GSK-XGBoost模型在井底风温预测中的应用*

为防治矿井热害，针对矿井井底风温在预测过程中精度较低的问题，提出1种网格搜索法结合K折交叉验证优化XGBoost的预测模型。通过分析确定影响井底风温的主要因素，使用网格搜索算法结合K折交叉验证，进行迭代缩小搜索范围并调参，选取最优参数配置，实现对XGBoost模型的优化，得到预测结果并与其他模型进行比较。研究结果表明:初始参数经优化后，当最大回归树深度为3且学习速率为0.1时，XGBoost回归模型性能最佳，与随机森林模型、BP神经网络模型、T-S模糊神经网络模型相比，平均相对误差分别降低了2.12%,0.88%,0.3%，均方根误差分别降低了0.66,0.24,0.11 ℃。

Application of GSK-XGBoost model in prediction of wind temperature at well bottom

In order to prevent the thermal hazard in the mine,aiming at the problem of the low accuracy in predicting the wind temperature at the well bottom of mine,a prediction model by using the grid search method combined with the K-fold cross-validation to optimize the XGBoost was proposed.The main factors affecting the wind temperature at the well bottom were determined through the analysis,then the grid search algorithm combined with the K-fold cross-validation were used to conduct the iteration to narrow the search range and adjust the parameters.The optimal parameter configuration was selected to realize the optimization of the XGBoost model,and the prediction results were obtained and compared with those of other models.The results showed that after the optimization of initial parameters,when the maximum regression tree depth was 3 and the learning rate was 0.1,the XGBoost regression model had the best performance.Compared with the random forest model,BP neural network model and T-S fuzzy neural network model,the average relative error was reduced by 2.12 %,0.88 % and 0.3 %,respectively,and the root mean square error was reduced by 0.66 ℃,0.24 ℃ and 0.11 ℃,respectively.