摘要
文章针对单坡式防波堤越浪量的问题,在欧洲CLASH项目数据库的基础上,以影响单坡式防波堤越浪量的十个主要因素为输入参数,平均越浪量为输出结果,基于集成学习的平均法组合策略,建立了单坡式防波堤越浪量的综合预测模型。最后,利用数据库自身的实验记录和交通运输部天津水运工程科学研究所的实验结果进行验证分析。验证结果表明,人工智能的预测结果和实验结果具有很强的相关性,相关系数在0.85以上,误差范围在5倍之内。此外,防波堤结构的复杂性和训练集数据的可靠性,会影响人工智能的预测精确度。
In order to assess overtopping of simple armoured slopes for dikes and breakwaters,an artificial intelligence model for predicting overtopping is established based on the European CLASH project database.Ten parameters which affect the overtopping are chosen as the input values and the mean overtopping is chosen as the output value.Then,based on the average method combination strategy of integrated learning,a comprehensive prediction model is established.Finally,the experimental results carried out in Tianjin Research Institute of Water Transport Engineering are used for verification and analysis.The verification results show that there is a strong correlation between the prediction results and the experimental results,with the correlation coefficients 0.85 and the error range is within 5 times.In addition,the accuracy of prediction will be affected by the complexity of simple slopes structure and the reliability of training data.
作者
刘诗学
王收军
陈松贵
栾英妮
刘路
彭程
LIU Shi-xue;WANG Shou-jun;CHEN Song-gui;LUAN Ying-ni;LIU Lu;PENG Cheng(Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control,Tianjin University of Technology,Tianjin 300384,China;Tianjin Research Institute for Water Transport Engineering,National Engineering Laboratory for Port Hydraulic Construction Technology,Key Laboratory of Engineering Sediment,Ministry of Transport,Tianjin 300456,China)
出处
《水道港口》
2019年第5期541-546,587,共7页
Journal of Waterway and Harbor
基金
中国科协青年人才托举工程(2018QNRC001)
中央级公益性科研院所基本科研业务费专项资金项目(TKS160219
TKS160206
TKS180102)
国家自然科学基金(51809133)
关键词
单坡式防波堤
越浪量
人工智能
欧洲CLASH
simple armoured slope breakwater
overtopping
artificial intelligence
european CLASH
