| 摘要: |
| 【目的】针对传统多沙水库冲淤预测模型难以准确、迅速预测某一具体水库调度运行方式下泥沙的冲淤变化过程，无法为制定合理水库调度运行方式提供决策依据的不足，构建一种计算效率高、能保证一定计算精度且相对简便的泥沙冲淤预测模型。【方法】将人工神经网络及粒子群优化算法引入到多沙水库冲淤预测中，构建基于自适应粒子群算法优化BP神经网络的多沙水库冲淤预测模型，并将该模型应用于冯家山水库库区泥沙冲淤形态、冲淤量的预测，验证其实用性。【结果】将多沙水库冲淤变化过程视为一个非线性动力系统，利用人工神经网络处理大规模复杂非线性动力学问题的优势，在采用自适应粒子群优化算法对BP神经网络的初始连接权值和阈值进行优化的基础上，成功构建了基于自适应粒子群算法优化BP神经网络的多沙水库冲淤预测模型。该模型在冯家山水库冲淤预测中的应用结果表明，模型计算值与实测值之间吻合良好，可满足实际水库管理的需要。【结论】所构建模型具有较强的合理性及较广的适用性，为多沙水库冲淤预测提供了一条有效途径。 |
| 关键词: 多沙水库 冲淤预测模型 BP人工神经网络 自适应粒子群优化算法 |
| DOI： |
| 分类号: |
| 基金项目:国家科技重大水专项（2009ZX07212-002-001-04，2008ZX07106-4-01） |
|
| Research and application on the deposition-scouring prediction model based on the neural network optimized by the adaptive particle swarm optimization algorithm in the heavily sediment-laden reservoir |
|
WU Wei, ZHOU Xiaode, WANG Xinhong
|
|
Key Lab of Northwest Water Resource and Environment Ecology of MOE,Xi'an University of Technology
|
| Abstract: |
| 【Objective】The traditional deposition-scouring prediction model of the heavily sediment-laden reservoir was difficult to predict accurately and quickly the process of scour and deposition in a specific reservoir operational mode.And the traditional mode was unable to provide decision basis to develop a reasonable reservoir operational mode.For above reason,it was necessary to build an efficient, high precision and relatively simple deposition-scouring prediction model.【Method】The artificial neural network and particle swarm optimization algorithm were brought into the deposition-scouring predication of the heavily sediment-laden reservoir.Based on the BP artificial neural network optimized by the adaptive particle swarm optimization algorithm,the deposition-scouring predication model was built in the heavily sediment-laden reservoir.Then the model was applied to predict the deposition-scouring amount and pattern of the Fengjiashan reservoir.【Result】The deposition-scouring process of heavily sediment-laden reservoir was regarded as a nonlinear dynamic system.Using the artificial neural network advantages in aspect of handling large and complex nonlinear dynamics problem,the deposition-scouring predication model of heavily sediment-laden reservoir was successfully built.Meanwhile,the adaptive particle swarm optimization algorithm was utilized to optimize the initial weight and bias values of BP artificial neural network to improve performance.The predication results of Fengjiashan reservoir showed that the calculated and measured values were in good agreement,so the model could meet the practical needs of reservoir management.【Conclusion】The model had strong rationality and wide applicability.An effective way was provided to predict the deposition-scouring process of heavily sediment-laden reservoir. |
| Key words: heavily sediment-laden reservoir deposition-scouring prediction model BP artificial neural network adaptive particle swarm optimization algorithm |
