摘要
在模糊优选BP神经网络模型的基础上,引入遗传算法,提出融入遗传算法的模糊优选神经网络预测模型,是对模糊优选BP神经网络模型的进一步发展。其基本思路是:在BP算法训练网络出现收敛速度缓慢时启用遗传算法优化网络的运行参数,把优化的结果作为BP算法的初始值再用BP算法训练网络,这样交替运行BP算法和遗传算法,直到达到问题要求的精度。在新疆雅马渡站年径流量的预报中,预测模型在预报精度和算法的收敛速度方面都达到了较好的效果。
Based on the BP neural network model of fuzzy optimization,the authors introduced a genetic algorithm and proposed an intelligent forecast model,which combined the fuzzy optimization with the genetic algorithm.The model was a funther development of the BP neural network model of fuzzy optimization.The steps were as follows:when the speed of BP algorithms became slow,the genetic algorithms were to be used to optimize the weights of neural networks instead of the BP algorithms;the weights were to be regarded as the initial values for next step of BP algorithms.An example proved that the intelligent forecast model would significantly promote the forecast precision and accelerate the calculation convergence speed.
出处
《水利学报》
EI
CSCD
北大核心
2003年第5期116-121,共6页
Journal of Hydraulic Engineering
关键词
模糊集
神经网络
遗传算法
径流预测
fuzzy sets
neural network
genetic algorithm
runoff forecast
