| 基于分级偏最小二乘回归的径向基函数神经网络 |
| |
| 引用本文： | 尹建川,东昉,侯建军,胡江强.基于分级偏最小二乘回归的径向基函数神经网络[J].大连海事大学学报(自然科学版),2007,33(3):113-116,121. |
| |
| 作者姓名： | 尹建川 东昉 侯建军 胡江强 |
| |
| 作者单位： | 大连海事大学航海学院 辽宁大连116026(尹建川，东昉，胡江强)，海军大连舰艇学院航海系 辽宁大连116018(侯建军) |
| |
| 摘 要： | 为提高径向基函数神经网络的泛化性能,提出一种利用分级偏最小二乘回归方法构造径向基函数神经网络的方法,逐步增加网络中的隐节点数直至达到合适的网络规模,消除了训练数据中存在的多重共线性对网络泛化能力的不利影响.所得径向基函数神经网络的泛化能力比偏最小二乘回归构造的径向基函数神经网络提高了约30%.船舶航向跟踪预测控制仿真验证了该算法的可行性和有效性.
|
| 关 键 词： | 径向基函数神经网络 偏最小二乘回归 泛化性能 预测控制 |
| 文章编号： | 1006-7736（2007）03-0113-04 |
| 修稿时间： | 2007-03-07 |
Two-stage partial least squares regression for constructing radial basis function networks |
| |
| YIN Jian-chuan,DONG Fang,HOU Jian-jun,HU Jiang-qiang.Two-stage partial least squares regression for constructing radial basis function networks[J].Journal of Dalian Maritime University,2007,33(3):113-116,121. |
| |
| Authors: | YIN Jian-chuan DONG Fang HOU Jian-jun HU Jiang-qiang |
| |
| Institution: | 1. College of Navigation, Dalian Maritime University, Dalian 116026, China; 2. Department of Navigation, Dalian Naval Academy, Dalian 116018, China |
| |
| Abstract: | To improve the generalization capability of radial basis function(RBF) networks,a two-stage partial least squares(TPLS) learning algorithm was developed.The algorithm grew RBF centers one by one with partial least squares(PLS) regression method,and the resulting parsimonious RBF-TPLS network improved the generalization capability by 30% than that of PLS algorithm network.The RBF-TPLS network was used to approximate and replace the cost function minimization in predictive control.The parsimonious structure and good generalization capability of RBF-TPLS network ensure the online application of the predictive control.Simulation results of predictive ship course-changing control demonstrate the feasibility and effectiveness of the proposed learning algorithm. |
| |
| Keywords: | radial basis function network partial least squares regression generalization capability predictive control |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
