年最大洪峰流量群居蜘蛛优化投影寻踪预测模型

发布时间：2018-03-15 20:49

本文选题：SSO算法　切入点：参数投影寻踪　出处：《南水北调与水利科技》2017年02期 　论文类型：期刊论文

【摘要】：年最大洪峰流量预测,受较多的复杂因素的影响,不确定性较强,用常规统计方法做出准确预报具有较大困难。从水文序列本身出发,提出将投影回归模型应用于年最大洪峰流量预测,为了更好获得投影寻踪模型参数和预测精度,提出了运用延迟相关系数法确定回归预测因子、群居蜘蛛算法优化投影寻踪模型最佳投影方向参数a、利用最小二乘法确定多项式的权系数c、岭函数个数M的群居蜘蛛优化投影寻踪年最大洪峰流量预测模型,结合长江宜昌站(1882年-2004年)的年最大洪峰流量资料进行实例预测,训练阶段平均绝对相对误差为8.61%,预测阶段平均绝对相对误差为10.51%,该模型预测效果较好,模型结果稳定,可有效应用于年最大洪峰流量预测。
[Abstract]:The forecasting of the maximum Hong Feng discharge in 2000 is influenced by more complicated factors, and the uncertainty is strong. It is difficult to make accurate prediction by conventional statistical methods. From the hydrological series itself, it is difficult to make accurate prediction. The projection regression model is applied to the annual maximum Hong Feng flow prediction. In order to obtain the parameters and prediction accuracy of the projection pursuit model, the delayed correlation coefficient method is used to determine the regression prediction factors. The optimum projection direction parameter a of colony spider algorithm is optimized, the weight coefficient of polynomial is determined by least square method, and the forecasting model of maximum Hong Feng flux in projection pursuit year of colony spider with the number of ridge function M is determined. Combined with the annual maximum Hong Feng discharge data of Yichang Station of the Yangtze River from 1882 to 2004, the average absolute relative error in training stage is 8.61 and the average absolute relative error in prediction stage is 10.51. The model has good prediction effect and stable results. It can be used to forecast the maximum Hong Feng flow in the year.
【作者单位】：华北水利水电大学水利学院;
【基金】：国家自然科学基金项目(51509088) 水利部公益性行业科研专项(201501008) 水资源高效利用与保障工程河南省协同创新中心(2013CICWP-HN) 河南省高校科技创新团队(14IRTSTHN028)~~
【分类号】：TV122.5

【相似文献】