Accident brief report rapid generation method based on automatic encoder

The invention provides an accident brief report rapid generation method based on an automatic encoder, and the method comprises the steps: 1, preparing data, cleaning and preprocessing the data, and guaranteeing the quality and accuracy of the data; 2, model training: inputting accident data into a model to train an automatic encoder, and extracting key features and semantic information in the data; 3, key information extraction: carrying out key information extraction on the accident related text through an LSTM-CRF model so as to generate a subsequent brief report; and 4, generating an accident brief report, and quickly generating the accident brief report based on the extracted information and context in combination with a predefined accident brief report template. According to the method, the powerful capability of the pre-training language model BERT-AE is utilized, and the aim of rapidly generating the power grid distribution network accident brief report is achieved through training of the automatic en