Research On Compressed Sampling Of Mechanical Vibration Signal Based On MWC

The vibration is usually accompanied by the change of the vibration and show through the noise produced by the vibration and vibration when the mechanical equipment failures.Through monitoring and analyzing these vibration signals,we can predict and locate faults in advance,understand the causes of failure and the extent of the deterioration,and reduce or avoid fault occurrence.Along with the continuous development of mechanical equipment in the direction of intelligence,precision and high-speed,the frequency band of mechanical vibration signal is wider and wider.Sampling with traditional Shannon Nyquist sampling theory will bring a series of problems,such as mass data transmission,storage and computation.Modulated wideband converter is an efficient solution for analog to information conversion,which breaking the limit of Nyquist sampling rate and can realize low speed sampling and accurate reconstruction.At the same time,the MWC also have the advantage of suitable for most type of signal,small computation,easily realization,low hardware costs.In this paper,the sampling system of MWC that applied to sample the data of mechanical vibration signal was studied,the main content is as follows:(1)The basic theory of compressed sensing is introduced,the advantages and disadvantages of the classic AIC design scheme are compared,and the working principle of MWC is analyzed.The design idea,sampling structure an d working principle of the single channel MWC sampling structure are described.A mathematical model of signal is built and simulation experiments are carried out to expl ore the influence of bands,bandwidth and signal to noise ratio on MWC signal reconstr uction process,compared the reconstruction performance of multi-channel MWC and single channel MWC.(2)The successful reconstruction probability of MWC sampling system is low or even failed under the condition of low signa l to noise ratio.To solve the problem,a signal reconstruction method based on independent component analysis under low signal to noise ratio in MWC is proposed.First,a modulated wideband converter is used to sample the input signal,then use ICA method to denoise the signal sampling data.Finally,the compressed sensing optimization algorithm is used to reconstruct the signal for de-noising data.The experiment results shows that,the proposed one can realize high probability signal reconstruction with fewer sampling channels under the condition of low signal to noise ratio;When the band number was 2 、4、 6,compared with traditional MWC,the number of channels that ICA-MWC required for high probability signal reconstruction and the overall sampling rate was reduced by 53.85% 、58.06% and 68.33% respectively.(3)With the band of mechanical vibration signals getting wider,traditional Shannon-Nyquist sampling theorem is used for data collection,which will bring a series of problems of transmission,storage and processing of massive data.Here,a method o f signal detection of mechanical vibration based on modulated wideband converter is proposed.Firstly,dividing the mechanical vibration signal spectrum into several segments according to the concentration extent of mechanical vibration signal energy in the frequency domain.Then,designing the reasonable parameter of MWC according to the prior knowledge of the highest frequency of mechanical vibration signal,the number and maximal bandwidth of divided spectrum segments,etc.Finally,using the MWC to sampling the mechanical vibration signal with low speed and recover the original signal from the undersampling data.The experiment results shows that,the method can effectively reduce the sampling rate of the back-end sampling device and realize high probability signal reconstruction from a small number of sampled data without losing the vibration information.