摘要
煮糖结晶过程是一个非常复杂的物理、化学过程,其内部机理复杂,各变量间相互耦合,很难建立机理模型。利用神经网络具有逼近任意复杂非线性函数的能力,将神经网络应用到煮糖结晶过程建模中。模型以煮糖结晶过程中糖液的过饱和度为输出,以影响过饱和度的温度,压力和各阀门的开度等等作为网络的输入,采用改进的BP算法,建立了煮糖结晶过程的动态响应模型。给出了模型的结构?网络学习流程图和仿真结果。仿真结果表明,此BP网络模型能较好地解决煮糖结晶过程的建模问题。
Cane Sugar＇s boiling and crystallization is a complicated physical and chemical process, and its internal mechanism is complex, and each variable is coupled with another one, so it is difficult to establish its mechanism model. This paper tries to establish a sugar crystal process model by using neural network identification technique. The oversaturation is taken as the output and the temperature, pressure and the valve opening and so on are inputs of neural network, then a dynamic model is made by applying improved BP algorithm. The paper gives out the structure of model, flow chart of network learning and the result of simulation. The result of simulation shows that this BP network model can solve the problem of sugar boiling processor modeling.
出处
《计算机仿真》
CSCD
北大核心
2009年第5期100-102,168,共4页
Computer Simulation
关键词
煮糖结晶
神经网络
建模
Sugar crystallization
Neural networks
Modeling
