摘要
针对SDN网络链路结构复杂,仅依靠人工经验诊断故障精确度低的问题,根据SDN结构特点,结合CNN算法和SVM分类器,提出一种基于CNN-SVM的SDN网络链路故障定位算法模型;根据CNNSVM算法流程,研究从超参数对模型的影响、模型故障定位结果两个方面设计完成了算法对SDN链路故障定位的仿真实验,并通过与标准CNN模型比较,对提出的算法进行了分析;实验结果表明,CNNSVM算法模型提高了传统CNN算法在SDN网络链路中的故障定位的精确度,具有良好的故障分类和定位的能力。
Due to the complexity of SDN network link structure and the low accuracy of fault diagnosis only relying on human experience,this paper proposes a fault location algorithm model of SDN network link based on CNN-SVM according to the characteristics of SDN structure,combined with CNN algorithm and SVM classifier.According to the algorithm flow of CNN-SVM,the simulation experiment of SDN link fault location is designed from two aspects:the influence of super parameters on the model and the fault location results of the model.The proposed algorithm is analyzed by comparing with the standard CNN model.The experimental results showed that the proposed CNN-SVM improves the accuracy of fault location of traditional CNN algorithm in SDN network link,and has a good ability of fault classification and location.
作者
程玮
CHENG Wei(Xiamen Ocean Vocational College,Xiamen Fujian 361100,China)
出处
《保山学院学报》
2021年第2期85-92,共8页
JOURNAL OF BAOSHAN UNIVERSITY
