基于遗传算法和BP神经网络的用水

为避免BP算法易陷入局部极小的缺陷，根据遗传算法具有全局寻优的特点，将二者结合起来形成GA-BP混合算法。以GA优化BP网络的初始权值和阈值，按负梯度方向修正网络权值及阈值，对网络进行训练。最后，用matlab编写GA-BP计算程序，以多组数据进行测试，并与纯BP算法进行分析比较，结果表明该方法可以有效、准确的应用于城市用水量预测。

Prediction of Water Demand with Genetic Algorithm and BP Neural Network

To avoid the BP algorithm's shortcoming of trapping to a local optimum and to take advantage of the genetic algorithm's globe optimal searching,a new kind of hybrid algorithm was formed based on GA and BP.First,the initialized weights and threshold of BP neural network(BPNN) was optimized with GA,and the network was trained by modifying the weights and threshold in the negative gradient direction.Finally,the program of GA-BP was compiled in Matlab language,and multi-group data were used to test it.Compared to BP algorithm,the results show that GA-BP is an effective and accurate method to predict city water demand.