Research on Alarm Information Oriented Fault Event Classification Method

China is building a new power system with new energy as the main body, which puts forward higher requirements for the intelligence of power grid dispatching, and the processing of power grid alarm information also needs to be more intelligent. In this paper, a power grid fault event classification method based on LSTM is proposed to realize the end-to-end discrimination of power grid fault event types and meet the needs of power grid intelligence. Firstly, the sample data set of power grid alarm information required for modeling is constructed; Then, the alarm information is vectorized by using the word2vec model in natural language processing technology; Then, using LSTM neural network, a power grid fault event classification model based on LSTM is constructed to determine the type of power grid fault equipment end-to-end and whether there is protection and circuit breaker failure; Finally, through the comparison of numerical examples, it is verified that this method has good expressiveness in fault event classification.