基于免疫神经网络的水泥分解炉温度控制  

The Temperature Control of Cement Decomposing Furnace Based on Immune Neural Network

作　　者：谢刚[1] 杨帆[1] 谢克明[1] 

机构地区：[1]太原理工大学信息工程学院,山西太原030024

出　　处：《太原理工大学学报》2007年第4期287-289,共3页Journal of Taiyuan University of Technology

基　　金：国家自然科学基金资助项目(60374029);山西省留学回国人员基金资助项目(2004-18)

摘　　要：通过对水泥分解炉的特性进行分析,确定了出口气体温度的数学模型。用BP神经网络建立水泥分解炉的控制系统,通过控制煤量和风量来使出口气体的温度保持在规定的范围内。针对BP神经网络通常权值更新方法中出现的收敛速度慢、易陷入局部最优值问题,利用改进的免疫算法来优化神经网络的权值。仿真结果表明该温度控制系统优于通常的神经网络PID控制系统,并具有良好的可靠性、自适应性和鲁棒性。This paper ensures the math model of gas temperature of cement decomposing furnace exit by analyzing its character. Establishes the control system with BP neural network, keeps the exit gas temperature in specified range by controlling the coal and wind quantity. To the convergence speed slowly and trapped in local optimal value which appears often in the common BP neural network weight value update, optimizes the weight value with modified immune algorithm. The simulation results show that the temperature control system is better than usual neural network PID with better reliability, adaptation and robustness.

关 键 词：免疫算法 神经网络 温度控制 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]