基于Chebyshev神经网络的非线性动态系统预测

发布时间：2018-01-15 06:32

本文关键词：基于Chebyshev神经网络的非线性动态系统预测　出处：《安徽大学学报(自然科学版)》2016年06期 　论文类型：期刊论文

【摘要】：针对非线性动态系统的预测常受到噪声或其他过程的耦合影响,使得规律变得难以发现的问题,提出了以一组Chebyshev正交基函数作为神经网络中各隐神经元的激励函数的新型的Chebyshev基函数神经网络预测模型.将该模型作为非线性动态系统预测模型,并采用基于粒子群和模拟退火组成的文化基因算法优化神经网络的权值,可以达到很高的预测精度和很好的预测结果.Chebyshev神经网络与传统的BP(back propagation)神经网络相比,工作量大大减少,加快了收敛性.文化基因算法用于确定权值的Chebyshev神经网络分别与粒子群和模拟退火优化的Chebyshev神经网络相比具有更好的拟合效果.
[Abstract]:The prediction of nonlinear dynamic systems is often affected by noise or other processes, which makes the rules difficult to find. A new prediction model of Chebyshev basis function neural network using a set of Chebyshev orthogonal basis functions as the excitation function of each hidden neuron in the neural network is proposed. The model is regarded as a nonline model. Dynamic system prediction model. A cultural genetic algorithm based on particle swarm optimization and simulated annealing is used to optimize the weights of neural networks. Chebyshev neural network can achieve high prediction accuracy and good prediction results. Chebyshev neural network is compared with the traditional BP(back propagation neural network. The workload has been significantly reduced. The Chebyshev neural network used to determine the weight of genetic algorithm has better fitting effect than particle swarm optimization and simulated annealing optimization Chebyshev neural network, respectively. Fruit
【作者单位】：新疆大学网络与信息技术中心;
【基金】：国家自然科学基金资助项目(51575469)
【分类号】：TP183
【正文快照】： 在实际的工程领域和科学研究中,许多系统都可以看做是一类存在较高不完整性和随机性的非线性动态时变系统[1],很难利用确定的机理模型进行描述与预测.由于神经网络具有高度的非线性映射特性和高度的并行计算能力,近年来被广泛用来解决实际的预测问题,在很多领域都取得了成功

【相似文献】