基于MFOA和LW的混沌时间序列鲁棒模糊预测

Robust fuzzy prediction of the chaotic time series based on the MFOA and LW

针对含有例外点的混沌时间序列的预测问题,提出了一种基于修正型果蝇优化算法(MFOA)和最小Wilcoxon方法(LW)的混合学习算法来训练T-S模糊模型,以达到准确建模和提高模型鲁棒性的目的。首先采用修正型果蝇优化算法优化模糊前件的高斯型隶属函数参数,利用其编程简单、收敛速度快的优点提高辨识精度和收敛速度。然后采用最小Wilcoxon方法辨识模型的结论参数,在训练数据中出现例外点时,LW方法的强鲁棒性可以有效克服传统最小二乘方法对例外点敏感的缺点。最后以Mackey-Glass混沌时间序列的预测为例进行仿真研究,通过比较不同的优化算法的辨识结果来验证修正型果蝇优化算法的优越性,并在系统存在例外点的情况下验证了所提方法的有效性和鲁棒性。

Focusing on the prediction of the chaotic time series containing outliers , a hybrid learning method based on the modified fruit fly optimization algorithm （ MFOA） and the least Wilcoxon （ LW） method is proposed in order to train the T-S fuzzy model .The purpose of this is to improve the accuracy and robustness of fuzzy modeling for nonlinear systems .Firstly, the MFOA is used to optimize the antecedent parameters of the Gaussian membership function with the advantages of ease of transformation of such a concept into program code and a high convergence speed , which can improve the identification accuracy and convergence speed in fuzzy modeling .Secondly , the least Wilcoxon method is applied to identify the consequential parameters of the model .When the outliers occur in the training data , the strong robustness of the LW with the outliers is effective for improving the sensitivity of the tradi-tional least mean square method .Finally, a simulation experiment is conducted on the prediction of the Mackey-Glass chaotic time series , and the comparisons of the prediction results by different optimization methods are done to verify the superiority of the modified fruit fly optimization algorithm and in the case of outliers existing , the simu-lation results show the effectiveness and robustness of this proposed method .