Matching and generation method of virtual terminals in smart substation based on reinforcement learning and semantic analysis

Abstract

Aiming at the problems of long time, heavy workload and error prone in virtual terminal connection of secondary equipment in smart substation, an automatic matching and generation method of virtual terminal in smart substation based on reinforcement learning and semantic analysis is proposed. Firstly, according to the similarity of input and output virtual terminals in data format and text description, the semantic similarity of virtual connections is calculated through historical virtual connection data, and the automatic matching model of virtual terminals between secondary devices is established; Secondly, the reinforcement learning method is introduced into the semantic similarity calculation model to automatically group the text, and the reinforcement learning algorithm is used on the semantic similarity calculation Siamese network model. It improves the semantic loss faced by the long-term and short-term memory model when extracting the semantics of sentences; Finally, combined with the actual situation of virtual terminal connection, the evaluation index for testing the virtual terminal automatic connection method is designed, and the example of a 220 kV smart substation in South China Power Grid shows that this method can effectively reduce the virtual terminal with wrong connection, reduce the manual workload, and effectively improve the virtual terminal connection efficiency.