混沌系统辨识的一种新的生长型神经气方法

IDENTIFICATION OF CHAOTIC SYSTEMS USING A NEW GROWING NEURAL-GAS NETWORK

在自组织神经网络基础上 ,根据生物群落自然增长的机制 ,提出了一种新的生长型神经气的自组织算法 ,用于混沌系统的自组织辨识 .该算法在学习样本的激励下能够动态地增加神经元 ,避免某些神经元可能出现的欠训练现象 ,从而极大地提高了网络整体训练的速度 .最后以 L

On the basis of self-organizing neural network, a novel growing neural-gas self-organizing algorithm to identify chaotic systems is proposed by imitating biological group population growing process. The algorithm can create new neurons according to the stimulation of learning samples. Thus, the problem of under-training for some neurons is well resolved, and the whole network training process is greatly improved. Simulations on Lorenz system as an example is performed.