非参数统计模型选择在SDVC数据压缩中的应用  

APPLICATION OF NONPARAMETRIC STATISTICAL MODEL SELECTION TO SDVC DATA COMPRESSION

作　　者：李鹏[1] 杨洪耕[1] 

机构地区：[1]四川大学电气信息学院,四川省成都市610065

出　　处：《电力系统自动化》2004年第20期47-51,共5页Automation of Electric Power Systems

基　　金：中华电力教育基金会许继奖教金资助项目(2002)~~

摘　　要：在电能质量监测仪的开发中,针对短时电压变化(SDVC)信号,提出了一种新的数据压缩和消噪算法。利用小波函数以及局部余弦基(LCB)构成模型库,首先对信号进行离散小波变换(DWT)和局部余弦变换(LCT),然后应用信息论的MDL(minimum description length)判据作为价格函数(cost-function)对模型进行初选,最后结合压缩性能评价指标选择最佳的信号模型,并以MDL确定的最佳分解系数子集作为模型参数。针对不同噪声水平和不同信号类型,这种算法具有数据自适应(data-adapted)能力,不需要进行任何先验的参数设置(例如阈值设置)就能确定保留分解系数的最佳个数,并能根据信号动态选择模型。计算实例表明该算法能够满足电力部门的要求,同时图表证明了MDL能够在信号保真度与信号压缩效率之间找到最佳的契合点。In view of the short duration voltage change (SDVC) signal, this paper proposes a new data self-adapted algorithm for data compression and denoising technique in the development of a power quality (PQ) monitor. A model library is made up of wavelet bases and local cosine bases (LCB). First, the signals are transformed via discrete wavelet transform (DWT) and local cosine transform (LCT) respectively. Secondly, the information theory-based MDL (minimum description length) criterion is used as the cost function to make a primary selection of models. Finally, the best signal model is selected by referring to the compression efficiency indicators, and the best subset of transform coefficients confirmed by MDL is treated as model parameters. This algorithm has the data self-adapted capability for different noise levels and different sorts of signals. It can not only select the 'best' number of retained decomposing coefficients without any parameter setting (such as threshold) or subjective judgments, but also choose the 'best' transform model according to the signals dynamically. Computer results show that the algorithm can satisfy the requirement of electricity utility industry, while it is proved by the diagrams that the MDL criterion gives the best compromise between the fidelity of signal and the efficiency of signal compression.

关 键 词：电能质量 数据压缩 消噪 离散小波变换 MDL 非参数模型选择 

分 类 号：TM933.4[电气工程—电力电子与电力传动]