摘要
为实现在实际的炉群多变量燃烧系统中,对各个燃烧的子系统的控制参数进行优化,提出了一种基于改进适应度函数的遗传单神经元控制算法,该算法克服了采用神经网络方法收敛速度慢、在求解过程中陷入局部极小点等缺点,利用遗传算法的全局寻优特性和神经网络对非线性函数较强的逼近能力,将改进的遗传算法和单神经元控制相结合,实现对一类非线性系统的参数进行优化。模拟实验和真实结果验证了这种方法是可行的。
Control parameters must be optimized during the application of multiple variable stove system in order that good performance of the sub-systems can be obtained.Neural net algorithm,which is widely used, is analyzed and its drawbacks of slow convergence and being unable guarantee global optimum solution are pointed out. Single neuron control based on genetic algorithm with improved fitness function is presented. Optimum characteristics of genetic algorithm and good performance with approaching nonlinear function of neural net algorithm are used in this method. Experiment results validated this control algorithm.
出处
《控制工程》
CSCD
2005年第S2期112-114,共3页
Control Engineering of China
基金
北京化工大学青年基金资助项目(QN0416)北京市教委资助项目(XK100100435)
关键词
遗传单神经元控制
改进适应度函数
遗传算法
单神经元
single neuron genetic control
improved fitness function
genetic algorithm
single neuron