摘要
使用多项式和切比雪夫(Tchebyshev)多项式分别对沉降监测数据进行回归分析以预测未来沉降值,其中切比雪夫多项式的外推效果较好;应用前向BP神经网络对两种不同的单因子输入模式进行非线性函数逼近,并进行了不同采样步长的比较,实例表明将时间点作为网络的输入对沉降进行预测效果较好。
In order to predict future subsidence value, by employing polynomial and Tchebyshev polynomial expression to carry out regressive analysis towards subsidence monitoring data, the Tchebyshev polynomial can obtain preferable extrapolated result. Applying feed-forward back-propagation network to carry out approximation of nonlinear function towards two different single factorial input mode, and proceeding comparison of different sampling step, experiments prove that taking point in time as network input can obtain preferable result.
出处
《海洋测绘》
2007年第4期23-27,共5页
Hydrographic Surveying and Charting
基金
国家自然科学基金项目(40574002)
广西区自然科学基金项目(0339072)
关键词
沉降监测
多项式回归
切比雪夫多项式
神经网络预测
subsidence monitoring
polynomial regressivion
tchebyshev polynomial expression
neural network prediction
