应用FPGA和小波变换提取局部放电信号

Extracting PD Signal by Applying the FPGA and Wavelet Transform

小波变换是抑制局部放电中强干扰信号的有效工具,然而其巨大的计算量制约了它在局部放电在线监测中的现场应用。FPGA(Fielded Programmable Gate Array)器件兼具并行、串行的工作方式,有较高的并行计算能力,在现场数字信号处理领域具有较高的实时性。在分析了局部放电信号和白噪声的特点后,提出折衷小波包硬阈值变换算法。根据该算法,以DSP Builder为平台,模块化地设计出基于FPGA的流水线结构。从模拟现场信号中提取局部放电信号的实验结果表明系统可行。实际局部放电现场信号的处理结果表明,系统能较好的提取局部放电信号,同时满足了的实时性要求,在变压器的局部放电信号在线检测等领域具有一定的工程应用价值。

Wavelet transform is one of the most effective methods for extracting partial discharge （PD） pulses from strong noise, however, since the complexity of computation, it is hardly used in PD online monitoring system and other high real-time fields when it is carried out on traditional embedded serial model microprocessors. While the FPGA （Fielded Programmable Gate Array） devices can work in parallel model and serial model, and it can be programmed to be special purpose chip, so it is more efficient in real-time field . In this paper, the Mallat algorithm of wavelet transform and wavelet package was introduced firstly, then, a scheme combining eclectic wavelet package with hard threshold processing was presented after analyzing the characteristics of PD signal and white noise at online transformer monitoring. According to this scheme, a pipe-line modularized wavelet transform system was de- signed subsequently on the basis of the FPGA using DSP Builder as the developing platform. An experiment of ex- tracting imitated PD was done on this system, and the results prove that the system is feasible. At last, an experiment of drawing a real PD signal from a 220KV transformer was done, the results indicate that most of the real PD pulses are separated successfully, which is correspondent with the practice. It can be concluded that the system is feasible in field of extracting PD signal and other relative fields.