In view of the current low efficiency of manual processing of massive monitoring alarm information and the need for deepening the application of power grid intelligence technology, an autonomous identification method of power grid equipment operation and maintenance alarm events based on natural language processing technology is proposed, which integrates neural network and unsupervised learning. The text of substation equipment alarm signal is vectorized based on word2vec algorithm, the time-density correlation between multiple alarm signals is established based on DBSCAN algorithm, and the "eventalization" model of alarm signal sequence is constructed based on TF-IDF algorithm. This paper proposes an application method based on natural language processing technology combining neural network and unsupervised learning algorithm to screen key "eventalization" alarms from a large number of discrete alarms, so as to realize the response efficiency and reliable identification of power grid monitoring alarm events.

eventalization; Natural language analysis; Neural network; Unsupervised learning; Density clustering