Design And Implementation Of Digital Signal Processing Algorithm For High Pressure Hydrogenation Flowmeter

The Coriolis mass flowmeter is widely used in various industries due to its high accuracy and strong anti-interference ability.However,under the high-pressure hydrogenation environment of 70 Mpa,the accuracy of the Coriolis mass flowmeter can only reach 0.5 level with a repeatability of 0.25%.This is mainly due to the impact of the flow meter sensor in the high-pressure environment,which increases the noise component in the sampling signal,thus reducing the algorithm performance.Additionally,when the measured fluid is liquid hydrogen,vaporization occurs,causing gas to mix into the measured fluid,and the compressibility of gas as well as relative motion with the liquid also contribute to the decrease in accuracy.In light of the aforementioned issues,this article will undertake the following specific research tasks:Firstly,this study compares three commonly used time-domain processing algorithms-zero-crossing detection,Hilbert transform,and orthogonal demodulation-in terms of their real-time performance,noise sensitivity,and susceptibility to frequency fluctuations.Additionally,based on the characteristics of steady-state single-phase flow signals,this study employs ALNF based on frequency estimation and feedback factor improvement,Fourier series fitting,and Kalman filtering to improve the computational accuracy of the zero-crossing detection algorithm.Secondly,the RWM signal model was used to simulate the sampling signal of the Coriolis mass flowmeter,and an improved Hilbert transform algorithm was used for phase difference calculation.Due to the impact of high pressure,the noise component in the sampled signal increases,resulting in a higher demand for noise reduction.In this thesis,a k-means clustering-based improved wavelet threshold denoising was used to improve the signal-to-noise ratio,where the k-means method classified wavelet coefficients into two categories,followed by threshold function processing of the wavelet coefficients.Through simulation verification,it was found that when the signal-to-noise ratio is greater than 21.5d B,the denoising effect is better than that of traditional algorithms.At the same time,the endpoint effect of the Hilbert transform was suppressed by using the periodic extension method,which improved the algorithm accuracy.Then,the Hilbert transform algorithm was refined and calibrated through experimental verification using the AMF006 AH series hydrogen mass flow meter.During the single-phase water flow calibration,the flow meter successfully recorded three sets of data at distinct flow rates,with a measurement error of less than 0.15% and a repeatability of less than 0.075%.Furthermore,in the hydrogen refueling calibration experiment,three testing devices were utilized to measure a cumulative mass of 5kg,with a measurement error of less than 0.15% and a repeatability of less than 0.075%,thus satisfying the precision requirements of level 0.15.Finally,machine learning algorithms were used to correct measurement results under gas-liquid two-phase flow conditions.Density descent was used instead of GVF to solve the problem of inaccurate measurement of GVF in practical applications.The data set was divided into a training set and a test set in a ratio of 9:1 using hierarchical sampling.The GA-SVR algorithm,RF algorithm,and M5 P model tree algorithm were used to analyze and verify the data set.It was found that the M5 P model tree had the best correction effect on the data set.In summary,this thesis provides a basic idea for signal processing of the Coriolis mass flowmeter under high-pressure hydrogenation conditions and provides methodological support for improving the practicality and stability of the flow meter.