Method Researches Of Power Quality Analysis By Wavelet Transforms And Matching Pursuits

Power quality problems, including steady-state and unsteady-state phenomena, are mainly caused by various non-linear loads in power system. Power quality signals are composed of steady-state components (such as harmonics and interharmonics) and unsteady-state disturbances (such as unsteady-state pulses and unsteady-state oscillations). In general, steady-state components are high in energies and large in time supports and periodic in waveforms while unsteady-state disturbances are low in energies and random in emergences. The spectra of steady-state components are compact and sparse in frequency-domain considering signal morphological component characteristics; on the contrary, the waveforms of unsteady-state disturbances are compact and sparse in time-domain. The accuracies of unsteady-state disturbance analysis of power quality signals are reduced by the steady-state components with high amplitudes and energies. At present, harmonic power measurements and power quality disturbance parameter estimations are of the current hot-points and difficult problems to be solved of power quality analysis. Firstly, harmonic power measurements in sub-bands were implemented by wavelet transforms. The different sources of measurement errors were revealed, and the error inhibition methods were proposed. The integer coefficient lapped data windows were constructed according to the behaviors of error variations. Considering of the calculations of traditional wavelet packets for harmonic power measurements in sub-bands, an optimal wavelet packet algorithm was proposed based on sub-band energy entropy threshold criterions.Secondly, frequency-domain matching pursuits (FDMP) algorithm was proposed to decompose and analyze the power quality steady-state components and unsteady-state disturbances. The separations of steady-state components and unsteady-state disturbances were achieved by FDMP, and the parameters of steady-state components were estimated with accuracies. The property of phase best matching at the midpoint of time-domain data windows was discovered while the analysis errors of periodic signals by matching pursuits were being studied. The theoretical proof of the property was produced and this property was adopted to estimate the sinusoidal waveforms and phases of periodic signals with accuracies. FDMP algorithm and the property of phase best matching have been evaluated with the experimental power quality data, and the unsteady-state disturbances of experimental power quality signals have been isolated in time domain during the experiment. And unsteady-state disturbances in sparse form could be analyzed and classified easily and accurately.Thirdly, regulation based morphological component analysis (RMCA) method was proposed to analyze unsteady-state disturbances of power quality signals. The time-frequency atoms and redundant dictionaries suitable of the analysis of unsteady-state disturbances were constructed. Some important methods of unsteady-state disturbance analysis were proposed:the methods of the locations and segments of unsteady-state disturbances, the methods for estimations of the frequencies and attenuation coefficients of unsteady-state disturbances, as well as the sub-space matching pursuits (SSMP) algorithm for the parameter estimations of oscillation unsteady-state disturbances. The accuracies of the analysis and parameter estimations of unsteady-state disturbances by RMCA were evaluated with experimental data. Contrast experiments showed that the analysis of unsteady-state disturbances by RMCA method could be more rapid in processing speed and more accurate in parameter estimations than that by conventional methods. Furthermore, SSMP algorithm could be used to estimate the parameters of oscillation unsteady-state disturbances with attenuation amplitudes or non-sinusoidal waveforms.Finally, the power quality signal data compression methods were studied according to the characteristics of power quality signals. The low frequency data obtained from wavelet orthogonal filter banks were compressed by Ray-period methods to realize lossless compressions, and the high frequency data were compressed by threshold methods to obtain a higher compression ratio. These data compression methods were implemented with practical measurement data.The measurements of harmonic power in sub-bands implemented by the anti-leakage algorithms proposed in this paper are accurate. The harmonic power distributions of power quality signals could be found out clearly and the harmonic sources could be identified correctly through the power measurement results in sub-bands. And harmonic pollution controls and electric energy consumption cost settlements could be implemented with accuracies based on the same results. The proposed power quality matching pursuit algorithm in frequency domain (FDMP) implemented the separations and parameter estimations of steady-state components; and the unsteady-state disturbances obtained by FDMP were sparse in time-domain. For unsteady-state disturbances, the proposed method RMCA could analyze multiple disturbances; meanwhile, accurate estimations of the parameters of unsteady-state disturbances could be obtained. So, a new and effective method for power quality analysis was provided.