Improved PLS and PSO methods-based back analysis for elastic modulus of dam. (May 2019)

• Record Type: Journal Article
• Authors: Yang, Lifu; Su, Huaizhi; Wen, Zhiping
• Is Part Of: Advances in engineering software. Volume 131(2019)
• Page Start: 205
• Page End: 216
• Publication Date: 2019-05
• Subjects: Dam -- Elastic modulus -- Optimized back analysis -- Partial least square -- Particle swarm optimization; Computer-aided engineering -- Periodicals; Engineering -- Computer programs -- Periodicals; Engineering -- Software -- Periodicals; Periodicals; 620.0028553
• DOI: 10.1016/j.advengsoft.2019.02.005
• Languages: English
• ISSNs: 0965-9978

Abstract

Highlights: Backward elimination-partial least square (BE-PLS) method was used to establish the statistical model of dam displacement. The relation between computations of the finite element model (FEM) and readings of the monitoring instrument was analyzed. FEM computations were consistent with the components extracted from monitoring data in objective function of back analysis. The rule of jumping out was used to prevent the particle swarm optimization algorithm from falling into the local solution. Abstract: The elastic modulus of dam body and dam foundation is the important mechanical parameter affecting the strength and stability of dam. According to dam monitoring data and numerical simulation, establishing the statistical model and the objective function, the improved particle swarm optimization (PSO) algorithm is applied to back analysis of elastic modulus of dam body and dam foundation. Firstly, based on backward elimination-partial least square (BE-PLS) method, the statistical model of dam deformation monitoring data is established, and the water load component is separated from monitoring data. Meanwhile, the finite element model (FEM) of the dam is built and the displacement field of the dam is calculated. Secondly, according to the difference between the FEM computations and the reading of the plumb-line, the relation between the two objects is deduced, and an accurate objective function is established. Thirdly, in the optimized back analysis, the PSO algorithm with modified parameters is used to search the optimal elastic modulus, and the rule of jumping out makes the algorithm avoid falling into the local solution. Finally, taking a gravity dam as an example, the results demonstrate that the statistical model established by BE-PLS has high fitting precision, and that the new PSO algorithm improves the global optimization ability and convergence speed and can avoid falling into the local solution. The optimized back analysis can obtain the elastic modulus of dam body and dam foundation with high accuracy.