Research On Smart Distribution Grid Acquisition System Based On Compressed Sensing

In recent years,the research on Compressed Sensing(CS)has been increasing.Its research focuses mainly on image reconstruction,radar imaging,data communication,power system and so on.This thesis focuses on the problem of redundancy in the amount of data collected in the distribution network and the need for a large amount of data storage space.On the basis of summarizing the theory of compressed sensing at home and abroad,an analog signal processing model(AIC)based on compressed sensing is used.Applying it to the power signal,the research work on the simulation framework,reconstruction accuracy,hardware design and implementation of the model was carried out.The main contents are as follows.Firstly,it compares and analyzes the theory of compressed sensing and the classical Nyquist sampling theory.It draws the advantages of CS theory in various aspects and the application of the theory.That is,the signal itself is sparse or sparse or compressible in some domain.At the same time,using MATLAB to simulate and analyze several reconstruction algorithms based on CS.Under the condition that each set of simulation data is repeated 1000 times,the reconstruction algorithm with good reconstruction performance is selected as the orthogonal matching pursuit method OMP.The reconstruction algorithm used in this study.Since the power signal is sparse after Fourier transform,the feasibility of CS for power signal is determined,and the model of the relevant power signal is given.Based on this,an analog information converter(AIC)simulation model based on compressed sensing theory was built,and an improved OMP reconstruction algorithm based on AIC model was proposed.And through the compression ratio,reconstructed signal-to-noise ratio,mean square error three indicators in each case were simulated 200 times,the AIC model in the power signal compression sampling reconstruction,the best compression ratio is 1/4 Refactoring works best.An integral ignition circuit(IF)model that also conforms to the CS theory is proposed,and an IF simulation model is constructed using MATLAB.At the same time,a reconstruction algorithm corresponding to the IF model is developed.Through simulation analysis,the IF model can adaptively sample the signal,adjust the sampling interval along with the magnitude of the signal amplitude,compress the sampled signal into a unit pulse sequence,and generate no quantization error.Through the comparison between AIC and IF model,it is concluded that the IF model is inferior to AIC model in signal reconstruction,parameter setting,etc.It is determined that the AIC model is used as the hardware design basis in this thesis.Finally,based on the AIC model block diagram and simulation model,hardware design and implementation.For each module in the AIC system,from the principle to the framework to the design of the circuit and the realization of related functions,detailed solutions,equipment selections,circuit schematics,and hardware in-kind are given.At the same time,a voltage level adjustment circuit is designed based on the generated sequence.Both pseudo-random sequence and AD sampling are controlled by FPGA.A voltage level adjustment circuit is designed based on the generated sequence;Mixing function hardware design using multiplier AD835:The hardware of the integral function is equivalent to the low-pass filter,and the OPA4353 is selected as the operational amplifier.And for each module and the required supply voltage of the chip is different,using the regulator type ASM1117 series designed a power circuit that can provide different voltage.Finally completed the overall hardware design and implementation of the analog information converter(AIC).After hardware testing and debugging,the feasibility of CS theory for power signal acquisition compression and reconstruction was determined.