摘要
非侵入式负荷分解方法是对系统中的负荷用电行为进行辨识的一种方法,是智能电网体系建设的重要环节。为了达到监测系统中负荷的目的,基于系统的暂态过程,对系统中的负荷进行辨识,即基于负荷的投切和换挡操作,提出用于负荷暂态过程检测和提取的算法;再选取合适的特征量表征每一类暂态过程,将得到的特征量经过归一化处理形成样本集,应用决策树中CART算法通过训练集生成负荷暂态类型判定的决策树,利用测试集来判定系统的暂态过程,以实现监测系统中负荷的目的,并通过改进决策树,提出了多层决策树算法,解决了算法误判的问题;最后通过实测数据验证了所提算法对系统中负荷的有效监测。
Non-intrusive load decomposition method is a technology that determines the state of load switching and is the key of smart grid system construction.In order to monitor load,it was identified based on the system transient process.The method for detection and extraction of load in transient process was proposed based on the load switching and shifting.The appropriate characteristic values were chosen to represent each transient process.The obtained characteristic values were normalized to form sample set.CART algorithm in decision tree was used to train the sample set and get the decision tree for judging the transient category.The test set was used to check the system transient process for monitoring the load.The multi-layer decision tree algorithm was proposed to solve the algorithm＇s misjudgment problem.The observed data verified the effectiveness of the proposed method for monitoring the load in grid system.
作者
曹敏
魏龄
邹京希
吴振升
付华
CAO Min;WEI Ling;ZOU Jing-xi;WU Zhen-sheng;FU Hua(Electric Power Research Institute,Yunan Power Grid Co.Ltd.,Kunming 650217,China;School of Electrical Engineering,Beijing Jiaotong University,Beijing 100044,China)
出处
《水电能源科学》
北大核心
2018年第8期177-180,共4页
Water Resources and Power
关键词
非侵入式负荷监测
暂态过程
特征提取
CART算法
多层决策树算法
non-intrusive load monitoring
transient process
feature extraction
CART algorithm
multi-layer decision tree algorithm
