减法聚类-ANFIS在网络故障诊断的应用研究  

Study on application of subtractive clustering and adaptive network-based fuzzy inference system in network fault diagnosis

作　　者：蒋静芝[1] 孟相如[1] 李欢[1] 庄绪春[1] 

机构地区：[1]空军工程大学电讯工程学院,西安710077

出　　处：《计算机工程与应用》2011年第8期76-78,86,共4页Computer Engineering and Applications

基　　金：陕西省自然科学基金(No.SJ08F14);空军工程大学电讯工程学院研究生创新基金

摘　　要：提出了一种基于减法聚类-自适应模糊神经网络(ANFIS)的网络故障诊断建模方法。减法聚类算法生成初始模糊推理系统,ANFIS建立网络故障诊断原始模型,应用混合算法对模糊规则的参数进行训练并建立最终的模型。仿真实验表明基于减法聚类-ANFIS的建模方法是有效的;通过仿真结果比较,减法聚类-ANFIS的网络故障诊断能力及收敛速度均优于BP神经网络,更适合作为网络故障诊断模型。A method for building network fault diagnosis models is proposed based on subtractive clustering and Adaptive Network- based Fuzzy Inference System（ANFIS）.The subtractive clustering is used to build initial fuzzy inference system,ANFIS is adopted to build network fault diagnosis original model,hybrid algorithm is used to train the parameter of fuzzy rule,and the final model is established.Simulation experiment results show that the modeling algorithm based on subtractive clustering-ANFIS is effective.Compared with the simulation results,the fault diagnosis ability and convergence speed of the subtractive clustering- ANFIS network are all better than the BP neural network,and much more suitable as network fault diagnosis model.

关 键 词：网络故障诊断 减法聚类 自适应模糊神经网络 模糊逻辑 神经网络 

分 类 号：TP306.3[自动化与计算机技术—计算机系统结构] TP393.07[自动化与计算机技术—计算机科学与技术]