3基于朴素贝叶斯算法的故障因子挖掘-南方电网技术.doc

文章编号： 中图分类号：TM76 文献标志码：A DOI： 基于大数据驱动的输电线路故障预警模型设计与分析 摘要：随着电网规模的迅速扩大，输电线路急剧增长，检测人员数量严重不足且检测质量良莠不齐，致使发展输电线路远程故障诊断成为建设智能电网的重点内容。基于大数据驱动的输电线路故障预警模型可有效解决以上问题，本文针对输电线路的上传数据单一、数量多、综合性差等问题，充分利用大数据技术挖掘潜在故障因子，采用朴素贝叶斯算法挖掘故障因子发生指数，随后结合可采集相关电气量、开关量、事件顺序信息、电网拓扑以及输电线路故障过程中故障录波装置所获取的数据，以基于时间序列相似性故障匹配对输电线路进行故障预警，诊断结果推送给相关工作人员，以辅助调度或运行人员提前制定解决方案、快速处理故障，加速电网恢复。 关键词：朴素贝叶斯算法；时间序列相似性匹配；大数据驱动；故障预警模型 Design and Analysis on Fault Warning Model of Transmission Line Based on Big Data Abstract: With the rapid expansion of the scale of power grid，the number of transmission line has increased dramatically. However, serious shortage of detective personnel and the quality of detection varies greatly, which leads to the development of remote fault diagnosis of transmission line to become the focus of the construction of the smart grid. Fault warning model based on large data can effectively solve the above problems. In order to solve the problem of single upload event and the amount of data but comprehensive poor, this paper uses large data technology to explore the potential failure factor and naive bayes algorithm as the data processing rule. Then the failure information of transmission line is tracked and diagnosed based on the time-sequence similarity matching, which is based on the collected electric data, the switch data, the event sequence information, power network topology and the data acquired by the fault recorder in the process of transmission line failure. Diagnostic results are pushed to the appropriate staff which can prepare the solution ahead of time, quickly troubleshoot and accelerate power grid recovery. Key words: naive bayes algorithm, time-sequence similarity matching, big data, fault warning model 0 引言 社会经济的发展，对电力的依赖性越来越强，而大规模停电事故往往始于电网某一线路的故障，并随着电网拓扑的强关联性及存在的脆弱环节而不断扩大。对于某些供电网架薄弱的地区，单一线路故障就可能导致负荷损失[1]。在不提高电价转移经济压力的前提下，运用先进的智能化技术，对输电线路故障进行预警，开展差异化运维，减少线路故障发生概率，有利于提高供电可靠性。 目前，针对输电线路故障的相关研究，大部分成果仅依赖现有数据信息对故障进行诊断，事实上，电力系统存在上传数据单一、数量多、综合性差等问题[2]，导致依靠此类数据难以实现故障预测的全面性和前瞻性，信息响应及时