基于RBF神經(jīng)網(wǎng)絡(luò)的生物發(fā)酵過程軟測(cè)量研究

2.26萬字
自己原創(chuàng)的畢業(yè)論文，僅在本站獨(dú)家出售，重復(fù)率低，推薦下載使用

摘要 發(fā)酵過程具有高度非線性、時(shí)變性和模型不確定性，關(guān)鍵變量如菌體濃度、基質(zhì)濃度和產(chǎn)物濃度難以在線測(cè)量，離線分析易導(dǎo)致測(cè)量滯后，且常引起發(fā)酵染菌，直接影響發(fā)酵品質(zhì)，這些使發(fā)酵過程控制變得非常復(fù)雜，難以獲得令人滿意的控制性能，成為制約發(fā)酵過程優(yōu)化控制的瓶頸問題。
近年來發(fā)展起來的軟測(cè)量技術(shù)突破了常規(guī)工業(yè)儀表在測(cè)量方面存在的問題， 軟測(cè)量的基本思想是把自動(dòng)控制理論與生產(chǎn)過程知識(shí)有機(jī)的結(jié)合起來，應(yīng)用計(jì)算機(jī)技術(shù)對(duì)難以測(cè)量或者暫時(shí)不能測(cè)量的重要變量，選擇另外一些容易測(cè)量的變量，通過構(gòu)成某種數(shù)學(xué)關(guān)系來推斷或者估計(jì)，以軟件來替代硬件的功能。
本文以青霉素發(fā)酵過程為研究背景，采用RBF神經(jīng)網(wǎng)絡(luò)對(duì)發(fā)酵過程進(jìn)行軟測(cè)量，RBF網(wǎng)絡(luò)可以任意精度逼近任意的非線性函數(shù)，且具有全局逼近能力，構(gòu)建發(fā)酵過程軟測(cè)量模型，能夠?qū)崿F(xiàn)青霉素發(fā)酵過程中菌體濃度、基質(zhì)濃度、產(chǎn)物濃度等參量的軟測(cè)量，為發(fā)酵過程的優(yōu)化控制提供先決條件，同時(shí)也為其它發(fā)酵過程關(guān)鍵參量的測(cè)量提供方法借鑒。

關(guān)鍵詞：發(fā)酵 軟測(cè)量 RBF神經(jīng)網(wǎng)絡(luò) 青霉素


Study on soft measurement of biological fermentation process based on RBF neural network
Abstract The fermentation process is highly nonlinear, time variability and model uncertainty, the key variables such as cell concentration, substrate concentration and product concentration is difficult to measure online. Offline analysis is easy to cause the measurement lag, and often causes fermentation bacteria, directly influence the fermentation quality. The fermentation process control becomes very complicated, it is difficult to obtain satisfactory control performance, become the bottleneck problems of optimal control of fermentation process.
Soft measurement technology developed in recent years to break the conventional industrial instruments in the measurement issues, The basic idea of soft measurement is the automatic control theory and combined with the production process of organic knowledge . Application of computer technology to the important variables are difficult to measure or temporarily can't measure. In addition to choose some easily measured variables, By constructing a mathematical relationship to predict or estimate, using software instead of hardware functions.
In this paper, the penicillin fermentation process as the research background use RBF neural network soft sensor on the fermentation process. The RBF network can approach any nonlinear function to arbitrary precision, and has the ability of global approximation of the fermentation process. Construction of the fermentation process soft measurement model, can achieve soft measurement of cell concentration, substrate concentration, product concentration and other parameters of penicillin fermentation process. It does not only provide the preconditions for the optimal control of fermentation process, but also provide a method of reference to others for the measurement of key parameters in fermentation process .
Key words fermentation soft measurement RBF neural network penicillin


目錄
摘要 I
Abstract II
第一章 緒論 1
1.1發(fā)酵工程 1
1.1.1概述 1
1.1.2發(fā)展趨勢(shì) 3
1.1.2研究現(xiàn)狀 4
1.2存在的問題 4
1.3發(fā)酵過程軟測(cè)量的研究現(xiàn)狀 5
1.3.1基于機(jī)理模型軟測(cè)量 5
1.3.2基于數(shù)據(jù)驅(qū)動(dòng)建模 6
1.3.3混合建模 7
1.4論文主要工作和意義 7
1.5本章小結(jié) 8
第二章 青霉素發(fā)酵過程及動(dòng)力學(xué)模型 9
2.1 青霉素發(fā)酵過程 9
2.2 青霉素發(fā)酵動(dòng)力學(xué)模型 12
2.3本章小結(jié) 14
第三章RBF神經(jīng)網(wǎng)絡(luò) 15
3.1 RBF神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu) 15
3.2 RBF神經(jīng)網(wǎng)絡(luò)特點(diǎn) 17
3.3 RBF神經(jīng)網(wǎng)絡(luò)算法 18
3.3.1利用模糊K均值聚類算法確定 18
3.3.2確定基函數(shù)寬度(方差) 19
3.3.3調(diào)節(jié)隱層單元到輸出單元間的連接權(quán) 19
3.4 本章小結(jié) 20
第四章 基于RBF神經(jīng)網(wǎng)絡(luò)軟測(cè)量的研究 21
4.1軟測(cè)量理論 21
4.1.1機(jī)理分析與輔助變量的選擇 21
4.1.2數(shù)據(jù)采集和處理 22
4.1.3軟測(cè)量模型及建模的常用方法 22
4.1.4在線校正 24
4.1.5軟測(cè)量性能的影響因素 24
4.2基于RBF神經(jīng)網(wǎng)絡(luò)軟測(cè)量 25
4.2.1基于RBF神經(jīng)網(wǎng)絡(luò)軟測(cè)量通用模型 25
4.2.3測(cè)量數(shù)據(jù)誤差處理 27
4.2.2數(shù)據(jù)樣本歸一化處理 27
4.2.3 RBF網(wǎng)絡(luò)訓(xùn)練算法 28
4.3本章小結(jié) 30
第五章 青霉素發(fā)酵過程軟測(cè)量實(shí)現(xiàn)方法 31
5.1青霉素發(fā)酵各參數(shù)之間的關(guān)系 31
5.2影響青霉素發(fā)酵因素 32
5.3實(shí)驗(yàn)與仿真 32
5.4 本章小結(jié) 36
第六章 總結(jié)與展望 37
6.1總結(jié) 37
6.2展望 37
致謝 39
參考文獻(xiàn) 40