论文部分内容阅读
传统原子库分解法在对信号进行分解时,往往先将信号分解为较易离散化的Gabor原子,再转化为衰减正弦原子,由于Gabor原子库并不符合振荡信号的本质特性,加之转化过程为近似转化,使分解存在双重误差.为克服以上不足,提出一种新的低频振荡分析方法,即基于混沌与Pseudo-Newton法组合优化的直接衰减正弦原子库分解方法.为了解决离散困难和六维空间寻优效率低的问题,采用混沌与Pseudo-Newton法组合优化方法,可以同时对6个参数进行寻优,极大的提高方法的速度.通过对自合成信号、仿真信号、华中电网能量管理单元实测信号等3个算例进行分析,并与其他2种辨识方法进行对比,表明该文算法在计算精度、辨识速度、抗噪能力方面都有较大优势.“,”Frequently, the signal is decomposed into Gabor atoms which are easy to be discretized, and then transformed it to damped sine atom in the traditional atomic decomposition method. But because the Gabor atomic library does not accord with the essential characteristics of the oscillation signal, and the transformation is an approximate process, so there is a double error in the decomposition. In order to overcome the above shortcomings, a new identification method was presented in this paper, which called direct damped sine atomic sparse decomposition method that based on the combinatorial optimization of chaotic and Pseudo-Newton. In order to solve the problems of discrete difficulties in the attenuation of the sine atom library and the low efficiency of the six dimensional space optimization, this method adopts the combination method of chaos and Pseudo-Newton. It can optimize six parameters simultaneously, which greatly improves the speed of the method. The analysis result of the signal of synthetic and the simulation and the power management unit measured signal of central China power grid, and the comparison result of Prony methed and traditional atomic decomposition method, prove the advantages in accuracy, speed, anti-noise ability and efficiency.