Research On Information Sensing Method Of Shock Wave Signal Based On Compressed Sensing

When the impact of artillery muzzle propagates to space around,it will cause damage to the surrounding targets,so protection measures should be designed according to the shockwave damage effect.To effectively evaluate the damage effect of the muzzle shockwave,it is necessary to improve the accuracy when sampling the muzzle shockwave signal.During the actual signal sampling process,target signal sampling rate and the actual sampling rate of the sampling device as well as the contradiction between the amount of data generated by sampling and the storage device capacity.Compressed sensing breaks through the limitations of classical sampling systems on signal sampling rate and can effectively reduce the signal sampling rate and data volume.However,most of the research related to compressed sensing is aimed at the digital domain,after sampling,compressed sensing is applied for transmission and reconstruction,the sampling method based on which can only reduce the data volume at the data transmission link and data processing terminal.The analog information converter is an application-oriented extension of compressive sensing in the analog domain which can reduce the data volume and reconstruct with high accuracy at the sampling front end.In this paper,we will analyze the mathematical model of random demodulator based on compressed sensing and then clarify the improvement directions.To make Specificity improvement for random demodulator by improving the correlation between the input signal and the mixed frequency sequence signal.The classical random demodulator consists of signal mixing,filtering,and sampling while m-sequence is commonly used in the mixing process.Although m-sequence has a wide spectrum distribution and good randomness,it is poorly correlated with the muzzle shockwave signal we investigated in this paper.Thus,its application in the sampling and reconstruction process of muzzle shockwave is not ideal.In this paper,instead of m-sequence that is used for mixing in the classical random demodulator,the Markov chain sequence will be applied and adjusted to increase its correlation between the mixing sequence and the buzzle shockwave signal from 0.0176 to 0.122.Then random demodulator system simulation will be performed to analyze the factors that affect reconstruction error of muzzle shockwave signal,which is through comparative experiments in the condition of decreasing sample rate from 2MSPS to250 KSPS,making 87.5% data volume reduction and ensuring the reconstruction error below 2.5%.Finally,the Hardware experiment will be conducted to implement the perceived down-sampling of muzzle shockwave signal by using FPGA as the main controller chip to build the hardware system and improve the low-pass filter circuitry of the random demodulator.When the muzzle shockwave signal is at 12.5% of the Nyquist sampling frequency,the reconstruction error is less than 5%.Based on results of theoretical derivation,system simulation and hardware experiments,the improvement for random demodulator in this paper can reduce the signal sampling rate and the overall data volume of the system providing a reference for the improvement of other signal sampling methods.