Research On Cyberspace Running Situation Awareness Based On Fuzzy Neural Network

With the continuous development of network communication technology,the network has brought great convenience to people's lives.While bringing convenience to people,the quality of cyberspace situation also affects all aspects of our lives.Ensuring the quality of cyberspace situation,giving a reasonable judgment on the current network situation,and giving a reasonable prediction of the network situation in the future is the focus of current network management research.Cyberspace situation awareness technology based on cyberspace situation data is the key to solving this problem.The cyberspace situation data has the characteristics of multiple sources,massive,complexity,uncertainty and ambiguity,so the cyberspace situation awareness is essentially the process of analyzing complex data.Combining fuzzy theory and neural network technology can make the fuzzy logic inference system greatly improve its fuzzy inference ability with the powerful parallel computing power of neural network.At the same time,the introduction of fuzzy theory to the neural network can enable the neural network to adaptively process fuzzy information.Therefore,this thesis is based on the fuzzy neural network for cyberspace situation awareness,and its main research contents are as follows:1.Determined the evaluation system of cyberspace situation quality indicators.Based on the principles of comprehensiveness,measurability,independence and authenticity of the index evaluation,this thesis integrates the two indicators of network performance and network flow,and selects the protocol flow indicators of the network layer and application layer respectively,from the network performance and service quality Two aspects reflect the operating quality of the network.In addition,in the process of determining the indicators,this article fully considers the measurability of the indicators.2.Proposed CSA-Fuzzy Neural Network.Combining the goals and requirements of network operation situation awareness,and considering that most of the network operation data is fuzzy,this thesis combines fuzzy logic inference system with neural network,and proposes CSA-fuzzy neural network as the structure of cyberspace situation awareness model.The CSA-fuzzy neural network is compared with the traditional fuzzy neural network,and the experiment proves that the fuzzy neural network model proposed in this paper has a faster training speed.3.Improved the fuzzy neural network learning algorithm.Genetic algorithm is susceptible to the influence of crossover probability and mutation probability,which leads to poor local search ability.Simulated annealing algorithm can control the appropriate cooling rate to ensure the local search ability.In this paper,the two algorithms are improved separately,and the improved algorithm is combined to propose the AUGA-MTSA hybrid algorithm.There are three aspects to the improvement of the algorithm: First,the use of a mixed temperature drop function allows the algorithm to have a faster convergence rate in the early stage of the search,and the temperature drops slowly in the later stage of the search,and the algorithm has a good local search ability.Second,in the process of generating the initial population,the distance D is added as a constraint condition to generate initial individuals uniformly distributed in the solution space,which is conducive to the global search of the algorithm.Third,two encoding methods are adopted to enable the algorithm to adjust the network structure adaptively.And the number of nodes is introduced into the fitness function,and the size of individual fitness is combined by the error loss and the number of nodes.The convergence efficiency of the improved algorithm is verified by simulation,and the verification results show that the convergence rate and convergence effect of the algorithm are better.4.Collect real cyberspace situation data,and preprocess and obfuscate the data.The proposed CSA-fuzzy neural network and the improved AUGA-MTSA training algorithm are used for training,determine the optimal parameters,finally determined the cyberspace situation awareness model.Then compared and verified the model's situation prediction accuracy rate.Experiments show that the model has a better convergence effect during the training process,and the situation prediction accuracy rate during the test process is higher.Finally,building the network situation awareness system,and use the model to complete the real-time evaluation of the cyberspace situation,and use the visual evaluation results guide the business application modules in the system.