论文部分内容阅读
针对非线性系统的多模型建模问题,基于差分进化算法提出了一种优化建模方法.从系统的输入输出数据出发,将样本空间分割与局部模型建立相结合,首先将PWA辨识问题转化为MIQP问题;然后采用自适应混沌差分进化算法对模型数量及模型参数同时优化;最后利用支持向量基求取分割曲面方程.仿真结果表明,该方法能以最优的线性子模型集准确地逼近非线性系统.
Aiming at the problem of multi-model modeling in nonlinear systems, an optimal modeling method based on differential evolution algorithm is proposed.Based on the input and output data of the system, the segmentation of sample space is combined with the establishment of local model. First, the PWA identification problem is transformed into MIQP problem, and then adaptive chaotic differential evolution algorithm is used to simultaneously optimize the model number and model parameters. Finally, the support vector base is used to obtain the segmentation surface equation. Simulation results show that the proposed method can approximate non-linearly with the optimal linear sub-model set Linear system.