基于自适应递阶遗传算法的洪水预报模型

Research on flood forecasting model based on adaptive hierarchical GA-neural Network

基于递阶结构的遗传算法可以同时对BP神经网络进行结构优化和权重求解。采用收缩解空间的方法可以有效提高算法的收敛速度和精度。采用先进行自适应递阶遗传算法获得最优网络结构及参数,再用BP算法对网络进行训练,并应用于洪水预报模型,可以获得较高的精度。

The genetic algorithm based on hierarchical structure （HGA） can optimize the structure and weights of neural network simuhaneously. The method of shorten space can improve the speed of convergence and calibration. We adopt AHGA to get the best structure and parameter of Network and train the network with BP algorithm. The research applied on flood forecasting model can achieve higher Model precision.