摘要
为满足配电系统精益化运维对二次设备状态检修的要求,提出了一种多源信息融合的配电终端状态诊断方法。通过分析配电终端的运行监测信息,提炼出三种方便采集的状态特征量,从而建立了配电终端状态诊断的证据体系。提出了基于相关系数的基本概率分配函数构造方法,为避免故障终端样本数量不足导致的训练偏差,利用基于专家经验的加权平均模糊隶属度代替相关系数完成实际计算。利用D-S证据理论合成规则实现多元信息融合,并根据基本概率分配决策准则来判断配电终端的状态。最后,对配电终端的状态诊断进行了实例分析,相关结果验证了所提方案的有效性。
To satisfy the requirements for secondary equipment condition-based maintenance from distribution network lean operation, this paper proposes a condition diagnosis method for Distribution Terminal Units （DTU） on the basis of multi-source information fusion concept. Three kinds of easily collected condition features are put forward by analyzing DTU monitoring information, thus the corresponding evidence framework is built. The Basic Probability Assignment （BPA） function constructor is presented, where the expertise-based weighted average fuzzy membership instead of correlation coefficients is utilized for calculation in order to avoid the training biases derived from the lower quantity of faulty DTUs. Multi-source information fusion is achieved by D-S evidence combining theory, and DTU condition is diagnosed according to the BPA decision rules. Finally, a case study of DTU condition diagnosis is carried out, where the results validate the effectiveness of this proposed method.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2018年第1期30-36,共7页
Power System Protection and Control
基金
国家自然科学基金项目(51407128)
国家电网公司总部科技项目(5216A016000T)
国网湖南省电力公司科技项目(5216A515008)~~
关键词
配电终端
状态诊断
D-S证据理论
模糊隶属度
基本概率分配
distribution terminal units
condition diagnosis
D-S evidence theory
fuzzy membership
basic probability assignment
