Abstract: The prediction of smoke movement and temperature distribution when a fire occurs is a popular technology in the field of construction and fire protection.At present,this prediction has not been combined with deep neural network technology.Aiming at the current situation that the prediction of fire smoke movement and temperature distribution is cumbersome and the prediction accuracy is low,a prediction model of fire smoke movement and temperature distribution based on trend feature vector is proposed.The deep learning methods are used to train and predict relevant data,which is of great significance to reveal the law of fire occurrence and development and can provide auxiliary information for fire-fighting and fire evacuation.The proposed model can extract the trend features in the fire time series data,and uses these features as a priori knowledge to accelerate and optimize the training process of the deep neural network.This paper designs LSTM-TFV (LSTM based on Trend Feature Vector) algorithm.Experimental results show that the proposed prediction model improves the accuracy of the prediction of fire smoke movement and temperature distribution,and realizes efficient and convenient fire time series data prediction.

Key words: Deep learning, Fire prediction, Smoke flow, Temperature distribution, Trend characteristics