【作者】 杨金标；

【导师】 康玲；

【作者基本信息】 华中科技大学 ， 水文学及水资源， 2011， 硕士

【摘要】 我国长江流域洪涝灾害频繁,而中游地区是防洪重点。三峡水库对长江中游地区的防洪作用明显,但调度方式还是以调度图和调度规则为基础的常规调度,不能充分发挥其防洪作用。本文研究将粒子群优化算法应用到三峡水库对城陵矶站的防洪补偿调度优化模型中,以改进三峡水库的防洪效果。粒子群优化算法是一种简单实用的群智能全局优化算法。自该算法提出以来,在各种工程领域都得到了广泛的应用。在水库优化调度调度领域也有一定的应用研究。但基本粒子群算法也有易早熟、收敛速度慢等缺陷。本文提出一种自适应粒子群改进算法,对基本粒子群算法进行了相关改进,提高其跳出局部最优的能力,并将其应用于设计洪水过程线推求、马斯京根洪水演进参数优化模型的求解中。本文以1954年大洪水为资料,建立三峡对城陵矶防洪补偿调度优化模型,将水库调度约束条件和三峡水库特有的调度规则相融合,以城陵矶站分洪量最小为目标。利用自适应粒子群改进算法求解该优化模型。结果表明,城陵矶站的分洪量较小,优化效果较为明显,为水库防洪优化调度模型的求解提供了一种新方法。本文又针对实际调度中需要关注的预报问题,提出利用人工神经网络建立城陵矶站流量短期预报模型,并对其进行相关试验和结果分析,指出该模型的缺陷,并加以改进。更多还原

【Abstract】 Flood is frequent in Yangtze River basin, and the middle reach is the most important area. The Three Gorges Reservoir has an important effect in flood control of this area, but the operation mode still relies on regulation diagram and operation rules. This cannot maximize the effsct. This paper applies the Particle Swarm Optimization (PSO) to the Three Gorges Reservoir compensation operation to decreas flood divertion to the Chenglingji station.PSO is a new, simple and praetieal optimization algorithm based on swarm intelligence. Since proposed, PSO has been widely used in various engineering fields including reservoir operation optimization. However, the classical PSO still has its inherent flaws, like fall into local optimum and low computing speed. This paper proposed an Improved Adaptive Particle Swarm Optimization (IA-PSO) to improve its ability of jumping out from local optimum.Then use it to solve the design flood flow curve, flood routing Muskingum parameter optimization model.Using the flood in 1954 as an example, this paper establishs the Three Gorges Reservoir compensation operation on Chenglingji station model, which combined the constraint conditions of the Three Gorges Reservoir and its operation rules in target of minimum flood volume of Chenglingji station.The IA-PSO is used to solve the model. The results show that the flood volume diverted to Chenglingji station is decrease, and the effect of the optimization method is obvious.It provides a new way to solve the optimal operation models.This paper also proposes a short-term inflow forecasting model of Chenglingji station by artificial neural network method, which is important on the flood real-time operation.Then pointed out the shortcomings of the model and make some improvements of this model based on the tests and results.更多还原