BACK ANALYSIS OF MECHANICAL PARAMETERS OF CONCRETE FACE ROCK-FILL DAMS BASED ON A MODIFIED PARTICLE SWARM OPTIMIZATION

【摘要】：Earth-rock Dams safety evaluation is significant to maintain the normal operation of dams, where back analysis based on the observed data of prototype plays an important role in guaranteeing the precision of evaluation. Taking example for a Concrete Face Rock-fill Dam(CFRD) in a real project, this paper attempts to introduce a novel intelligence optimization algorithm, one-population self-adaptive Particle Swarm Optimization(PSO), to back analyze four considerably sensitive mechanical parameters of Duncan-Chang EB model(K, n, Kb, m) on the basis of measured displacements. PSO has been widely utilized in various fields due to its essential advantages like simple conception, easy implementation, fast calculation etc., while it is seldom applied to back analysis of mechanical parameters of dams. The modified PSO, One-Population Self-Adaptive PSO, further improved the search speed and enhanced the ability of guaranteeing the convergence to the global optimization solution, which benefits the back analysis calculation more in the calculation affectivity and efficiency compared with the standard PSO. Most of the former back analyses of earth-rock dams were according to the displacements of different measuring points at a certain time or the displacements at one measuring point along the time-history, thus the back analyzed parameters can't reflect material properties very well. In this paper, back analysis is based on the displacements at multiple measuring points along time-history during construction and operation period. Through analyzing the rationality of parameter results, it is possible to provide a reliable reference for the safety evaluation of dams as well as design and construction of dams to be constructed. This optimization algorithm was also proved to be effective in the back analysis of the mechanical parameters of CFRD based on displacements.