摘要
针对可调引射混合式低压加热器是一个多输入多输出的非线性系统,具有时变和非线性特性,采用广义生长-剪枝RBF(GGAP-RBF)神经网络对其进行双变量神经网络自适应控制。该方法用GGAP-RBF网络对加热器非线性模型进行实时辨识,并将系统的Jacobian信息反馈给BP神经网络控制器,从而保证了控制器对被控对象的精确控制。通过加热器的控制对比试验,结果表明该方法能在动态条件下实现对加热器的自适应控制,并且具有较好的动静态性能。
A method of double - variable neural network adaptive control method based on GGAP - RBF network is proposed. The ad- justable ejecting - mixing low - pressure heater is with some characteristics including multi - input and nmhi - output, and nonlinearity time - varying uncertainty. In this method, the heater nonlinear model is real - time identified by GGAP - RBF network, and system Ja- cobian information is real - time feedback to BP neural network controller so that the controller can exactly control the heater dynamics. Through comparisons, the results show that it can realize the double - variable control for the heater under dynamic conditions, and the pooosed control scheme has good dynamic and static performances.
出处
《控制工程》
CSCD
北大核心
2012年第3期420-424,共5页
Control Engineering of China
基金
重庆市科委重大科技攻关项目(CSTC2009AB108)