摘要话务量，也称电信负载量，在代表着通信设备承载能力的同时，另一方面也体现了用户对通信的需求的要求。除此之外。话务量随着用户数量，通信频繁程度以及每次通信所耗时长的变化而变化，因此对于通信用户话务量的研究来制定适当的话务量方案，成为每个企业，开发商充分利用资源，获取利益的重要手段。86526

虽然现在已经步入大数据时代，对于数据的分析主要依据收集到的大量数据，然而，有些时候在这些数据中有效信息并不充足，甚至不能满足统计分析的需求。本文综合整理对通信运行指标“话务量”的统计要求，采用以“少数据，贫信息”著称的灰色建模技术，首先详细阐述灰色建模的不同方法（GM模型，马尔可夫模型）、基本步骤，建立相应的模型方程，使系统的研究有一个合理的模型基础，满足研究的研究框架。

本文从通信运行指标的角度入手，选取话务量这一方面。通过对现有的信息，数据以及人们对未来的希望兼顾考虑，结合GM(1,1)模型以及灰色加权马尔可夫模型深入研究后续通信运行发展工作。详细的说，即先仅通过2014、2015两年的1至5月月话务量数据并且建立GM（1,1）模型来预测2016年1至5月的话务量数据，随后单独用12,13,14,15四年的每月数据分别预测2016年相对应的数据，并且用加权马尔可夫的方法来修正预测的残差，从而来提高预测的精确性。最后将最终预测值与最初预测的数据比较相对模拟误差，凸显出本方法的相对准确性。

毕业论文关键词：灰色模型；通信；话务量；GM（1,1）；马尔可夫链

Abstract Traffic is known as the telecommunications load。 It often represents the carrying capacity of communicating equipments。 On the other hand, it also reflects the user's demands for communication。 In addition to this, traffic is influenced by the number of users, frequent communicating degree and the time variation of each communication, so researches on communications of user traffic become very important to develop appropriate traffic scheme, so as to each enterprise, developers to make full use of resources and access to benefits 。

Although we have entered the era of big data, the analysis of data is mainly based on a large number of data which we collect, sometimes the data information is not sufficient, even can’t meet the demand for statistical analysis。 After synthesizing the arrangement of operating index of communication the traffic statistical requirements, grey model technology which was known as "little data, poor information" was used。 Firstly, it elaborates grey modeling of different methods (GM (1, 1) model, Markov model), the basic steps, the establishment of appropriate model equations, so that the system has a reasonable model based, it meets the research framework。

The writer starts from the angle of communicating indexes, choosing traffic which is an aspect of the indexes。 Apart from this, combined with the comprehensive consideration of existing information, data and the people's future hope, the subsequent communicating operation and development work was studied deeply by GM (1,1) model and grey weighted Markov model。 More specifically, author firstly establishes GM(1,1) model to predict traffic data from January to May in 2016 by correspondent data in 2014 and in 2015。 Then the four years of monthly data which are from 2012 to 2015 are used to forecast the 2016 corresponding data。 In addition, the predicted residual was modified positively by using weighted Markov chain to improve the prediction accuracy。 Finally, comparing the final predicted value with the initial prediction of the data in the relative simulation error is to highlight the relative accuracy of the method。

Keyword:  Grey  model; Correspondence; Traffic; GM（1,1）model; Markov model