Research Of The Early Warning Mechanism Of Network Operation Situation Based On Fuzzy Analysis

Nowadays,the network has entered the homes of millions of ordinary people,which has added color and convenience to people's lives.On the other hand,with the increasing size of network models,various forms of networking such as sensor networks,ADHoc networks,and space-based networks With the addition of new networks,the scale of the topology becomes larger and larger,and it is difficult to accurately obtain;the number of network devices is large,the networking methods are various,and the number of times to receive information becomes more and more.The emergence of this situation has increased the difficulty for network managers to maintain the normal operating state of the network.Under this circumstance,network operation situational awareness technology should be born from time to time.This article is based on network operation situation awareness technology,from the perspective of network business,and adopts a method based on time series and expected business losses to construct a comprehensive network operation situation early warning mechanism.According to network operation events and reflected state changes,time series are used The model performs big data analysis and forecasting.This article focuses on how to build a reasonable and effective early warning mechanism,how to accurately and perfectly complete the conversion of indicators between levels,and how to build a real-time and accurate early warning mathematical model.The research content and innovations of this article are as follows:1.Established an early warning model of network operation situation based on video services.Based on the idea of hierarchical network management,the hierarchical network management is a bottom-up data processing process.The bottom layer is responsible for data collection and processing,while the top layer is demand management and analysis.The model combines the fuzzy comprehensive evaluation method based on entropy weight-analytic hierarchy process,improved RBF neural network and other technical means to establish a real-time and accurate network operation situation early warning model structure.2.Proposed a scheme of using entropy weight method to modify the fuzzy comprehensive network operation situation assessment of the analytic hierarchy process.In the original analytic hierarchy process,the weight is relatively easy to be constrained by the knowledge of professionals,lacks feedback on the actual collected data,and has a great subjective nature.The entropy method is used to modify the analytic hierarchy process,and the method of integrating the subjective plan and the objective plan is used to make the weight calculation method more reasonable.It is a superior model by integrating fuzzy theory into the model,transforming quantitative attributes into qualitative analysis,and improving the accuracy of the evaluation model.3.Proposed adaptive k-value clustering and adaptive learning rate RBF neural network.In previous neural networks,in clustering methods,the k value,which is the number of hidden layer nodes,and how to choose the learning rate during iterative learning is a relatively big problem.From the perspective of k-value clustering and network learning rate,this paper uses the contour coefficient to confirm the k value.The optimal learning rate of the RBF network is derived from the formula,and is replaced and calculated in each iteration.The improved RBF neural network avoids the inconvenience of artificially selecting the learning rate,and at the same time allows the network to perform iterative learning normally,and can also increase the iterative speed of the network.