基于ISSA-XGBoost模型的多特征融合露天矿卡车行程时间预测

针对露天矿运输系统卡车行程时间预测问题，提出了一种基于特征选择及改进麻雀算法优化XGBoost的露天矿卡车行程时间预测模型。模型充分考虑了卡车特征、道路特征、气象特征以及时间特征对卡车行程时间的影响，并使用皮尔逊系数法深入分析影响因素的贡献度。针对麻雀算法中全局搜索能力薄弱的问题引入反向学习和螺旋搜索策略，以提高算法的收敛性能。最后，使用改进的麻雀算法对XGBoost的关键参数进行寻优，进而构建露天矿卡车行程时间预测模型。选取国内某大型露天矿卡车调度系统采集的数据进行仿真模拟，并将所提出模型与SVM、BP、RBF和RF等其他机器学习模型进行对比。结果表明：所提出模型的预测误差均低于其他模型，相关系数可达0.981 9。开发的模型和分析结果可以极大地帮助决策者规划、运营和管理更高效的露天矿运输系统。

Multi-feature fusion open pit mine truck travel time prediction based on ISSA-XGBoost model

For the problem of truck travel time prediction in an open pit mine transportation system, an open pit mine truck travel time prediction model based on feature selection and an improved sparrow algorithm to optimize XGBoost is proposed. The model fully considers the influence of truck, road, weather, and time characteristics on truck travel time, and uses the pearson correlation coefficient to analyze the contribution of the influencing factors. The paper introduces reverse learning and spiral search strategies to address the weak global search capability issue in the sparrow search algorithm. Finally, an improved sparrow algorithm is employed to optimize key parameters of XGBoost, resulting in the construction of the predictive model for open-pit mine truck travel time. In the experiment, the travel time of a truck collected by a truck dispatching system of a large open pit mine is selected for simulation, and the proposed model is compared with other machine learning models such as SVM, BP, RBF, and RF. The experimental results show that the prediction errors of the proposed model are all lower than those of other models, and the correlation coefficient can reach 0.9819. The developed model and analysis results can greatly help decision-makers to plan, operate, and manage a more efficient open pit mine transportation system.