摘要
本文研究了基于综合应用人工神经网络和演化算法的位移反演分析方法,并将此方法应用于三峡茅坪溪沥青混凝土心墙堆石坝的变形反演分析中。以茅坪溪一期工程原型观测成果为依据,反演出能够正确反映坝体变形特性的邓肯-张EB模型参数,从而预测了施工期末和蓄水期末该坝的变形特性,并基于总应力法研究了心墙水力劈裂破坏发生的可能性,为茅坪溪堆石坝运行期的安全生产提供依据。研究结果表明,茅坪溪心墙堆石坝不会发生水力劈裂破坏,该坝蓄水后虽然水压上升,但心墙仍是安全的。
Analyzing and forecasting the security of hydraulic project in future with the result of observing the prototype of the project at the beginning of construction is an important method to guarantee security in production and operation in hydraulic project. A new approach of displacement back analysis is studied by combining the neural network and evolutionary algorithm. The proposed approach is applied to the displacement back analysis of Maopingxi Rockfill Dam with asphalt concrete core in Three Gorges Project. Based on the field observation results of the stage-I of the dam, a group of DuncanChang EB modal parameters which can reflect the deformation characteristic of the dam is investigated by back analysis, and then, the deformation characteristic at the end of the construction phase and the reservoir filling phase is forecasted and the possibility of hydraulic fracturing of the core is discussed with total stress method, which offers the guarantee for the safety of Maopingxi Rockfill Dam in the reservoir filling phase. The results indicate that Maopingxi Rockfill Dam will never encounter hydraulic fracturing, as the hydraulic pressure rises after the dam stores water, the core is in safety.
出处
《中国安全生产科学技术》
CAS
2007年第3期12-16,共5页
Journal of Safety Science and Technology
基金
国家科技支撑计划(编号:2006BAK04B01)资助
关键词
人工神经网络
演化算法
位移反演分析
水力劈裂
artificial neural network
evolutionary algorithm
displacement back analysis
hydraulic fracturing