Reconstruction And Disturbance Identification Of Power Quality Data Based On Compressive Sensing

With the rapid development of smart grid and the vigorous implementation of new energy strategy,the problem of power quality pollution is becoming more and more serious.Effective detection and classification of power quality can further solve the problem of power quality.However,the traditional power quality analysis and processing process is based on the Nyquist sampling theorem,which not only requires high precision hardware acquisition equipment,but also a large number of redundant data bring huge storage and transmission pressure.Through parallel compression and sampling,the theory of compressed sensing breaks the constraint of traditional sampling theorem,and reconstructs the original signal by a small amount of observed values,which greatly improves the precision of the reconfiguration of the power quality signal and relieving the storage and transmission pressure.In this paper,the theory of compressed sensing is introduced in detail,including the sparse representation of signal,the design of measurement matrix and the design of signal reconstruction algorithm.The reconstruction algorithm and sparse representation of the signal are the focus of this paper.According to the different power quality disturbance signals,the sparsity analysis is carried out,and the corresponding signal model is constructed by MATLAB platform,which provides a theoretical basis for the later analysis.In view of the compression sensing reconstruction algorithm,a regularized adaptive compressed sampling matching pursuit(RACSMP)algorithm is proposed to compression reconstruct power quality data.It is determined that the Fourier transform base is used as the sparse transformation base and the Gauss random measurement matrix is used as the measurement matrix,the RACSMP algorithm can achieve accurate reconstruction of power quality signal under the condition of unknown sparsity.First,by calculating the correlation coefficient,the atom index value is stored in the candidate set to complete the preliminary screening of the atom.Then the backtracking idea is introduced to update the support set through the regularization process to achieve the second time screening of the atoms.Finally,the adaptive step size and step size are used to do the adaptive approximation of sparsity.The experimental results show that the average reconstruction progress of all kinds of single disturbance signals is up to 96.81% in the noisy circumstance with the signal to noise ratio above 20 dB,the RACSMP algorithm is more efficient,fast and stable than the traditional greedy algorithm for reconstructing the power quality data.Due to the power quality signal in the power system is often uncertain,and the actual signals are always complex,how to determine the type of the power quality signal is the key point of the research.Based on the sparse representation under the framework of compressed sensing theory,a new approach of sub-dictionary concatenate learning(SDCL)with label information is proposed to identify the power quality disturbances signal.Firstly,different types of testing and training of the PQD signal samples are dimension reduced feature extraction with principal component analysis,add the label information to train samples.Secondly,the different categories of power quality samples are trained into redundant sub dictionary and concatenated into structured dictionary.Finally,using dictionary learning algorithm to optimize the structured dictionary and the object class is determined through minimizing the redundant error.Simulation results show that the recognition effect of SDCL method is better than that of SVM and SRC,and having good anti-noise robustness.The average recognition rate of the single power quality disturbance is up to 94.20% and multiple power quality disturbance is up to 93.47% in the noisy circumstance with the signal to noise ratio above 20 dB.In a word,this paper makes an intensive study on two aspects: the data reconstruction of the single power quality disturbances and the identification of the complex power quality disturbances,which provides an effective method for the actual power quality signal processing and classification detection.