Text Classification Method for Urban Rail Transit Fire Accident Cases Based on Word2Vec + LSTM

Fire accidents are the type of accidents in urban rail transit that occur frequently, are extremely dangerous, and cause huge losses, which and bring many difficulties to the safe operation and management of urban rail transit. Aiming at the data processing of urban rail transit fire accident cases, this paper first collects urban rail transit fire accident cases through multiple channels and divides them into five categories according to the fire causes: electrical reasons, mechanical reasons, artificial arson, improper operation, and accidental fire, and then establishes A text classification model based on Word2Vec feature processing and LSTM classification prediction is implemented to automatically classify fire texts with high accuracy, thereby solving the problems of time-consuming, low-quality, and strong subjectivity in manual text labeling. Furthermore, it provides high-quality data support for realizing refined risk prevention and control and accident management. Finally, this paper uses model comparison experiments to further verify the superiority of LSTM as the model backbone network.