基于多源异构数据融合的电力工程知识图谱架构算法

Power engineering knowledge graph architecture algorithm based on multi⁃source heterogeneous data fusion

针对电力行业数字化进程中出现的对于多源异构类型数据的分析效率低、概念逻辑混乱等问题,开展了基于数据融合的电力工程知识图谱架构算法研究。在收集电力行业数据和外部数据并整理相关领域专业术语的基础上,构建了电力工程知识图谱,采用CRF算法将非结构化文本信息转化为结构化信息。最终将典型相关分析(CCA)与深度神经网络相结合,利用逐层语义匹配的方法,架构出满足域私有网络和域共有网络的深度语义匹配模型。通过设置对照组进行对比测试结果表明,文中提出的多源异构数据融合模型可以有效提高不同领域实例数据的融合精度,较两种对比算法分别提高了8.32%和11.7%,具有较为理想的综合性能。

Aiming at the problems of low efficiency of multi-source and heterogeneous data analysis and confusion of concept logic in the process of power industry digitization,the algorithm of power engineering knowledge map architecture based on data fusion is studied.On the basis of collecting power industry data and external data,and sorting out the related professional terms,the power engineering knowledge map is constructed.CRF algorithm is used to transform unstructured text information into structured information.Finally,combining Canonical Correlation Analysis(CCA)with deep neural network,using the method of layer by layer semantic matching,the deep semantic matching model satisfying domain private network and domain common network is constructed.The results show that the proposed multi⁃source heterogeneous data fusion model can effectively improve the fusion accuracy of case data in different fields,which is 8.32%and 11.7%higher than the two comparison algorithms,respectively,and has ideal comprehensive performance.