Abstract: To address the challenges of low efficiency in manual verification and reliance on non-standard virtual terminal definitions in substation configuration description(SCD) files in intelligent substations, an automatic verification method for virtual circuits based on an improved Bi-directional long short-term memory(Bi-LSTM) network is proposed. Firstly, the Word2vec model is used to vectorize virtual terminal description texts, effectively resolving the issue of unstructured text in virtual terminal descriptions and enabling their efficient recognition and analysis in subsequent processes. Secondly, an attention mechanism combined with the Bi-LSTM network is incoporated to comprehensively extract the semantic features of the virtual terminal description texts. By calculating the similarity between non-standard and standard virtual terminals, intelligent mapping of non-standard virtual terminals is achieved and a standard virtual terminal library is constructed as foundational data support. Subsequently, verified SCD files and the standard virtual terminal library are used to build a virtual circuit template library, which effectively stores and manages virtual circuit information for various intelligent substation devices. Finally, based on the virtual circuit template library, automatic verification of virtual circuits are realized, significantly improving the efficiency and accuracy of the verification process. Experimental results demonstrate that the proposed method not only enhances the adaptability and accuracy of virtual circuit verification in intelligent substations, but also substantially reduces the workload associated with manual verification, greatly improving the level of automation in intelligent substation operations. This research provides crucial technical reference for the future verification and management of virtual circuits in intelligent substations.