Research On Sampling And Detection Algorithm Of Compressed Sensing Power Quality Signals

With the continue expanding of the industry scale and the rapid development of science and technology, a variety of non-linear, shock, volatility load substantial increase in modern system, which makes the power system suffered increasingly serious power quality damage. Nowadays, how to execute effective power quality analysis and research has become an urgent and important work, which catches wide attention by electric field at home and abroad. Two questions are mainly faced to enhance and improve power quality, on the one hand collection, compression, storage and transmission problems of massive power quality data need to be solved, on the other hand detection, location and identify problems need to be solved.This paper firstly introduced the definitions, standards, classification and research status and development trends of the power quality, and constructed a variety of single power quality disturbances and multi-disturbance signal models.Secondly, this paper discussed generation, research status, basic theory and implementation of compressed sensing theory in depth. Compressed Sensing theory was applied on sampling and compression of power quality disturbance signal for the first time on the basis of comparing advantages and disadvantages of traditional sampling methods before the compression supported by Nyquist and feature extraction method used home and abroad. Two kinds of one-dimensional compression sampling and reconstruction method based on compressed sensing theory was designed in this paper. The performance indicator under different compression sampling ratio by two methods was given. Simulation and experiment show that the two methods could sample and compress at the same time and could reconstruct power quality disturbance with high signal to noise ratio.Thirdly, this paper mapped power quality signal to two-dimensional image, and first researched compression sampling algorithm of unsteady and steady power quality disturbances according to the principle of image sparse representation. Simulation analysis and experimental verification of the algorithms was performed for measured signal of calibration source with multiple disturbances, which show that reconstruction meets the requirements of power quality analysis. On this basis, a series of simulation experiments was carried out for the comparative analysis of reconstruction results of the two dimensional Fourier and wavelet matrix.Finally, simulation and experiment were conducted using the time-frequency analysis method wavelet transform widely researched in current documents, and analyzed its advantages, disadvantages and performance issues in processing power quality disturbances. This paper studied and designed the integrated approach of compression sampling and parameter estimation of power quality disturbance with combining compressed sensing theory and atoms decomposition. Using a small amount of measurement data achieved from compression and sampling of Short-term and steady-state disturbances signal directly, we could achieve detection and reconstruction at the same time and draw a comprehensive parameter estimation. This paper also analyzed algorithm performance with the noisy and comparative analysis the lack of the detection using wavelet transform and the advantages by the proposed method, which provided a solid foundation of theory and practice for subsequent identification and classification of the disturbance.