基于混合优化算法的砂砾料面板堆石坝压实质量评价模型及工程应用研究  

Study on the compaction quality evaluation model of gravel face rockfill dam based on Hybrid Optimization Algorithm and its engineering application

作　　者：刘世伟 杨宇 宿辉 孙熇远 赵宇飞[3] 杨涛 LIU Shiwei;YANG Yu;SU Hui;SUN Heyuan;ZHAO Yufei;YANG Tao(Hebei Key Laboratory of Intelligent Water Conservancy,Handan 056006,Hebei,China;School of Water Conservancy and Hydropower,Hebei University of Engineering,Handan 056006,Hebei,China;China institute of Water Resources and Hydropower Research,Beijing 100044,China)

机构地区：[1]河北省智慧水利重点实验室,河北邯郸056006 [2]河北工程大学水利水电学院,河北邯郸056006 [3]中国水利水电科学研究院,北京100044

出　　处：《水利水电技术（中英文）》2023年第7期141-149,共9页Water Resources and Hydropower Engineering

基　　金：河北省重大科技专项(E2020402087);河北省高等学校科学技术研究项目(QN2021030);国家自然科学基金项目(51779250)。

摘　　要：良好的压实质量评价模型是砂砾料面板堆石坝压实质量实时有效控制的关键。【目的】为改善现有模型的预测精度和泛化能力,【方法】依据新疆阿尔塔什混凝土面板堆石坝工程实例,采用主成分分析法优化数据样本空间,提出自适应差分进化算法与极限学习机相结合的混合优化算法,构建了基于混合优化算法的砂砾料面板堆石坝压实质量评价模型,并与现场实测结果以及其他模型预测结果进行对比分析。【结果】结果显示:该评价模型的预测结果与工程实际值的平均绝对误差MAE为0.00708,均方误差MSE为0.0000923,采用原始数据预测的MAE和MSE分别为0.0106、0.000223;与ELM、BP、RBF等模型对比显示,该评价模型预测结果与实测结果的皮尔逊相关系数为0.824,平均绝对百分比误差MAPE为0.62%,ELM、BP、RBF模型预测结果与实测结果的皮尔逊相关系数分别为0.447、0.43、0.556,MAPE分别为1.18%、1.59%、1.01%。【结论】结果表明:碾压参数、料源参数和气象参数是影响坝体压实质量的关键控制影响因子;通过主成分分析,降低了样本空间维度,提升了模型训练效率;与实测结果相比,该评价模型预测结果最优,相比于其他三类模型而言,误差减小了1倍;不同样本空间范围的预测结果曲线更加平滑,表明该评价模型预测结果更加稳定。相关研究成果可为面板堆石坝施工质量实时管控提供理论依据。A good compaction quality evaluation model is the key to real-time and effective control of compaction quality of sand-gravel face rockfill dam.[Objective]In order to improve the prediction accuracy and generalization ability of existing models.[Methods]According to the project example of Xinjiang Altash Concrete Face Rockfill Dam,the principal component analysis method is adopted to optimize the data sample space,and a hybrid optimization algorithm combining adaptive differential evolution algorithm and extreme learning machine is proposed.A compaction quality evaluation model of sand-gravel face rockfill dam based on hybrid optimization algorithm is constructed,and comparison and analysis with the field measurement result and other model prediction result are carried out.[Results]The average absolute error MAE between the prediction result of the model and the actual value of the project is 0.00708,and the mean square error MSE is 0.0000923.The MAE and MSE predicted by the original data are 0.0106 and 0.000223,respectively.The comparison between the model in this paper and the ELM,BP,RBF and other models shows that the Pearson correlation coefficient between the predicted result of the model and the measured result is 0.824,and the mean absolute percentage error MAPE is 0.62%.The Pearson correlation coefficients between the predicted result of the ELM,BP and RBF models and the measured result are 0.447,0.43 and 0.556,respectively,and the MAPE is 1.18%,1.59%and 1.01%,respectively.[Conclusion]According to the research result,the rolling parameters,material source parameters,as well as meteorological parameters are the key c ontrol factors influencing the dam compaction quality.Through principal component analysis,the space dimension of the sample is reduced and the model training efficiency is improved;Compared with the result measured,the prediction result of the model in this paper is the best,and the error is reduced by one time compared with the other three types of models;The prediction result curves of diff

关 键 词：面板堆石坝 砂砾料 压实质量 主成分分析 混合优化 

分 类 号：TV5[水利工程—水利水电工程]