基于特征事件的PON网络故障分析诊断与应用

摘要：为提升无源光网络（PON）运行稳定性以及相关故障的快速定位能力，提出了基于事件特征的综合故障处理方法。该方法首先通过事件去噪提升查准率，然后通过事件聚合提升查全率，并基于关联数据，借助算法演进完成相关网络问题的快速感知、预警与定位。以PON网络为例，通过关键事件（告警）的选取，并基于上述流程实现PON链路故障的快速高效定位。典型网络分析和仿真表明，该方案去噪比与聚合查全率都能达到80%以上，能够显著提升PON网络运维效率。

关键词：去噪；聚合；查准率；查全率；无源光网络；信号丢失；特定光网络单元信号丢失

Abstract: To enhance the operational stability and rapid fault localization capability of passive optical networks (PON), this paper proposes an integrated fault processing method based on event features. Firstly, the method enhances precision through event denoising, and then improves recall through event aggregation. Combined with auxiliary data, the method employs iterative optimization algorithms to achieve rapid detection, early warning, and accurate localization of network anomalies. For PON, the scheme selects key alarm events and applies the above processing flow to efficiently locate link faults. Typical network analysis and simulation results demonstrate that the proposed scheme achieves over 80% denoising accuracy and event aggregation recall, which can significantly boost PON operation and maintenance efficiency.

Keywords: de-noising; aggregation; precision; recall; PON; loss of signal; LOSi