基于数据挖掘的变电站告警数据分析和巡检策略研究

Research on substation alarm data analysis and patrol inspection strategy based on data mining

鉴于目前高压变电站告警数据多、运维人员巡检效率低的情况,提出了一种基于关联规则的告警数据分析方法。通过关联规则得到大量告警事件间隐含的先导—后继关系,分析故障发生后运行人员工作站记录的告警报文,这有助于运维人员找出故障的根本原因;并参考相关的电力设备状态量评定标准,综合告警事件间的关系与各巡检单元权重系数,确定系统频繁告警后巡检的优先顺序,以提高巡检效率。最后,通过西北电网某变电站主变近一年的告警数据来具体说明这种方法,研究发现该站主变存在着由于冷却器就地操作、直流控制电源故障引起的频繁冷却器油流停止告警,需要重点巡视冷却装置控制系统;并提出了改进的巡检策略来提高巡检效率。该方法为变电站实施新的巡检策略提供理论支撑。

Considering that there are many alarm data in high voltage substation at present and the inspection efficiency of operation and maintenance personnel is low,this paper presents a method based on association rules for alarm data analysis.After identifying the pilotsuccessive relationship covered in the numerous alarm data through the association rules,and alarm messages recorded by operator’s workstations after the failure are analyzed,which is helpful for the operation and maintenance personnel to find out the root cause of the failure.Meanwhile,referring to the interrelated equipment status assessment standards,alarm events relationships and weights of each inspection unit determine the priority of the inspection after OWS frequently alarms,which contributes to improve inspection efficiency.Finally,this method is illustrated through the alarm data of a station transformer of northwest power grid for nearly one year,and the reasons that the transformer frequently alarm are raised.It is found that the main transformer of the station has frequent oil flow stop alarm caused by the local operation of the cooler and the fault of DC control power supply.It is necessary to focus on inspecting the control system of the cooling device.An improved inspection strategy is put forward to improve the inspection efficiency.This method provides theoretical support for the implementation of new inspection strategy in substations.