基于大数据模糊关联分析的网络运行态势感知

Analyzing and forecasting network status accurately and controlling network operating situation in real time is an important guarantee to build a credible, manageable and controllable network, and is also a problem which needs to be solved by modern network management system urgently. The technology based on network situation awareness is the development direction in future network management. In this project, fuzzy theory, artificial intelligence and big data are combined organically. Effective solutions and approaches for uncertain problems in big data analysis are explored. The technical supports are provided for refined and integrated network management and control..This project makes systematic research on the key technologies of network situation awareness from full and multi-level perspectives of network infrastructure, network services, network information system and service guarantee system. Compared with traditional methods, it is an innovative research which is user-oriented, based on business and aimed at runtime. This project studies big data fuzzy clustering in situation awareness from the perspective of theory and methods, and it is of great innovation and research value to technologies such as the mechanism for complex system state characterized by fuzzy clustering and the optimal segmentation of the characterization sets, calculation of the similarity of feature vectors, active learning in situation awareness as well as fuzzy evaluation and grey correlation of network situation, and methods of network warning and resource optimum reconfiguration based on operating situation...The project involves multidisciplinary cross and integration field, and has important significance for the subject development and industry application.

准确地分析和预测网络状态，实时掌控网络运行态势，是构建可信、可管、可控网络的重要保障，也是现代网络管理急需解决的问题。基于网络态势感知是未来网络管理的发展方向。本项目将模糊理论、人工智能与大数据有机结合，探索大数据分析中不确定性问题的有效解决方法和途径，为实现精细化和综合化的网络管控提供技术保障。.项目从网络基础设施、网络业务、网络信息系统和服务保障系统全方位多层次的角度出发，对网络态势感知的关键技术展开系统研究。与传统方法相比，是一种基于业务、面向用户、针对运行时刻的创新性研究。项目从理论和方法上研究大数据模糊聚类对复杂系统状态的表征机理、态势特征集合的最优分割及相似度计算、态势感知的主动学习及模糊评判和灰色关联等技术，以及基于运行态势的网络预警和资源优化重构等方法，具有较强的创新和研究价值。.本项目的研究是一个涉及多学科交叉、融合的领域，对于学科发展和行业应用都具有非常重要的意义。

当今信息时代，网络已经成为现代社会严重依赖的神经枢纽。保障网络正常、稳定、高效地运行，实时掌控网络运行状况，科学地分析和预测网络状态及变化趋势，是构建可信、可管、可控网络的重要保障，也是现代网络管理和运维管理急需解决的问题。. 本项目将模糊理论、人工智能与大数据有机结合，从多层次、多角度对网络态势感知的相关理论和方法展开系统研究。构建了网络运行态势评估分析的指标体系，研究了提取网络态势特征的自适应模糊聚类方法，以及对特征信息进行模糊关联和分析的关键技术。将模糊系统与神经网络及人工智能其它技术相结合，应用到网络运行态势大数据分析中，建立了网络态势模糊神经网络分析系统，并对模糊神经网络的模型结构优化和学习算法进行了深入研究，有效解决了态势感知的一致性和稳定性问题。项目还对网络运行态势预测的关键技术和方法展开研究，从网络业务和用户感受的综合角度出发，研究了网络运行态势的预警机制，设计了该问题的综合评价指标和系统的评价框架，解决了评价过程中的关键技术，基于网络运行态势的分析和预测，研究了网络资源的重构与动态调整方法，将遗传算法和模拟退火算法相结合，保证了网络资源调整的合理性和有效性。本项目取得的上述成果，为解决网络运行大数据分析中不确定性问题提供了有效的解决方法和途径，为实现精细化和综合化的网络管控提供技术保障。. 在项目研究期间内，发表了期刊和会议论文12篇，发明专利3个，并以本项目为依托，培养了博士、硕士研究生近20人。.关键词：网络态势感知，模糊聚类，模糊关联分析，模糊神经网络