基于近似匹配模型的电网多模态数据检索研究

Research on power grid multimodal data retrieval based on approximate matching model

电网运行过程需要多种模态形式数据的支持,且多模态数据的复杂性较高,存储量增大,导致其检索难度较高,易出现较多相似冗余数据。为解决这一问题,提出基于近似匹配模型的电网多模态数据检索方法。引入树匹配理论构建模型,分析元数据与近似查询树映射关系。根据电网多模态数据近似匹配过程,计算查询树与元数据之间的编辑距离。依据极大似然估计原则,得到先验参数最优估计结果。按照由大到小顺序排序待检索文档与查询数据,由此完成多模态数据检索。实验结果表明,该方法检索精准度与实际数据存在最大为0.50 kB的误差,检索精准率最高为98%,说明使用该方法检索结果完整且精准率高。

The power grid operation process needs the support of multi-modal data,and due to the high complexity of multimodal data and the increase of storage capacity,it is difficult to retrieve it,and more similar redundant data are prone to appear.To solve this problem,a multimodal data retrieval method based on approximate matching model is proposed.The tree matching theory is introduced to build the model,and the mapping relationship between metadata and approximate query tree is analyzed.According to the approximate matching process of multi-modal data,the edit distance between query tree and metadata is calculated.According to the principle of maximum likelihood estimation,the optimal estimation results of a priori parameters are obtained.The documents to be retrieved and the query data are sorted in the order from large to small,thereby completing the multi-modal data retrieval.The experimental results show that the maximum error between the retrieval accuracy of this method and the actual data is 0.50 kB,and the highest retrieval accuracy is 98%,indicating that the retrieval results using this method are complete and the accuracy is high.