基于相似度计算的学习型模板库在虚回路设计和校验中的应用

Application of similarity-calculation-based learning template library in design and check of virtual circuit

智能变电站二次回路设计基本相通,由于虚端子描述尚不统一,使得新建站虚端子连接无法复用,仍采用人工点对点连接,导致效率不高。提出自动创建虚回路模板库,通过关键字匹配实现虚回路的自动设计和完整性及正确性校验,其核心在于学习型模板库海量学习已有配置描述(SCD)文件,采用中文分词技术进行关键字提取,引入经典的RKR-GST算法完成虚端子描述的相似度计算,从而进行关键字的匹配整合,从而创建、整理和完善虚回路模板库。试验证明,通过该方法对新建智能变电站的虚端子连接效率和准确性都有较大提升。

Though the design principle of smart substation secondary circuits is basically same,because the virtual terminal description has not been unified,the virtual terminal connection of new substation is still manually implemented with low efficiency. A method is proposed to automatically create the virtual circuit template library for realizing the automatic design and the integrity and correctness check of virtual circuits based on the keyword matching. Massive SCDs（Substation Configuration Descriptions） stored in the learning template library are explored,the Chinese word segmentation technology is applied to extract the keywords and the classic RKR-GST algorithm is introduced to calculate the similarity of virtual terminal descriptions for the matching and integration of keywords as well as the creation,organization and improvement of the virtual circuit template library. Experiments show that,with the proposed method,the efficiency and accuracy of virtual terminal connection for new smart substation are greatly improved.