基于树增强朴素贝叶斯分类的台区关口表挂接关系在线校验方法

An Online Verifying Method for the Connection Between Transformer Area and Its Gate Meter Based on Tree-Enhanced Naive Bayes Classification

针对台区关口表挂接关系人工现场核查耗时耗力的问题,提出基于树增强朴素贝叶斯分类的在线校验方法。对于一个台区,首先从用电信息采集系统提取台区用户和变压器(关口表)的电压曲线;接着,分别计算一段时间内每个用户与变压器三相电压曲线之间的皮尔逊相关系数;然后,统计所有用户皮尔逊相关系数在不同区间的分布数量;最后,以台区用户相关系数在不同区间的分布数量为输入属性特征,以台区关口表挂接关系是否正确为输出属性,采用树增强朴素贝叶斯模型构建台区关口表挂接关系在线校验模型。该方法在某地市电网公司试运行,其应用效果表明,可有效提升台区关口表挂接关系数据的正确性。

In order to solve the problem of the time and cost consumption of the manual verification of the connection between the transformer area and the gate meter, an online verification method based on tree-enhanced naive Bayes classification was proposed. For a transformer, the voltage sequence data of low-voltage users and transformer(gateway meter)are firstly extracted from the electricity information acquisition system.Then, the Pearson correlation coefficients of voltage curves for a period of time between each user and the three-phase of the transformer are calculated respectively. Then, the distribution number of all users Pearson correlation coefficients in different intervals is counted. The Pearson correlation coefficient distribution in different interval number is the input attribute feature, and whether the connection of the transformer area and its gate meter is correct for the output attributes. Tree-enhanced naive Bayes classification was used to construct the model of verifying the connection between transformer and its gate meter. This method has been used in a municipal power grid company, and its application results show that it can effectively improve the correctness of the connection between the transformer area and the gate meter.