摘要
趋势面从宏观上揭示了研究对象的特性,在各领域发挥着重要作用。BP神经网络可以对复杂系统进行无限逼近,进而进行预测。建立了基于贝叶斯正则化BP神经网络的数字高程模型趋势面,与二次多项式建立的数字高程模型趋势面进行比较分析,证明了该方法的可行性和有效性。
Trend can open out the characteristic of research object, and exert important effect in many domains. BP neutral network can approach complex system adinfinitum, and process forecast. In this paper, DEM trend based on Bayesian Regularization BP neutral network is constructed. Compared with DEM trend using quadratic polynomial , the feasibility and validity are proved.
出处
《海洋测绘》
2009年第4期32-34,41,共4页
Hydrographic Surveying and Charting
关键词
贝叶斯正则化
BP神经网络
数字高程模型
趋势面
Bayesian regularization
BP neutral network（BPNN）
digital elevation model（DEM）
trend
