Abstract: Phase space reconstruction and least square support vector machine parameters optimization are optimized separately in traditional methods，and the prediction model can’t achieve the best performance.A parameters simultaneous optimization method is proposed，which uses the interdependent relationship between phase space reconstruction and least square support vector machine parameters to improve the model performance.Firstly，the parameters are designed based on uniform design.Secondly，parameters are optimized based on least square support vector machine and the best parameters are obtained.The experiment results on sunspots time series show that the proposed algorithm’s prediction precision is higher and computational complexity is lower compared with the traditional parameters optimization method.It provides a new way for chaotic time series prediction model parameters optimization.

Key words: least square support vector machine, time series prediction, uniform design, optimization

摘要： 传统上相空间重构与预测模型参数优化分开优化，割裂两者的联系，模型预测性能难以达到最优。利用相空间重构和预测模型参数的互相关系，提出一种混沌时间序列预测模型参数同步优化方法。首先采用均匀设计方法对影响模型预测精度的参数进行均匀设计，然后采用自调用最小二乘支持向量机进行参数同步优化，得到最优参数。以经典混沌时间序列太阳黑子年平均数为例进行了验证，结果表明，相对传统的参数优化算法，参数同步优化算法时间复杂度低、预测精度高，为混沌时间序列预测模型参数优化提供了一种新的思路。

关键词: 最小二乘支持向量机, 时间序列预测, 均匀设计, 优化