一种新的T-S模型混合辨识算法

A novel hybrid T-S model identification algorithm

提出一种新型混合辨识算法HIA,以解决传统T-S模型辨识方法中所存在的不完全优化问题,如FCM与最小二乘法相结合的辨识方法就存在这样的问题.HIA通过将FCM、和声搜索算法以及最小二乘法相结合,并引入了误差反馈机制,以实现对所有参数的整体优化,并避免陷入局部极小点.论文将HIA应用到陀螺稳定平台的T-S模型辨识中,通过与传统辨识方法比较MSE值可以看出,HIA能够获得更高的辨识精度.这表明,对于实际的非线性系统,HIA能够有效解决传统辨识方法的不完全优化问题.

To overcome the drawback of regular T-S model identification techniques,such as the FCM and least-squares method,a new Hybrid Identification Algorithm（HIA） is proposed in this paper.The HIA can simultaneously optimize all the model parameters and avoid being trapped into the local minima by merging the FCM,Harmony Search（HS） and the least-squares method together and using the error feedback mechanism.Our HIA is employed in the T-S modeling of the Gyro-stabilized platform.By comparing the MSE peformance,the HIA can indeed yield a superior MSE performance over the conventional identification methods.The identification results show that the HIA can effectively overcome the incomplete optimization problem of the conventional identification methods.