基于KPCA-WOA-ELM的爆破飞石距离预测

为提高爆破飞石距离预测的精度和效率,构建了一种基于核主成分分析法（KPCA）和鲸鱼算法（WOA）优化的极限学习机（ELM）爆破飞石距离预测模型。以国内某露天煤矿爆破工程为例,选取影响爆破飞石距离的7个因素。通过KPCA对影响因素间非相关性关系进行降维,提取出包含原始信息95.76%的4个主成分作为模型输入。然后,采用WOA对ELM进制参数寻优,避免了局部最优解问题。结果表明,KPCA-WOA-ELM模型的平均相对误差、均方根误差R_MSE、决定系数R²和平均绝对误差R_MAE分别为4.271%、6.681、0.985和6.413,均优于对比模型。说明该模型可实现对爆破飞石距离的准确预测,为确定爆破作业中的爆破安全区提供依据。

Prediction of Blasting Flyrock Distance Based on KPCA-WOA-ELM

In order to improve the prediction accuracy and efficiency of blasting flyrock distance, a prediction model of blasting flyrock distance based on kernel principal component analysis (KPCA) and extreme learning machine (ELM) and optimized by a whale optimization algorithm (WOA) was established. Taking a blasting operations in open-pit coal mine as an example, seven influencing factors of blasting flyrock distance were selected. KPCA was used to reduce the dimension of the non-correlation relationship between the influencing factors, and four principal components containing 95.76% of the original information were extracted as the model input. Then, WOA was used to optimize the ELM system parameters to avoid the problem of local optimal solution. Results indicate that the average relative error, root mean square error R_MSE, coefficient of determination R²and average absolute error R_MAEof KPCA-WOA-ELM model are 4.271%, 6.681, 0.985 and 6.413, respectively, which are better than those of the comparison model. KPCA-WOA-ELM model can accurately predict blasting flyrock distance, and it could provide a basis for determining the blasting safety zone in blasting operation.