摘要本文研究了软测量技术的原理及建模步骤、神经网络的构成及学习算法。其中,重点讨论了数据预处理中的主元分析法、粗糙集的原理及算法,神经网络结构中的 BP网络以及 RBF 网络。在青霉素发酵过程中,产品成分难以在线测量,且系统具有非线性和不确定性,通过基于神经网络的软测量方法对其建立了 BP、PCA-BP、RS-BP、RBF、PCA-RBF、RS-RBF吹冰种模型。利用 Matlab神经网络工具箱编程,对模型进行仿真研究,比较各类模型性能差异,选定RS-BP 网络作为青霉素发酵过程的模型。32629 毕业论文关键词 软测量技术 BP网络 RBF网络 主元分析法 粗糙集 青霉素发酵
Title Study on Soft Sensor Modeling Method Based on Neural Networks
Abstract The principle and modeling process of soft senor model, and the composition and learning algorithm of neural networks are presented in this paper. Among them, the focus is the data preprocessing including the principle and algorithm of principal component analysis (PCA) and rough sets (RS), the structure of BP network and RBF network. In the process of penicillin fermentation, the product is hard to be measured online, and the system is nonlinear and uncertain, to solve these problems, six kinds of soft sensor models based on neural networks including BP、PCA-BP、RS-BP、RBF、PCA-RBF and RS-RBF are established. Use neural network toolbox of Matlab to simulate these models, compare differences between performances of them. Finally, RS - BP is selected as the model of penicillin fermentation process.
Keywords Soft sensor BP RBF PCA RS Penicillin fermentation process
目次