一种混沌对角递归神经网络混合学习算法研究

Hybrid Algorithm for Identification of Chaotic Diagonal Recurrent Neural Networks

针对一种混沌对角递归神经网络辨识,提出了一种混合学习算法。首先,采用遗传算法来获得混沌对角递归神经网络的拓扑结构和连接权值的全局次优解。之后,用混沌BP算法对网络的连接权值进行精调。最后,将这种混合优化算法应用到非线性时间序列的建模中。仿真结果表明了模型和算法的有效性。

A hybrid algorithm for identification of chaotic diagonal recurrent neural networks is proposed in this article.First of all,genetic algorithm is adopted to optimize the topology structure of neural networks and global suboptimum weight distribution.Then the connect weights of neural networks are accurate adjusted using chaotic BP method.The two methods are combined together to modeling a nonlinear system based on chaotic diagnonal recurrent neural networks.Simulation results indicate that the presented approach is effective and feasible.