基于聚类分析的配电网无整定保护技术

提出一种基于聚类分析的配电网无整定保护方法，采集被保护馈线在不同运行条件下的故障信息，提取多种故障特征量，组成历史特征样本集；借助聚类分析算法将历史特征样本集划分为故障类与非故障类，并得到相应的聚类中心；实时监测被保护馈线运行状态，通过计算实时特征样本与聚类中心的空间距离对实时特征样本的属性进行判别，从而实现保护判断。仿真结果表明，该方法无需设置保护整定值，不受系统运行方式与过渡电阻大小的影响，对于不同故障条件下的接地故障均能实现准确判断。

A novel clustering analysis based non-setting protection technique for distribution systems

A novel non-setting protection method for distribution systems fault based on clustering analysis is proposed in this paper.The fault information of the protected feeder under different operating state is extracted to construct the historical feature sample set.The historical feature samples are divided into fault group and non-fault group by clustering algorithm and the cluster centers are obtained.By comparing the distances between the real-time feature sample and the cluster centers,the attribution of the real-time feature sample is obtained as well as the protec-tion result.The simulation result indicates that the non-setting protection method can realize the protection without the setting value.And the impacts of operating state and transition resistance can be neglected.