基于改进遗传神经网络模型的通信网络故障识别和告警相关性分析方法

随着通信网络自身规模、带宽和复杂度的不断增加以及用户对网络服务质量要求的不断提高,迫切需要智能化的网络管理系统对当今高速通信网络进行有效和可靠的管理,而故障管理正在变的比以往任何时候都更加困难和重要。当网络产生某一故障或失效时,往往在短时间内产生成千上万个告警信息,因而分析这些告警的相关性也变得更加复杂。现有的一些告警分析系统均在不同程度上存在可扩充能力差,难于应付复杂局面,缺乏学习能力等不足。本文提出了一种基于改进遗传神经网络模型的故障识别和告警相关性分析方法。实验表明,这种方法可克服一般告警相关性分析方法的局限,不仅简单,而且在网络学习和训练效率上也高于传统的BP算法,标准遗传算法和一般的自适应遗传算法。

A FAULT-IDENTIFICATION AND ALARM-CORRELATION METHOD BASED ON IMPROVED GA-NN MODEL IN COMMUNICATION NETWORKS

As a result of the rising demand for services and the resulting increase in size, bandwidth and complexity, fault management in today's high speed communication networks is becoming even more difficult. When a network problem or failure occurs, it is possible that a very large volume of alarm messages is generated, while alarm correlation is a potentially complex problem. Though some existing alarm correlation systems nowadays have different drawbacks such as lack of scalability, hindered by solving complexity, or no learning process, etc. This paper presents a fault-identification and alarm-correlation method based on improved GA-NN model in communication networks. The experimental results show that this method is simple, which not only overcomes the disadvantages of normal alarm correlation ways, but also improves the dynamic character, training accuracy and efficiency greatly than BP algorithm, BGA algorithm and AGA algorithm do.