绝缘子泄漏电流的自适应SPIHT数据压缩

摘要

A fault in any one string of insulators may cause a power network accident, so insulator online monitoring is of vital importance. Sampling frequency of leakage currents must be high enough to find abnormalities in insulators, yielding large amount of data, hence the data compression of leakage currents is of great significance. An adaptive version of SPIHT is proposed, which can partition sets of wavelet coefficients according to their significance, thus effectively reduces the number of judgments on whether the coefficients in one set are all 0, making itself well suited for coding signals with high-level noise such as leakage currents. In view of the high noise level and periodic redundancy of the leakage currents, adaptive SPIHT and DPCM are utilized for the coding of wavelet detail and approximation coefficients, respectively. The compression efficiency can be further improved if context-based adaptive binary arithmetic coding is employed. The proposed algorithm is evaluated on testing data of insulator leakage currents, showing compression performance significantly better than SPIHT. Compared with existing two-dimensional algorithms for data compression in power systems, it can do compression after data sampling within just one AC power working cycle, so it is more suitable for the real time or online situations.%任何一串绝缘子故障都可能诱发电网事故，因此绝缘子的在线监测十分重要。为了发现绝缘子放电等异常，泄漏电流的采样频率须比较高，数据量大，这就要求对这些采样数据进行压缩。本文提出自适应SPIHT算法，该算法可以根据小波系数集合的显著性自适应地进行集合划分，有效地减少了原始SPIHT算法判断集合中系数是否全部为0的次数，尤其适合压缩泄漏电流这类高噪声信号。针对泄漏电流周期冗余和高噪声的特点，本文利用自适应SPIHT和DPCM分别对泄漏电流的小波细节分量和近似分量进行编码。如果再应用上下文自适应二进制算术编码，那么编码性能还可以进一步提高。对绝缘子泄漏电流的实测数据检验了算法的压缩性能，和SPIHT算法相比，压缩性能显著提高。该算法同已有的用于电力系统数据压缩的二维算法相比，允许采完一个工频周期的数据后就进行压缩，更适用于实时或在线的场合。