通讯设备的定位与维修速度是影响通信专线稳定性的关键因素。为提高告警设备故障定位及检修效率,本文设计与实现了一个通信设备劣化性能与告警数据间关联规则自动生成系统。首先,采用Auto-Encoder模型对连续性能数据进行离散化处理,并利用转移矩阵寻找转移概率较小之离散性能数据,将其定义为劣化性能,这一过程旨在挖掘设备的劣化性能特征,统一劣化性能与告警数据的维度,生成与告警数据统一维度的时空二维布尔型数据集。接下来,本文采用深度优先的相互关联规则挖掘方法,将统一维度的二值化数据集作为输入,生成劣化性能与告警数据之间的时序传导规则。同时,为了提升数据处理效率,本文利用Jaccard系数在时间轴上对劣化性能数据进行压缩,合理降低数据规模,从而加速关联规则的生成过程。最后,通过结合离线学习和在线更新策略,系统能够实时输出劣化性能与相关数据之间的关联规则,从而推测并定位关联告警设备。本文基于运营商实际的政企专线数据集,进行了相关的运行测试。测试结果表明,该系统能够准确、快速地生成设备劣化性能数据与告警数据之间的关联规则,并实现根据在线劣化性能数据实时预测并定位关联告警设备的功能。The positioning and repair speed of communication equipment are key factors affecting the stability of communication leased lines. To improve the fault location and maintenance efficiency of alarm devices, this paper designs and implements an automatic system for generating association rules between the degradation performance of communication equipment and alarm data. First, the Autoencoder model is used to discretize continuous performance data, and a transition matrix is applied to identify discrete performance data with a small transition probability, which is then defined as degradation performance. This process aims to extract the degradation performance features of the equipment and unify the dimensions of degradation performance and alarm data, generating a spatiotemporal two-dimensional Boolean dataset with the same dimensions as the alarm data. Next, this paper uses a depth-first mutual association rule mining method, taking the unified-dimensional binary dataset as input to generate temporal propagation rules between degradation performance and alarm data. To enhance data processing efficiency, the Jaccard coefficient is employed to compress the degradation performance data along the time axis, reasonably reducing the data scale, thus accelerating the rule generation process. Finally, by combining offline learning and online updating strategies, the system can output association rules between degradation performance and related data in real-time, thereby predicting and locat
暂无评论