丁二酸发酵过程软测量模型的参数优化研究  

Parameter Optimization of Soft Sensor Model of Succinic Acid Fermentation Process

作　　者：袁安平[1] 张湜[1] 姜珉[2] 陈可泉[2] 

机构地区：[1]南京工业大学自动化与电气工程学院,南京210009 [2]南京工业大学制药与生命科学学院,南京210009

出　　处：《化工自动化及仪表》2009年第5期13-17,共5页Control and Instruments in Chemical Industry

基　　金：国家自然科学基金资助(20606017)

摘　　要：针对生物转化法制备丁二酸发酵过程机理复杂、高度非线性、生物参数难以实时在线测量等特点,采用基于结构风险最小化的支持向量机对发酵过程建模。与神经网络方法相比,所建模型的预测效果更好。同时采用粒子群算法对支持向量机模型中的关键参数进行优化,通过MATLAB软件仿真的结果表明,与粒子群算法结合的支持向量机模型,相比于单一的支持向量机模型,以及人工神经网络模型,它的泛化能力更好,性能更为优越,这对实现丁二酸发酵过程的优化控制具有重要作用。In accordance with the features of complicated mechanism,non-linear and hardship to get real-time and on-line biology parameters in succinic acid fermentation process,a support vector regression(SVR) model based on the structural risk minimization was established for estimating the concentration of succinic acid.The model had the better performance than that established by artificial neural networks(ANN).Then the particle swarm optimization(PSO) algorithm was applied to optimize the key parameters in SVR model simultaneously.The simulation results show that higher prediction accuracy and better performance of the PSO-SVR model than those based on simple SVR model and artificial neural networks.The PSO-SVR model is of important significance for optimal control in the succinic acid fermentation process.

关 键 词：丁二酸发酵 软测量模型 支持向量机 粒子群算法 参数优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]