Research Of Network Traffic Classification In Power Communication Network

Power communication network is an important part of the grid.It is the basis to achieve grid scheduling automation and management modernization.It provides grid operation,management,infrastructure and other aspects of the service for the grid,to meet the dispatching telephone,administrative telephone,network automation,relay protection,automatic safety devices,computer networks,video transmission and other business needs,especially the IP communication services carried by the integrated data network.However,with the diversification of today’s integrated data network traffic,resulting in various types of network problems,and it has brought great distress and inconvenience to network management and operations,in order to provide higher quality service,use the bandwidth more efficient,traffic classification is necessary.In order to solve the above problem:First,this paper carried out research on the concept of power communication network,business,communication technology and the type of the integrated data network traffic by finding relevant references.Second,analyzes and compares the network traffic classification methods,and studies the problems in the various types of network traffic classification algorithm.According to the business situation and the characteristics of power communication network,we choose network traffic classification method based on support vector machine algorithm to classify.Then because of the particular requirements of the power industry for traffic classification accuracy,this paper designed a characteristics dimension reduction algorithm based on Relief and PLS to reduce the interference of the redundant features and the correlation between characteristics,so it can improve the accuracy of the classification algorithm.Also we collect some sample data from the integrated data network of a power enterprise,then sample the date and extract the feature from them.Respectively,the SVM algorithm and SVM improved algorithm is applied to the data samples for a comparative analysis of simulation and experimental results show that the proposed SVM algorithm can enhance the accuracy of the original algorithm and its usefulness to the electric power communication network traffic classification.At last,this paper designs and implements the electric power communication network traffic classification system,it simplifies flow classification operation and has an important significance to electric power communication network management.