摘要
为了提高爆破振动预测精度,提出了一种鲸鱼优化算法(whale optimization algorithm,WOA)和信息准则(information criterion,IC)优化的人工神经网络(artificial neural network,ANN)爆破振动预测模型。根据二维指标变量法将地质参数定量化,建立了包括3个定量参数和10个定性参数的更完整的数据集。利用信息准则对模型复杂度的反馈,构建了一个提高模型泛化能力的双层优化结构,分析改进ANN模型的激活函数和训练算法最优组合,并引入鲸鱼算法优化模型初始权值和阈值的选取,降低模型输出结果的偏差和波动。对比分析WOA-IC-ANN模型与传统经验公式、ANN模型、IC-ANN模型、WOA-ANN模型预测结果的差异。研究表明,WOA-IC-ANN模型的预测结果与实际吻合更好,误差显著降低,具有较好的泛化能力。研究成果可用于隧道爆破工程的振动预测,并为类似工作提供借鉴和参考。
In order to improve the accuracy of blasting vibration prediction,a whale algorithm(WOA)and information criterion(IC)optimized artificial neural network(ANN)blasting vibration prediction model was proposed.Geological parameters were quantified according to a two⁃dimensional index variable method,and a more complete data set including 3 quantitative parameters and 10 qualitative parameters was constructed.Using the information criterion to feedback the complexity of the model,a two⁃layer optimization structure was established to improve the generalization ability of the model.The optimal combination of activation function and training algorithm for the improved ANN model was analyzed,and the whale optimization algorithm was introduced to optimize the selection of initial weights and thresholds of the model,the deviation and fluctuation of the model output results were reduced.Differences between the WOA⁃IC⁃ANN model and traditional empirical formulas,ANN model,IC⁃ANN model,and WOA⁃ANN model were analyzed and compared.It was shown that the prediction results of the WOA⁃IC⁃ANN model are in better agreement with reality,with significantly reduced errors and good generalization ability;the research results can be used for vibration prediction in tunnel blasting engineering,and provide reference and guidance for similar work.
作者
高宇璠
傅洪贤
GAO Yufan;FU Hongxian(School of Civil Engineering,University of South China,Hengyang 421001,China;School of Civil and Architectural Engineering,Beijing Jiaotong University,Beijing 100044,China)
出处
《振动与冲击》
北大核心
2025年第4期229-237,共9页
Journal of Vibration and Shock
基金
轨道交通安全协同创新中心项目(2022JBXT007)
国家自然科学基金联合基金项目(U1934210)。
作者简介
第一作者:高宇璠,男,博士,讲师,1993年生;通信作者:傅洪贤,男,博士,教授,1966年生,E-mail:fhxllj@163.com。
