【摘要】 为解决岩爆数据集不均衡和模型参数寻优困难等问题，提出1种基于K-means SMOTE与改进蜣螂算法优化随机森林(random forest, RF)的预测模型。首先，分析岩爆发生机理构建指标体系；其次，使用K-means SMOTE算法对岩爆数据集进行均衡化处理，采用Robust标准化消除量纲；最后，引入Tent混沌映射和非线性递减策略组合改进蜣螂优化(improved dung beetle optimizer, IDBO)算法，寻优RF超参数，建立岩爆烈度等级预测模型(IDBO-RF)并与其他模型对比验证其有效性。研究结果表明：数据均衡处理后，各模型准确率提高10.85%～16.02%;设计的IDBO-RF预测模型平均准确率约为94.37%,较RF、GWO-RF、DBO-RF模型分别提高约7.76百分点、1.69百分点、1.11百分点；IDBO-RF预测模型准确率最高约为96.43%,优于RF、GWO-RF、DBO-RF模型。研究结果可为解决岩爆预测问题提供一定参考。更多还原

【Abstract】 To address the issues of imbalanced rockburst datasets and the challenges in optimizing model parameters, a predictive model based on K-means SMOTE and an improved dung beetle optimizer(IDBO) algorithm for optimizing random forest(RF) is proposed.Initially, the mechanism of rockburst occurrence is analyzed to construct an indicator system.Subsequently, the K-means SMOTE algorithm is employed to balance the rockburst dataset, and Robust Standardization is used to eliminate dimensionality.Finally, the Tent chaotic map and a nonlinear decreasing strategy are incorporated to improve the dung beetle optimizer algorithm for optimizing RF hyperparameters, resulting in the establishment of a rockburst intensity prediction model(IDBO-RF).The model’s effectiveness is verified through comparison with other models.The research findings indicate that, following data balancing, the accuracy of various models improves by 10.85% to 16.02%.The designed IDBO-RF prediction model achieves an average accuracy of approximately 94.37%,which is an improvement of about 7.76 percentage point, 1.69 percentage point, and 1.11 percentage point over the RF,GWO-RF,and DBO-RF models, respectively.The IDBO-RF prediction model attains the highest accuracy of approximately 96.43%,outperforming the RF,GWO-RF,and DBO-RF models.These results can provide reference for solving the problem of rockburst prediction.更多还原