基于数据挖掘的配电网规划问题关联性知识提取研究

Research on Correlation Knowledge Extraction for Distribution Network Planning Based on Data Mining

传统配电网规划中,配电网问题的发现以及相关影响因素之间关联性的建立依赖于规划人员的专业知识,忽视了配电网历史数据本身反映的变化规律,制约了规划问题查找的客观性。通过数据挖掘方式提取配网数据的特征,实现配电网问题的关联性知识提取及验证。构建配电网问题关联规则的数据挖掘模型,采用灰色关联法获取模型输入的电气特征指标,通过Apriori算法求取配电网问题与其影响因素之间的强关联性规则,并采用提升度和卡方检验,对输出的强关联性规则进行合理性校验。基于实际地区电网数据,挖掘配电网线路重过载、轻载问题与相关电气特征指标的关联关系,为配电网规划项目的制定提供有效依据,从而提高配电网规划的效率及精细程度。

In the traditional distribution network planning,the discovery of distribution network problems and the establishment of the correlation between related influencing factors depend on the professional knowledge of planners,ignoring the change law reflected by the historical data of the distribution network itself,which restricts the search for the objectivity of planning problems.Through data mining,the characteristics of distribution network data can be extracted,and the extraction and verification of relevant knowledge of distribution network problems can be realized.A data mining model for the association rules of distribution network problems was constructed,and the electrical characteristic index of the model input was obtained by using the grey relational method;the Apriori algorithm was used to obtain the strong correlation rules between the distribution network problems and their influencing factors,and the lifting degree and chi-square test were used to verify the rationality of the output strong correlation rules.Based on the actual regional power grid data,the relationship among heavy overload problems,light load problems and related electrical characteristic indicators of distribution network lines was mined to provide an effective basis for the formulation of distribution network planning projects,thereby improving the efficiency and refinement of distribution network planning.