论文部分内容阅读
小波神经网络(WNN)的理论基础是小波函数的重构理论,只含有一个隐含层的3层小波神经网络,该网络可以通过多种形式来解决一个接近非线性映射问题。常用的遗传算法(GA)是在达尔文生物进化论的基础上逐渐形成的一种计算机模型,可以通过模拟自然进化来选择最佳的问题解决方法,基于遗传算法优化的小波神经网络包含了神经网络理论中最精华的部分,由于其结构简单、可塑性强,可以使权值收敛到某个值后,误差为全局最小值。通过建立金华市区红层地下水的GA-WNN模型,实现了对其进行科学的预测,为以后地下水预警体系建设打下了坚实的基础。
The theoretical basis of wavelet neural network (WNN) is the reconstruction theory of wavelet function, which only contains a hidden layer 3-layer wavelet neural network. The network can solve a near-nonlinear mapping problem in many forms. The commonly used genetic algorithm (GA) is a computer model gradually formed on the basis of Darwin’s biological evolution theory, which can select the best problem solving method by simulating natural evolution. The neural network based on genetic algorithm optimization includes neural network theory In the most essential part, due to its simple structure and strong plasticity, the error can be the global minimum after the weight converges to a certain value. Through the establishment of GA-WNN model of groundwater in the red ground of Jinhua city, it has realized its scientific prediction and laid a solid foundation for the future construction of groundwater warning system.