| 摘要: |
| 基于小波变换理论进行混合有源电力滤波器（HAPF）的谐波检测，但分析Mallat小波分解算法可知其存在缺陷：若直接对所有数据进行小波变换，计算量大且误差明显；提高采样频率会使单位时间内处理过多数据。提出在小波变换前将采样数据与窗函数相乘，在有效滤去高频分量的同时减少小波分解层数。基于Blackman窗、Hamming窗和Hanning窗的比较分析，选取Hamming窗作为小波分析前的辅助窗，可使99.963 % 的能量集中在主瓣内。基于PSCAD／EMTDC与Matlab进行仿真实验，结果显示，该方法响应时间约为10 ms，且小波分解次数减少，比较仿真结果可知检测精度提高。 |
| 关键词: 混合有源电力滤波器 谐波检测 小波变换 Hamming窗 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| HAPF harmonic detection based on windowed wavelet transform |
|
SUN Haoyun1, MU Longhua2
|
|
1.Department of Electrical Engineering，Tongji University，Shanghai 201804，China;2.Department of Electrical Engineering，Tongji University，Shanghai 201805，China
|
| Abstract: |
| When Mallat algorithm is used in HAPF（Hybrid Active Power Filter） harmonic detection，if the data are directly processed by wavelet transformation，the calculation load will be huge and the error obvious；if higher sample rate is applied，the data to be processed in unit time will be large. It is proposed to multiply the sampling data by Hamming window before wavelet transform，which filters the high frequency components and reduces the wavelet decomposition levels. Based on the analysis of Blackman window，Hamming window and Hanning window，Hamming window is selected to concentrate 99.963 % of energy in the main lobe. Simulations with PSCAD／EMTDC and Matlab prove that，the HAPF harmonic detection based on windowed wavelet transform has less wavelet decomposition levels and better accuracy，with the response time of about 10ms. |
| Key words: hybrid active power filter harmonic detection wavelet transform Hamming window |
