Research On Construction Of Network Operation And Maintenance Quality Evaluation Model And Its Key Technologies

In order to improve quality of service and enhance their competitiveness,network operation and maintenance quality evaluation is always one of the key issues for operators.The traditional network operation and maintenance evaluation method relies too much on domain expert experience in key steps such as indicator selection and weight assignment,which makes it difficult to obtain accurate and comprehensive evaluation results.In addition,the quality of network operation and maintenance involves data from multiple manufacturers and devices,and the number of users of communication devices is also increasing.It brings a large amount of heterogeneous data to the maintenance personnel.These situations does present challenges to traditional evaluation methods.This paper focuses on the above problems.Firstly it summarizes the generalization process of quality evaluation model construction,which includes four steps: the selection of evaluation indicators,the process of data quantification and standardization,the assignment of indicator weight,and the establishment of evaluation models.Aiming at the steps of selecting the evaluation indicator,the strategy based on correlation coefficient is proposed.The original indicator set which is high-dimensional is classified by comparing the correlation between the indicator data series,and the concept of centrality in graph theory is used to select the key indicators in each cluster.For the steps of assigning the weight,the traditional standard deviation method is improved by adding the overlap coefficient and the empirical coefficient.And then the weight assignment algorithm based on subjective and objective combination is formed.With the prior knowledge of the field,the consistency and practical value of the standard deviation method is strengthened.In addition,combined with the data processing method and evaluation comprehensive formula which are both without human participation,the network operation and maintenance quality evaluation model are completed.The visual interface for the evaluation results is also displayed.In the experiment,the real data is used to verify the effect of the indicator selection strategy and the weight assignment algorithm proposed in this paper.The indicator set which uses the strategy based on correlation coefficienton is 31% less than the set using manual method on the information overlap rate.The weight assignment algorithm based on subjective and objective combination improves the weight concentration caused by the fluctuation law of the indicator data itself,and makes the indicator weight closer to the one given by the experts.