GPS HEIGHT CONVERSION BASED ON BAYESIAN REGULARIZATION BP ARTIFICIAL NEURAL NETWORK
Zhang Qiuzhao 1) ;Zhang Shubi 1,2) ;Liu Jun 1) ;Wang Guanghui 1) ; and Wang Bo 1)
1)School of Environment Science and Spatial Informatics of CUMT, Xuzhou 221008 2)Key Laboratory of Resources and Environmental Information Engineering of Jiangsu Province, Xuzhou 221008
Abstract In view of the defects of traditional BP artificial neural network, the Bayesian regularization method is used for improving the generalization ability of artificial neural network. Through an engineering example, the BP artificial neural network method with Bayesian regularization is compared with LM method and polynominal curve fitting method respectively. It is found that the improved method has better fitting precision, stability and better generalization ability,so it has good practicability in GPS height conversion.
Zhang Qiuzhao ,Zhang Shubi,Liu Jun et al. GPS HEIGHT CONVERSION BASED ON BAYESIAN REGULARIZATION BP ARTIFICIAL NEURAL NETWORK[J]. , 2009, 29(3): 84-87.
Zhang Qiuzhao ,Zhang Shubi,Liu Jun et al. GPS HEIGHT CONVERSION BASED ON BAYESIAN REGULARIZATION BP ARTIFICIAL NEURAL NETWORK[J]. jgg, 2009, 29(3): 84-87.
