摘要
结合原子分解与和声搜索算法,提出了一种电能质量扰动信号自适应分解及特征参数辨识方法。针对Gabor原子分解匹配追踪算法计算量大、实时性差的问题,首先利用傅里叶变换进行频谱分析,估计出扰动信号频率、幅值等参数,并以估计值作为搜索的初始解,加快算法的收敛速度;然后根据电能质量扰动信号特点将Gabor原子库分解为类基波库、脉冲库、谐波库、振荡库4个子库,依次搜索各子库,降低搜索的复杂度;再次,利用和声搜索算法快速、准确的全局搜索和协同搜索的特点对匹配追踪算法进行改进,加快了搜索速度;最后,依据获得Gabor原子索引参数实现电能质量扰动信号参数辨识。算例仿真表明,所提方法在保留匹配追踪算法优良重构性能的前提下,计算复杂度显著降低,搜索效率和收敛速度加快,扰动参数辨识精度得到提高。
Combined with atom decomposition and harmony search （HS）, an adaptive signal decomposition method is proposed to estimate parameters of power quality disturbances （PQD）. Some measures were taken to reduce computation time of matching pursuit （MP） algorithm in atom decomposition. First of all, in order to accelerate convergence speed, initial solution of MP is given by the approximate signal parameters which were obtained by discrete Fourier transform. Besides that, based on characteristics of PQD, the Gabor dictionaries are divided into four subsets named as similar fundamental dictionary, pulse dictionary, harmonic dictionary and oscillation dictionary. The complexity of MP was then reduced by changing searching parameters. Furthermore, the MP algorithm was optimized by harmony search, which has advantages of fast speed and more accuracy in global search and cooperative search. Finally, the PQD parameter estimation is achieved by the index parameter of the Gabor dictionary. The simulation results show that the novel method could reduce the computation complexity, improve the search efficiency, accelerate the convergence speed, and enhance the accuracy of PQD parameter estimation, while also retains the advantage of signal reconstruction through MP.
出处
《电网技术》
EI
CSCD
北大核心
2015年第1期194-201,共8页
Power System Technology
基金
国家863高技术基金项目(2012AA050215)~~
关键词
电能质量扰动
原子分解
匹配追踪算法
和声搜索算法
参数辨识
power quality disturbances atom decomposition matching pursuit harmony search parameter identification
