Research Of The Methods That Evaluates And Analyzes The Effectiveness Usage Of Government Network

In the progress of informatization construction of institutions, such as the education sectors and universities, the goverment pay a large amount of funds for advanced network infrastructures to support informatization construction. However, these equipments always can be used for a long time. It is difficult to evaluate which is reasonable or not for the investment using traditional input-output method. This paper proposed another method. Based on the method of effective usage of network evaluated and analysised the network in order to provide reasonable decision for the managers.On the basis of researching a lot of internal and external documents, it refers to determine evaluation indexes, collecte data using SNMP protocol, and construct a mathematical model for evaluation and analysis based on the BP neural network. At last, it can be obtained a calculation results represented the operation status of the network.Which is a good indicator for the managers to make decisions. On the base of this mathematical model, this paper also give the algorithm to implement a management information system by programming. This system not only can compare the comprehensive indicators, but also can provide a management tool for the network administrators.In the progress of building the model, which difficult is determing the training data sets.At present, subjective ecaluation method is usally used to obtain output data. But, it is difficult for some issues to use it. Considering the results have fuzzy concept to a certain extent, such as excellent, good and bad. This paper refers to fuzzy mathematics theory, using this theory to calculate the output data sets for BP neural network. The BP neural network was combined with fuzzy mathematics theory to implement their own advantages. The fuzzy mathematics method, which is based on operation experiences and knowledge of human beings, can make comprehensive dusicions and solve some difficult issues for conventional mathematical theory. Moreover, the BP neural network can process plenty of concurrent information and display more intelligent performance.