遗传神经网络在径流预报中的应用研究

Study on application of genetic algorithm neural network for RUN-OFF forecasting

针对遗传算法的不足,在有效改进传统遗传算法的基础上,利用改进遗传算法对BP网络从拓扑结构到初始权阈值进行了全面优化。将改进遗传算法和BP网络相结合的技术应用于水科学领域,建立黑河水库枯水季入库径流量预报模型。拟合预测结果表明,将本文研究的方法用于水资源的实时预测是可行的、高效的。

For the weakness of genetic algorithms, this improved genetic algorithm optimizes BP network in terms of effective, which improves the speed and accuracy of forecasting. Improved genetic algorithm combined with BP neural network is proposed to establish the prediction model of dry season runoff into the Heihe reservoir in the field of science and technology in water. Fitting forecasting results show that the method for water resources real-time forecast is feasible and efficient.