基于自适应核密度估计理论的抗差状态估计的性能分析及算例验证  

The Performance Analysis and Sample Verification for Robust State Estimation Method Based on Adaptive Kernel Density Estimation Theory

作　　者：刘阳升[1] 林济铿[1] 郭凌旭 蔡凝露 仝新宇 孟宪朋[5] 张耀先[6] 江伟[7] 

机构地区：[1]同济大学电子与信息工程学院,上海市嘉定区201804 [2]天津电力公司调度中心,天津市河北区300010 [3]天津市电力设计院,天津市河西区300200 [4]天津电力公司城西供电公司,天津市河北区300110 [5]国网河北省电力公司,河北省石家庄市050051 [6]天津泰达电力公司,天津市塘沽区300457 [7]福建省电力有限公司电力科学研究院,福建省福州市350007

出　　处：《中国电机工程学报》2016年第14期3845-3856,共12页Proceedings of the CSEE

摘　　要：针对基于自适应核密度估计的电力系统抗差状态估计新模型,提出了相应的求解策略,并通过其数学性质分析该方法对正常量测、可疑量测和不良数据的辨识性能。该算法对于正常量测具有与加权最小二乘法类似的收敛性好、渐进无偏的特点;对于不良数据,在迭代过程中通过自适应地降低其核密度带宽而等效降低其权重直至为零,相应减小或消除其影响而具有良好的抗差能力;对于可疑量测,在迭代中自适应核带宽逐渐减小,部分量测的等效权值逐步增大而被辨识为好数据,部分量测的等效权值减小而逐步被辨识为不良数据,从而实现了可疑数据的平滑逐步辨识,避免了"非好即坏"的判定而提高了对于不良数据的辨识能力。多个算例表明:新算法相较于现有其他抗差方法而言,具有较强的抗差能力,较高的计算精度及较快的计算速度。所提方法具有较强的工程应用前景。Aimed at the new proposed robust state estimation model based on adaptive kernel density estimation theory, a solving strategy was presented and whose identification performance ,an normal measurements, suspected measurements and bad data were analyzed through mathematic properties： For the nornlal measurements, the algorithm proposed is well convergent, unbiasedly and asymptotically similar with weighted least square （WLS） method; for the bad data, since the equivalent weights of which are adaptively decreased, the algorithm proposed is robust in identifying bad data; for the suspected measurements, during iteration process, the weights of part of which are gradually increased and are classified as normal data, while the weights of the rest are progressively decreased and are identified as bad data, thus the suspected measurements are smoothly and gradually identified to be normal or bad data and the identifying capability of bad data for the algorithm is raised due to avoiding a black or white judgment. Examples illustrate that： the algorithm proposed is much more robust in identifying bad data and with higher computing precision and speed compared with others. The new state estimation algorithm owns the powerful potential to be applied for practice power system.

关 键 词：抗差状态估计 自适应核密度估计 核密度带宽 不良数据辨识 

分 类 号：TM74[电气工程—电力系统及自动化]