摘要
提出一种结构等价型模糊神经网络的学习算法．等价型神经网络根据模糊系统的推理规则，决定等价的神经网络结构参数，因而网络结构特殊．采用的学习算法是用误差反传对局部节点的权值进行调整，收敛速度快．实验表明，将其用于火灾探测系统中，能够准确、及时地探测各种标准试验火，并具有较强的抗干扰能力．
A learning algorithm for structure equivalent neuro-fuzzy network is presented. Accordingto the reasoning rules of fuzzy system, the equivalent structure parameters of neural network aredetermined for a special structure of network. Because network parameters have physical meanings, it isdifferent from conventional black-box network. The learning algorithm applies the feedback errors toad just weights of nodes and gives fast convergence. It also has been applied in a fire detection system.The results of experiment show that the fire detection system can detect various normal test firesaccurately and with strong anti--interference capability.
出处
《华中理工大学学报》
CSCD
北大核心
1998年第8期77-80,共4页
Journal of Huazhong University of Science and Technology
基金
武汉晨光计划资助
