基于相似度分段及重采样的自适应波形数据压缩

Adaptive waveform data compression based on similarity segmentation and resampling

针对快速傅里叶变换(FFT)对稳态和暂态频繁切换的智能家电的电流数据压缩时,存在单周波非整数点造成的谐波失真及时间相位累计误差、分段衔接失真等问题,提出了一种基于相似度分段及重采样的自适应波形数据压缩方法。通过自相关系数自动分段得到稳态和暂态,对稳态数据重采样后用FFT和ZLIB级联压缩,对暂态数据采用ZLIB无损压缩,并提出综合压缩系数评价指标。以微波炉、空调、洗衣机进行测试,实验表明,在保证失真度的情况下,优选综合压缩系数较小的采样率; 5种采样率下的3种重采样方法中重采样到单周波128点综合效果较好,压缩比高于21. 73,均方根误差百分比在1. 04%之内。

In order to solve the problems of harmonic distortion,time-phase cumulative error and segmentation connection distortion caused by non-integer points of single-cycle,when FFT transform compresses the current data of smart home appliances with frequent switching between steady state and transient state,an adaptive waveform data compression based on similarity segmentation and resampling is presented. The auto-correlation coefficient is used to automatically segment the steady state and transient state. FFT and ZLIB cascade compression are used after re-sampling of steady-state data. ZLIB lossless compression is applied to transient data,and the comprehensive compression coefficient evaluation index is proposed. Tested in microwave oven,air conditioner and washing machine,the experiment results show that the sampling rate with small comprehensive compression coefficient is preferred under the condition of ensuring distortion;the three resampling methods under the five sampling rates have a good effect on the 128 points of single-cycle:compression ratio is higher than 21. 73,and mean square error is within 1. 04%.