摘要
提出了一种采用模糊小波基函数神经网络的控制器,该控制器采用小波基函数作为模糊隶属函数,利用神经网络实现模糊推理,并可对隶属函数进行实时调整,从而使控制器具备更强的学习和自适应能力。还提出了控制器参数的混合学习算法,即先采用混沌算法离线优化,再采用BP梯度算法在线调整。对锅炉主蒸汽温度控制的仿真结果表明了此法的可行性和有效性。
A controller, which makes use of fuzzy wavelet basis function neural network, is being proposed. Wavelet basis functions are used as fuzzy membership functions together with neural networks for realizing fuzzy reasoning as well as for real time adjustment of the membership functions, to promote the controller＇ s learning and self-adaption capability. A hybrid learning algorithm for the controller＇ s parameters is moreover being proposed, i.e. in a first step, chaos optimizing algorithm is used for off-line optimization, followed by on-line adjustment with BP gradation algorithm. Simulation results of a boiler＇ s fresh steam temperature control shows the feasibility and effectiveness of the proposed method. Figs 3 and refs 6.
出处
《动力工程》
EI
CSCD
北大核心
2006年第2期233-237,共5页
Power Engineering
基金
上海市重点学科建设项目(编号P1303)
关键词
自动控制技术
模糊神经网络
小波基函数
混合学习算法
主汽温控制
仿真
automatic control technique
fuzzy neural network
wavelet basis function
hybrid training algorithm
fresh steam temperature control
simulation
