论文部分内容阅读
为克服传统BP神经网络在渗流压力预测过程中收敛慢、计算量大和易陷入局部极小等缺陷,依据渗流压力的影响因素,研究了模型的结构和输入输出因子,建立了基于遗传算法和LM算法相结合的GA-LMBP神经网络的大坝渗流压力预测模型,即通过遗传算法(GA)的选择、交叉和变异操作得到BP网络的一组全局最优近似解(即网络的初始权值和阈值),再以该近似解为初值,利用LM算法对BP网络进行优化训练,将训练好的网络用于渗流压力的预测。实例应用结果表明,在相同精度的要求下,GA-LMBP神经网络模型收敛速度快、预测精度高,对大坝渗流压力的预测效果更佳,是值得采用的一种模型。
In order to overcome the shortcomings of traditional BP neural network, such as slow convergence, large amount of calculation and easy to fall into local minimum in the process of seepage pressure prediction, the structure and input and output factors of the model are studied based on the factors of seepage pressure, Algorithm is used to predict the seepage pressure of dam based on the GA-LMBP neural network. That is to say, a set of global optimal solution (ie, the initial weight and the sum of the network) of the BP network is obtained through the selection, crossover and mutation operation of the genetic algorithm Threshold), and then use the approximate solution as initial value, optimize the BP network by using LM algorithm, and use the trained network for the prediction of seepage pressure. The practical application shows that the GA-LMBP neural network model has the advantages of fast convergence, high prediction accuracy and better prediction of dam seepage pressure under the same precision. It is worth to adopt a model.