摘要
文章提出了一种神经网络辨识的混合学习算法。采用具有递阶结构的遗传算法来获得神经网络拓扑结构和连接权值的全局次优解,之后由BP算法来进一步调整神经网络的连接权值,从而实现神经网络的自动优化设计。仿真结果表明,所得的神经网络结构简单、精度高,并具有良好的泛化能力。
A novel hybrid algorithm for identification of a neural network is proposed.The hierarchical genetic algorithm is used to optimize the structure and weights of the neural network,so the suboptimal solution is obtained.After that,the BP algorithms is used to tune the weights,consequently,the automatic optimization design of the neural network is real-ized.The simulation results show that the neural network trained by the proposed approach possesses the characteristics of simple structure,high precision and good generalization ability.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第28期33-36,共4页
Computer Engineering and Applications
基金
江苏省教育厅自然基金项目(编号:03KJB510041)
关键词
递阶遗传算法
BP算法
神经网络
结构辨识
参数辨识
hierarchical genetic algorithms ,BP algorithms ,neural networks,structure identification,parameter identification