Data Compression Technology Of Power Quality Based On Wavelet And Compressed Sensing

With the expanding of the power system, various non-linear and impact load increase, which damages power quality more and more seriously. To solve the problem, power sector must analysis the signals in power quality accurately and efficiently. Under the background of massive electric power data, it’s very necessary to compress the data.In the thesis, aiming at problems existed in the transient data compression, the research is carried out. The data compression method of power quality based on wavelet and compressed sensing is proposed in order to analysis power quality correctly and timely. The main work of this paper is as follows:First, the theory of wavelet transform and compressed sensing is studied deep.According to the IEEE standard, some common models of transient disturbance signal in power quality is built.Second, aiming at the problems of storing and transmitting large data in power system, the thesis presents a compressed scheme of the transient signal in power quality which is based on two-dimensional integer wavelet transform with lifting scheme, run length encoding and Huffman coding. The scheme maps one-dimensional power quality data to two-dimensional space, which greatly increases the redundancy. Two-dimensional integer wavelet transform with lifting scheme is more suitable for decomposing two-dimensional transient data.The obtained wavelet coefficients are dealed with in turn by thresholding, run length encoding and Huffman coding to further improve the compression ratio of the data.The simulation results show that the compression ratio of the signal is improved under the condition that the signal to noise ratio and the mean square error are not changed too much. Compared with the traditional two-dimensional discrete wavelet algorithm, the error can be limited to a very small range when the compression ratio is increased by about one time.Last but not the least, to break through the traditional Shannon sampling theorem,this thesis introduces compressed sensing theory, comes up with a more efficient compression and reconstruction scheme which is based on the fast fourier transform and generalized orthogonal matching persuit algorithm for power quality transient signal. In this scheme, the fast fourier transform is used to sparse the signal, and thesparsity is reasonably increased by threshold processing. Then observate randomly the sparsed signal by the identity matrix. The original signal is reconstructed by using the generalized orthogonal matching pursuit algorithm. The simulation results show that after threshold processing, the compression ratio of some disturbance signal is improved by 1~4 times, and all the signal to noise ratio is increased by about 30%. In addition, compared to the classical orthogonal matching pursuit reconstruction algorithm, the signal reconstruction accuracy is increased by 35%under the premise of setting both measurement dimension to 100. So this method can provide a new way for related technicians to solve the mass data storage and transmission in power quality.