摘要
针对GA与BP网络优缺点互补的特点,摒弃以往GANN模型的不足,采用以GA优化BP网络权值与阈值,再用BP网络对GA搜索到的近似最优解进行微调的方法,建立了一种能充分发挥两者优势的混合GANN模型。将模型应用于大气折光实时改正,建立了基于混合GANN的大气折光实时改正模型。通过与其他模型的分析对比表明,该模型科学、精确、易于实现,是一种实用的大气折光实时改正模型。
A new GANN model is presented based on the complementation of advantages of GA NN.With this model the disadvantages of GANN are eliminated,at the same time the advantages of GA and then the precision is improved with BP.The model is applied to the real-time correction of atmospheric refraction coefficient and a mixed GANN real-time correction model is established.Compared with others,the results indicate that the model is scientific,accurate,practical and useful.
出处
《长江大学学报（自然科学版）》
CAS
2005年第10期382-384,共3页
Journal of Yangtze University(Natural Science Edition)
关键词
全站仪
大气折光系数
神经网络
遗传算法
实时改正模型
total station
atmospheric refraction coefficient
neural network(NN)
genetic algorithm(GA)
real-time correction model
