Research On Situation Awareness Learning Algorithm Based On Fuzzy Neural Network

With the continuous development of computer network and data science,traditional network analysis and network management have exposed their shortcomings.At present,it has become a new demand in this field to establish a complete perception model based on the running situation of network.Therefore,this thesis puts forward a core problem of network running situation awareness.The situation awareness of network running will make an overall judgment of the network situation through the observing,processing,analysis and prediction of the important situation factors,as well as the subsequent adjustment of network resources.In this process,combined with the characteristics of network data itself such as fuzziness and large quantity,fuzzy logic method is adopted to analyze network operation data,and artificial intelligence method is used to model it,which will be an inevitable new direction of network running situation awareness research.Therefore,this thesis focuses on the research core of network running situation awareness,and also improves the establishment of network awareness model and the selection of indicators.In the prediction of network situation,it adopts the fuzzy neural network(FNN)method to establish an analysis model of situation awareness,and focuses on the theoretical research and experimental simulation of learning parameter algorithm of the model.The main contents of the thesis are summarized as follows:1.Network running situation assessment.Firstly,this thesis proposes a complete situation awareness model and selects the important factors of network running by combining the network topology information and network traffic information.Then the objective weighting method of principal component analysis is used to assign weights to each index,and the fuzzy evaluation method is used to evaluate the data of each feature.In the fuzzy evaluation part,this thesis proposes a membership function determination method based on clustering method,and objectively obtains a more accurate membership function.2.Parameter learning algorithm of fuzzy neural network.Situation awareness for network problem,this thesis constructed based on the T-S type fuzzy neural network model of running situation awareness in joined the nonlinear model of the structural design of output the activation layer to increase the power of expression of the model.As to algorithms,it respectively discusses the BP algorithm and particle swarm optimization(PSO)algorithm applied to study the characteristics of model parameters and improvement methods.The thesis as well proposes a combination algorithm PSO-BP of the two thoughts of PSO and BP learning algorithm.In this thesis,the experimental simulation and result analysis of each parameter learning algorithm are also completed,and the optimization rate and iteration times of each algorithm are verified,and the correctness of the model is proved by the test data.