混沌时间序列的混合粒子群优化预测

提出一种混合粒子群优化算法,即在改进粒子群优化算法全局搜索模型参数的基础上,利用梯度下降法进一步确定径向基神经网络模型参数,以提高网络的收敛精度和网络性能.采用基于RBFNN的混合粒子群优化算法进行离散Henon和连续Mackey-Glass混沌时间序列预测仿真,结果表明该算法能快速精确地预测混沌时间序列,是研究复杂非线性动力系统辨识和控制的一种有效方法.

Predicting chaotic time series using hybrid particle swarm optimization algorithm

A hybrid particle swarm optimization(HPSO) is proposed,in which the gradient descent is combined with modified particle swarm optimization(MPSO).The MPSO is determined by linearly decreasing inertia weight and constriction factor weight to speed up global search.Crossover and mutation operation is embedded to avoid the common defect of premature covergence.By using the proposed HPSO algorithm based on RBFNN,simulation for the chaotic time series prediction of discrete Henon and continuous Mackey-Glass chaotic time series is made to test the validity.Simulation results show that the HPSO can accurately predict chaotic time series and an effective approach is provided to study the properties of complex nonlinear dynamic system.