多元经验模态分解的发电机组多通道电信号滤波

Multichannel Electrical Signal Filtering of Generators Based on Multivariate Empirical Mode Decomposition

发电机组多需要满足严苛运行环境的需求,受大量不确定因素的影响,系统三相电流信号具有非线性以及非平稳特性。为实现各种扰动影响下的发电机组运行状态监测,将基于多元经验模态分解(Multivariate Empirical Mode Decomposition,MEMD)的方法用于风电系统电信号去噪。相对于传统单通道模态分解去噪算法,该方法能同时实现多通道信号的多尺度分解并保证相同模态数,有效地解决模态混叠问题。同时,将动态均方差(Consecutive Mean Square Error,CMSE)算法用于信号重构,只需计算各模态能量值,实现简单。仿真验证结果表明,基于MEMD的多通道滤波算法能够有效地实现多工况运行下的三相电流鲁棒滤波。

Most of large-scale generator sets need satisfy the requirements of harsh operating environment, the three-phase current signals of the system have non-linear and non-stationary characteristics because of the large number of uncertainties. In order to achieve the supervision under the generator operating conditions with a variety of disturbances, the method based on Multivariate Empirical Mode Decomposition（MEMD） is used to the electrical signals of wind power system de-noising. Compared with traditional single-channel modal decomposition de-noising algorithm, the proposed method can not only achieve multi-scale decomposition of multi-channel signals and ensure the same modal number simultaneously, but also overcome the problem of modal aliasing more effectively. Meanwhile, the Consecutive Mean Square Error（CMSE） algorithm is used for signal reconstruction, and it is simple to implement that only the energy value of each mode needs to be calculated. The simulation results show that the multi-channel filtering algorithm based on MEMD can realize the three-phase current robust filtering under multi-working conditions effectively.