【摘要】 在混沌算法神经网络的预测模型中 ,适当选择非线性反馈项 ,能使网络的动力学在权空间具有混沌行为 ,网络系统在学习和训练过程中能够跳出能量的局域极小达到全局极小或其近似 .本文基于EP进化算法建立一种自适应机制 ,使得网络能够根据学习和训练的结果优化非线性反馈项 .应用这种算法的神经网络对基于 Mackey-Glass方程和 Lorenz系统的时间序列进行在线预测 ,结果表明 ,网络具有很好的自适应预测性能更多还原

【Abstract】 For a prodiction model with chaotic algorithm, the network system can escape from the local minima and converge to the global minimum or its approximate solution during the learning and prediction process by selecting the suitable nonlinnear feedback term. The dynamics of network become chaotic one in the weight space. In this work, we use the EP evolutionary computation to establish a kind of self\|suitable prediction model. The model is tested for the time series which generated with Mackey\|Glass equation and Lorentz system by on\|line method. The simulation results indicated that the network has a good self\|suitable prediction performance.更多还原