基于“自适应遗传算法”的磁轴承系统辨识

Identification of Magnetic Bearing System Based on Adaptive Genetic Algorithm

磁轴承系统辨识是对其进行自适应控制、鲁棒控制、扰动抑制及故障诊断、容错的基础。磁轴承系统输出信号中含有难以确定统计特性的有色噪声,会对磁轴承系统线性模型辨识的准确性造成影响。针对这一问题,提出一种采用"自适应遗传算法"的系统辨识方法。这种方法以输出误差为准则,利用"自适应遗传算法"对系统传递函数参数进行优化,算法准确性与噪声统计特性无关,且避免了标准"遗传算法"中的"早熟"现象。与目前应用广泛且有较高辨识精度的卡尔曼滤波相比,输出误差的均方根下降了71.6%。实验结果表明,这种方法可以成功避免噪声的"有色"特性对辨识精度的影响,充分验证了方法的有效性。

System identification is the basic for applying adaptive control,robust control,disturbance rejection,fault diagnosis and fault tolerant basis to magnetic bearing system.The statistical characteristic of the colored noise is difficult to determine which is contained in the output signal of the magnetic bearing system.As a result,the identification accuracy is affected seriously.In order to solve this problem,this paper puts forward a kind of ＂adaptive genetic algorithm＂ for magnetic bearing system.In this algorithm,output error is chosen as the criterion.The parameters of system transfer function are optimized according to adaptive genetic algorithm,and appearing to standard genetic algorithm,the ＂premature＂ could be avoided.Appearing to Kalman filter,which is a widely used identification algorithm with high accuracy,the algorithm in this paper has a better precision,that the mean square root of output error decreased by 71.6％.Experimental results show that this algorithm can successfully avoid influence of colored noise with unknown characteristics,which just validate the effectiveness of the algorithm.