水库径流预报的蚁群优化神经网络算法应用研究

The Applied Research on Artificial Neural Networks Based on Ant Colony Optimization for Reservoir Runoff Forecast

水库优化供水调度效果如何很大程度上取决于对不确定因素的预测,针对水库径流预报问题,介绍一种基于蚁群优化的神经网络算法。该算法充分利用了神经网络的较强的记忆、联想能力和蚁群算法的正反馈性,同时在蚁群算法中增加扰动策略,克服了蚁群算法在求解过程中出现初期信息素匮乏、易陷入局部最优解的问题,并将其应用于滦河下游各水库的径流预报中。实例计算表明,本文建立的基于蚁群优化的神经网络模型是合理可靠的,训练精度较高,可对水库径流进行预测。

The results of reservoir water supply optimal dispatching depend to a great extent on the prediction of uncertain factors. Based on the problems of reservoir runoff forecast, an artificial neural network based on ant colony optimization is presented. This new hybrid algorithm that perturbation strategy is added in ant colony optimization can combine the strong memory and associative a- bility of artificial neural networks with positive feedback of ant colony optimization, and can overcome the shortcomings in the respect of lack of pheromone at the initial stage and easiness of local optimization of ant colony optimization. It is applied to six reservoirs op- timization water supply dispatching in the Luanhe River. A practical example calculation shows that this new hybrid algorithm is ra- tional, reliable and the high efficiency when it is used to solve the problem of reservoir runoff forecasts, at the same time, the training precision is high. So it can be used to forecast the reservoir runoff.