Intelligent Noise Reduction for Alarm Events

In this paper, we propose a decision tree-based intelligent noise reduction technique that utilizes machine learning algorithms to accurately remove noise from alarm events by comprehensively considering multiple metrics. A hybrid ratio of two algorithms, LightGBM and CatBoost, and systematic threshold selection experiments are used to optimize the model performance, and an auto-start touting mechanism is introduced to improve the stability and robustness of the system. Experiments demonstrate that the method performs well in terms of noise reduction rate, maintaining above 85% with an average of 88.6% and up to 91%. Overall, this intelligent noise reduction technique has achieved remarkable results in terms of alarm event processing efficiency, cost reduction, and accuracy improvement, providing reliable services for practical applications.