基于IGWO-CatBoost模型的岩石爆破块度预测

Prediction of Rock Blasting Fragmentation Based on IGWO-CatBoost Model

针对无法准确预测矿山岩石爆破后块度大小的问题,提出一种基于改进灰狼算法(IGWO)优化的CatBoost块度预测模型。采用一种新的非线性收敛因子,引入动态权重策略,改进已有的灰狼算法(GWO),通过4个测试函数和5种优化算法验证了IGWO的寻优能力。对公开数据库和现场采集的32组数据进行预测分析。首先,采用随机森林算法进行特征重要性筛选,利用IGWO对CatBoost进行参数寻优,建立IGWO-CatBoost爆破块度预测模型;然后,将预测结果与在相同条件下建立的CatBoost、XGBoost、LightGBM模型进行对比分析。经过IGWO调参,CatBoost模型的预测准确度得到有效提高,IGWO-CatBoost模型的预测准确度均优于其他3种预测模型。对比结果表明,IGWO-CatBoost模型具有很好的预测能力和适应性。

Aiming at the problem of inaccurate prediction of block size after rock blasting in mines,a CatBoost block size prediction model based on improved grey wolf algorithm(IGWO)optimization was proposed.A new nonlinear convergence factor and a dynamic weight strategy were introduced to improve the existing grey wolf algorithm(GWO).The optimization ability of IGWO was verified by four test functions and five optimization algorithms.Thirty-two sets of data collected from the public database and the field were predicted and analyzed.Firstly,the random forest algorithm was used to screen the feature importance,and IGWO was used to optimize the parameters of CatBoost to establish the IGWO-CatBoost prediction model for blasting fragmentation.Then,the prediction results were compared and analyzed with the CatBoost,XGBoost and LightGBM models established under the same conditions.The prediction accuracy of the CatBoost model is effectively improved by IGWO,and the prediction accuracy of IGWO-CatBoost model is better than the other three prediction models.The comparison results show that IGWO-CatBoost model has good prediction ability and adaptability.