摘要
为了提高神经网络的训练速度和泛化能力 ,同时解决一般模糊神经网络由于输入增多而导致模糊规则膨胀的问题 ,提出了多输入模糊神经网络的结构和算法。此算法用取大取小运算部分代替网络的积和运算 ,同时提出一种获取重要规则的方法。最后将多输入模糊神经网络应用于建筑投标报价系统。仿真结果表明 ,本网络具有较快的训练速度和较高的泛化能力。
In order to improve the training speed and the generalization ability of the fuzzy neural network and to solve the problem that the fuzzy rules will expand in the traditional fuzzy neural networks when the number of inputs increases, this paper proposes a structure and an algorithm of the multi-input fuzzy neural network. In the algorithm maximum and minimum partly replace product and sum. The paper proposes a means to obtain the fuzzy rules, too. At last this paper applies the multi-input fuzzy neural network to the construction bidding system. The results of the simulation indicate that the multi-fuzzy neural network has a high training speed and a much better generalization ability.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2003年第10期1249-1252,1291,共5页
Systems Engineering and Electronics
基金
大连理工大学知识科学研究中心资助课题
