摘要
针对露天矿卡车从装载点到卸载点的运输循环总时间预测问题,在考虑卡车特征、道路特征和天气特征影响的情况下,通过支持向量机(SVM)、Adaboost和随机森林(RF)3种算法构建了露天矿卡车运行状态时间预测模型。以某大型露天矿山的卡车调度系统采集数据为例,结果表明,针对运行状态时间预测,采用SVM、Adaboost和RF 3种算法预测结果均优于传统平均法,最佳机器学习方法的平均绝对误差降低了81.52%;同时,将运行状态作为行程时间预测的最小单元进行组合预测,比用单一方法预测整个运输过程时间更好,其平均平方误差和平均绝对误差分别相对降低了76.96%和29.90%。
In view of the total transportation cycle time prediction of truck from loading point to unloading point in open-pit mine,a time prediction model of truck running state was constructed by support vector machine(SVM),Adaboost and random forest(RF),considering the influence of truck characteristics,road characteristics and weather characteristics.Taking the data collected from the truck dispatching system of a large open-pit mine as an example,the results show that for the time prediction of the running state,the prediction results of SVM,Adaboost and RF algorithms are all better than that of the traditional average method,and the average absolute error of the optimal machine learning method is reduced by 81.52%.At the same time,the combined prediction of taking the running state as the minimum unit of running time prediction is better than that of the single method to predict the whole transportation process time,and the average square error and average absolute error are reduced by 76.96% and 29.90%respectively.
作者
顾清华
马平平
闫宝霞
卢才武
GU Qinghua;MA Pingping;YAN Baoxia;LU Caiwu(School of Resources Engineering,Xi’an University of Architecture and Technology,Xi’an,Shaanxi 710055,China;China Nonferrous Metals Industry Xi’an Survey and Design Institute Gd.,Ltd,Xi’an,Shaanxi 710043,China)
出处
《矿业研究与开发》
CAS
北大核心
2020年第7期149-155,共7页
Mining Research and Development
基金
国家自然科学基金资助项目(51774228,51404182)
陕西省自然科学基金资助项目(2017JM5043)
陕西省教育厅专项科研计划项目(17JK0425)。
关键词
机器学习
露天矿
卡车运行状态
时间预测
Machine learning
Open-pit mine
Truck running state
Time prediction
