Reconstruction Algorithms And Sampling System Of Power Quality Based On Compressed Sensing

With the rapid transformation of energy system and the growth of new types of load,power quality will face more severe challenges in the power system characterized by high penetration of renewable energy and wide application of power electronic equipment.Futhermore,the highperformance power quality monitoring system puts forward higher requirements for the speed and accuracy of sampling hardware and also brings massive data.The traditional sampling system and compressed storage mode are facing great challenges.In this paper,considering the sparse features of power quality signals,compressed sensing theory is used to realize low-speed sampling and precisely reconstruct the signal.The efficient signal reconstruction algorithms and an implementation scheme of compressed sampling system for engineering application are proposed.Firstly,to solve the problems of relying on sparsity and limited performance of existing reconstruction algorithms,a power quality reconstruction algorithm is proposed based on stagewise weak orthogonal matching pursuit(SWOMP).In this algorithm,the correlation function of the atom and residual is used to achieve selection,which can improve the reconstruction speed,realize the blind sparsity and reduce the restriction on observation matrix.Then the reconstruction performance for power quality signals is studied.Meanwhile,the influence of the number of measurements and selection threshold on the reconstruction performance are analyzed.The reference value of the number of measurements under different error levels and the optimal range of selection threshold are given.Furthermore,a backtracking stagewise weak orthogonal matching pursuit(BSWOMP)algorithm is proposed through introducing the idea of backtracking.The algorithm improves the accuracy of reconstruction by re-evaluating the support set.The comparison shows that BSWOMP algorithm can achieve higher compression ratio and reconstruction accuracy while retaining the advantages of SWOMP algorithm,and has more obvious advantages in power quality signal reconstruction.Secondly,considering that the sparse characteristics of transient power quality signals are not obvious under Fourier transform,a transient power quality signal reconstruction algorithm based on complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)is proposed.The algorithm uses CEEMDAN decomposition to stabilize the transient disturbance and preprocesses the decomposed components to obtain the best sparse representation.Then,combined with compressed sensing theory,the signal is measured and reconstructed.Simulations are carried out to verify the performance of this algorithm for typical transient power quality signals.Besides,compared with the orthogonal matching pursuit and empirical mode decomposition algorithm,the results show that the reconstruction algorithm based on CEEMDAN can achieve high-precision reconstruction of transient disturbance characteristics and ensure high compression ratio in processing transient power quality signals.In the end,as for the shortcomings of traditional power quality sampling system,a compressed sampling system for power quality signals based on compressed sensing is studied.The compressed samping system has two parts,including the front-end sampling and the backend reconstruction.Thereinto,the front-end sampling is realized by random demodulator.Through theoretical analysis and mathematical derivation,the corresponding observation matrix of front-end is given.In addition,the BSWOMP algorithm is used in the back-end reconstruction which is realized when the observation matrix and sampling values are obtained.The simulation model of compressed sampling system is built.The sampling and reconstruction experiments are carried out for steady-state and transient power quality signals and the anti noise ability of the system is analyzed.The results show that the proposed compressed sampling system can realize low-speed sampling and high-precision signal reconstruction in the process of sampling and reconstruction and has strong anti-noise ability.