【摘要】 为了更好地预测爆破块度分布，采用人工旅鼠优化算法（artificial lemming algorithm, ALA）对支持向量回归（support vector regression, SVR）超参数进行优化，构建ALA-SVR模型。选取台阶高度、底盘抵抗线、填塞长度、排间距、孔间距、岩石坚硬程度为输入参数，使用R-R函数描述爆破块度分布，并将控制R-R函数的n和x₀作为输出参数。利用某石灰岩采石场的20次实际爆破工程数据对ALA-SVR进行训练和测试，结果表明：SVR经元启发算法优化后预测性能具有明显的提升，人工旅鼠算法优化后的SVR（ALA-SVR）其预测结果的平均相对误差E_MRE（mean relative error, MRE）、均方根误差E_RMSE（root mean square error, RMSE）和相关性系数（R²）分别为3.983 2%、0.837 4和99.942 6%,优于在相同条件下建立的鹰鱼优化算法（hawkfish optimization algorithm, HFOA）-SVR、雪橇犬优化算法（sled dog optimizer, SDO）-SVR预测模型，具有较高的预测精度和适用性。更多还原

【Abstract】 To enhance the prediction accuracy of blasting fragmentation distribution, the artificial lemming algorithm（ALA） is employed to optimize the hyperparameters of support vector regression（SVR）, thereby constructing the ALA-SVR model. Bench height, burden, stemming length, row spacing, hole spacing, and rock hardness are selected as input parameters. The R-R function is utilized to characterize the distribution of blasting fragmentation, with the parameters n and x₀ governing the R-R function designated as output parameters. Data collected from 20 actual blasting operations at a limestone quarry are used to train and test the ALA-SVR model. The results indicate that the predictive performance of the SVR model is significantly improved following optimization by metaheuristic algorithms. Specifically, the mean relative error（MRE）, root mean square error（RMSE）, and correlation coefficient（R²） of the ALA-SVR predictions are 3.9832%, 0.8374, and 99.9426%, respectively. These metrics outperform the HFOA-SVR and SDO-SVR predictive models established under identical conditions, demonstrating that the proposed ALA-SVR model possesses high prediction accuracy and broad applicability.更多还原