基于决策树的变电设备带电感应报警方法

A Decision Tree-based Live Induction Alarm Method for Substation Equipment

为了提高变电设备运行的安全性与可靠性,并使报警系统能够迅速对变电设备的异常运行状态作出报警反应,开展了基于决策树的变电设备带电感应报警方法的研究。首先,采集与预处理变电设备带电感应数据;然后,提取与变电设备运行状态和故障模式相关的特征向量,构建决策树模型实时监测变电设备的运行模式和运行状态;最后,设定报警阈值,实现对变电设备故障的实时报警。实验测试结果表明应用所提方法后,报警反应速度加快,平均报警时间均在2 s内。

In order to improve safety and reliability of substation equipment operation and enable prompt response of alarm system to to abnormal operating conditions,a decision tree-based live induction alarm method for substation equipment was studied.First the live induction data of substation equipment were collected and preprocess.Second by extracting feature vectors related to operational status and fault modes of substation equipment,a decision tree model was constructed to monitor the operational mode and status of substation equipment in real-time.Finally the alarm threshold was determined to achieve real-time alarm for substation equipment faults.The experimental test results showed that the application of the proposed method achieved faster alarm response with average alarm time of less than 2 seconds.