露天矿计划阶段内离散块体物料运距预测算法

Haul distance prediction algorithm of discrete block in open-pit mine planning stage

矿岩运距是衡量露天矿山卡车运输经济性的重要指标之一,但多年因受运输系统复杂、多变等多种综合因素限制,在进度计划编制期间进行逐条带的运距计算问题一直并未得到有效解决。特别是需要进行离散化的物料规划时,传统逐阶段的运距量测方法无法对离散块体进行逐个推估。因此,为有效解决计划阶段内离散块体物料运距计算问题,根据露天矿外排土场出入口选址及排土运输干线相对固定等空间特征,将非线性预测理论、因子分析法等技术方法应用于实际的矿山优化问题中,研究了排土场计划阶段内块体运距波动变化的主要控制因素;结合外排土场内运距波动变化的影响要素,提出了采用加权最小二乘支持向量机技术(WLS-SVM)拟合非线性运距曲线的预测算法,并对权向量、核参数等主观经验指标进行了动态修正,以实现对计划阶段内(两阶段工程位置间的)离散块体物料过程运距的时变预测;最终,利用拟合出的时变曲线,实现两连续工程位置间逐个物料块体运距的高效计算。实验结果表明:预测算法的误差期望分别为:训练集0. 93 m,测试集0. 84 m,且在实验中表现出模型精度易受样本规模影响的特性,且当样本规模N>90时,可控制绝对误差水平稳定在0. 8%～1. 2%。结论表明:试图应用非线性预测理论处理计划阶段内缺乏时变路网条件的运距预测问题是可行的,提出的算法模型对解决计划阶段内的逐块体运距赋值问题具有现实有效性。常规的露天矿山运输优化问题其优化目标和特征参数之间往往存在紧密耦合关系,无法直接应用特定的优化、规划模型进行计算求解,尝试采用全新的解耦思想和求解思路对于有效解决露天矿山运输优化问题是十分有益的,也应该是今后露天矿山系统工程学科研究的热点问题之一。

The ore-rock transportation distance is one of the important indexes to measure the trucking economy in open-pit mine.Due to the complex and changeable transportation system,the calculation of the each strip’s transportation distance has not been effectively solved during the scheduling in the open-pit mines.In particular,when discrete material flow scheduling is needed,the traditional method of stage-by-stage transportation distance measurement cannot estimate discrete blocks one by one.Therefore,in order to effectively solve the problem of calculating the transportation distance of discrete blocks in the scheduling,the authors applies the nonlinear pre-diction theory,factor analysis method and other technical methods to the actual mine optimization problem ac-cording to the relatively fixed spatial characteristics,such as the location of the entrance and exit of the external mine dump site.The main controlling factors of the transportation fluctuation in the scheduling stage of the external dump are studied.Combined with the impact factors of the fluctuation of the transportation distance in the external dump,this paper presents an algorithm to fit the nonlinear haul distance curve using weighted least squares support vector machine(WLS-SVM).In order to improve the accuracy of the algorithm,the subjective experience indexes such as weight vector and nuclear parameter are modified dynamically to realize the fluctuation prediction of process distance.Finally,the efficient calculation of the distance between two consecutive engineering positions is realized.The experimental results show that the error of the prediction algorithm expectations are:training set is 0.93 m and test set is 0.84 m.The experiment shows the model precision is easily affected by the sample size.When the sample size N>90,the algorithm can control the absolute error level steady at between 0.8%and 1.2%.The study shows that the application of the theory of nonlinear prediction for lack of time varying network conditions to the stage’s distance prediction is feasible,and the proposed algorithm model is effective in solving the problem of the assignment of discrete block in the planning stage.In the conventional open-pit mine transportation optimization problem,its optimization goal tends to have tightly coupled relationships between characteristic parameters,and the authors cannot directly apply a certain optimization,programming model to calculate.For this kind of problems,the authors try to adopt new ideas of decoupling and solving ideas for effective optimization problem,which is good for transportation problem of open-pit mine,and it should also be a hot research topic in open-pit mine system engineering in the future.