| In recent years,the optical transmission network,as part of the new infrastructure,is an important communication foundation for the development of the digital economy.In order to ensure its service quality,scientific planning and rational design are required during the network construction process.The optical network performance evaluation is to guide the optimization of the network.An important basis for planning,construction and security work.At present,there are some problems in the performance evaluation of optical networks.First,subjective factors account for a large proportion in the performance evaluation process based on the subjective weighting method,which does not take into account the changes in the index data of the actual operating network,which leads to the inability to guarantee the objectivity of the evaluation results.Secondly,with the complexity of the optical network structure and the diversification of business types,the complexity and time cost of evaluating the network performance become higher.The application of big data technology makes it easier to obtain data resources,and it is imminent to introduce machine learning to study a more effective optical network performance evaluation model.In view of the above problems,this paper focuses on the following two aspects in the evaluation of optical transmission network performance.First,aiming at the time-varying characteristics of optical network traffic,an improved AHP based on the entropy weight method is proposed to determine the weight of each index,and then the fuzzy comprehensive evaluation method is used to obtain the quantitative evaluation value of the optical network,and then the bucket aggregation is used to obtain the quantitative evaluation value of the optical network.The method classifies the evaluation value,and then gives an intuitive qualitative evaluation.The effectiveness of the method is verified by the comparison of simulation experiments.Compared with the traditional AHP,the improved AHP based on the entropy weight method can better consider the relative contribution between indicators,adapt to the time-varying conditions of flow,and can dynamically obtain weights when the flow changes,thereby obtaining more objective and accurate results.evaluation results.Secondly,an optical network performance evaluation model based on particle swarm optimization(PSO)optimized BP neural network is proposed.Taking index data as input and network rating as output,the PSO algorithm is used to optimize the BP neural network to obtain an evaluation model based on multi-dimensional performance indicators.The experimental results show that compared with the traditional BP neural network model,the optical network performance evaluation model based on the PSO-BP neural network has higher accuracy and reliability,which provides a new idea for the application of machine learning to optical network performance evaluation. |
