基于空间多点输入的LSSVM非高斯风压预测  

Non-Gaussian wind pressure prediction based on LSSVM with spatial multipoint input

作　　者：楼志挺 李春祥[1] LOU Zhiting;LI Chunxiang(Department of Civil Engineering,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学土木工程系

出　　处：《上海大学学报（自然科学版）》2019年第6期1013-1022,共10页Journal of Shanghai University:Natural Science Edition

基　　金：国家自然科学资助项目(51378304)

摘　　要：为提升最小二乘支持向量机对建筑表面风压的预测精度与泛化能力,提出了增加输入点数的方法.通过增加输入点数,增加提供给最小二乘支持向量机的信息量,优化最小二乘支持向量机预测模型性能;对不同时长的实测风压分别进行内插预测与外插预测.结果表明:采用内插预测和外插预测时增加输入点数均可提高风压预测性能,但是内插预测性能的提升效果优于外插预测;当增加训练集长度时,增加输入点数对预测性能的提升效果不明显.In order to enhance the generalization performance and prediction accuracy of Least Square Support Vector Machines for the wind pressure on the building surface, the method of increasing the number of input points has been proposed in this paper. Through increasing the number of input points, more information can be providedfor least squares support vector machines, which in turn optimizes the performance of the forecasting model of least squares support vector machines. Then interpolation prediction and extrapolation prediction of measured wind pressure at different lengths are carried out respectively. Experiment results show thatwhen interpolation prediction and extrapolation prediction are used, the performance of wind pressure prediction can be improved through the increase of the number of input points, but with interpolation prediction,the performance is better than that of extrapolation prediction. When the number of training sets increases, increasing the number of input points has little effect on prediction performance.

关 键 词：多点输入 最小二乘支持向量机 内插预测 外插预测 非高斯风压 

分 类 号：TU311[建筑科学—结构工程]