论文部分内容阅读
针对RBF网络序贯学习算法参数多、计算复杂等问题,深入分析RBF网络隐节点贡献度计算方法,提出基于主成分和周期性的贡献度计算方法,改进RBF网络GAP学习算法,细化算法中增加、删除和替换隐节点的条件,控制隐节点数量,自适应调整RBF网络结构.实验结果表明,相比传统RBF网学习算法,该算法在可靠性和泛化能力上都有显著提高.
Aiming at the problems of RBF network sequential learning algorithm such as many parameters and complicated computation, this paper deeply analyzes the calculation method of the contribution degree of RBF network hidden point, proposes the calculation method based on the principal component and periodic contribution degree, improves the GAP learning algorithm of RBF network, The conditions of adding, deleting and replacing hidden nodes are controlled, the number of hidden nodes is controlled, and the RBF network structure is adaptively adjusted. The experimental results show that the proposed algorithm can significantly improve the reliability and generalization ability compared with the traditional RBF network learning algorithm.