用 2
1.2.3 近紅外光譜在發(fā)酵領(lǐng)域中的應(yīng)用 3
1.3 支持向量機(jī) 4
1.3.1 支持向量機(jī)概述 4
1.3.2 支持向量機(jī)基本思想 5
1.3.3 核函數(shù) 6
1.4 本文研究主要內(nèi)容 7
1.5 本章小結(jié) 7
第二章 固態(tài)發(fā)酵試驗(yàn)與數(shù)據(jù)采集及預(yù)處理 9
2.1 試驗(yàn)材料與方法 9
2.1.1 試驗(yàn)主要設(shè)備 9
2.1.2 樣本制備 9
2.1.3 pH值測(cè)定 10
2.2 光譜信息的采集 11
2.3 光譜預(yù)處理 12
2.3.1 平滑處理 12
2.3.2 基線(xiàn)校正 13
2.3.3 歸一化處理 16
2.4 本章小結(jié) 16
第三章 基于SVR的近紅外光譜的pH值軟測(cè)量方法研究 17
3.1 小波變換 17
3.1.1 小波變換的基本理論 17
3.1.2 離散小波變換 17
3.1.3 離散小波去噪 19
3.2 主成分分析 21
3.3 支持向量回歸 23
3.3.1 支持向量回歸理論 23
3.3.2 支持向量回歸實(shí)驗(yàn) 24
3.3.3 性能評(píng)價(jià) 26
3.4 參數(shù)優(yōu)化 26
3.4.1 參數(shù)簡(jiǎn)介 26
3.4.2 網(wǎng)格法尋優(yōu) 27
3.5 本章小結(jié) 29
結(jié) 論 31
致 謝 32
參考文獻(xiàn)： 33
附錄 37