With the advent of the intelligent era， computer simulation algorithms such as machine learning and deep learning have played a role in many fields such as aerospace， medical treatment， education and communication. For the traditional industry of mining， the concept of smart mine has become a research hotspot of relevant researchers in recent years. Intelligent technologies such as machine learning have been used in pedestrian detection， gas prediction， coal rock identification has been successfully applied to practical production， but the intelligent acquisition of parameters in intelligent ventilation system is still in a blank. Therefore， under the background of smart mine， aiming at the problem that the mine intelligent ventilation system can’t obtain the wind speed in time and then complete the subsequent ventilation system solution and optimization， the training set required by the neural network is obtained by using the simulation of tunnel wind speed distribution in ANSYS. Based on manual measurement and wind speed sensor monitoring data， the prediction model of roadway average wind speed based on gated recurrent unit neural network was constructed. Firstly， the neural network model was proposed， and then the Adam optimization algorithm was used to preprocess the data such as outlier processing and normalization. After the structural processing of the wind speed at the monitoring points of the roadway with different shapes， it was used to train the neural network to find out the strong nonlinear relationship between the wind speed at each point and the average wind speed， so that the predicted wind speed is close to the actual average wind speed of the roadway. Finally， the prediction model of roadway average wind speed based on GRU neural network was constructed. Taking the measured data of Wangjialing coal mine as the model test set， the results show that the GRU neural network model has high precision and strong generalization ability， and can obtain the average wind speed of roadway， which will provide a roadway average wind speed prediction model with advanced technology， scientific process and accurate results for the mine intelligent ventilation system. Moreover， the strong prediction ability of in-depth learning will provide intelligent data for the solution and optimization of ventilation network， it can be extended to the acquisition of ventilation parameters in other metal mines to popularize the intelligent acquisition of ventilation parameters.

http://www.goldsci.ac.cn/article/2021/1005-2518/1005-2518-2021-29-5-709.shtml