Research on Measurement Error Correction of Hydrogen-Doped Natural Gas Ultrasonic Flowmeter Based on Artificial Neural Network

This study analyzed the metrological performance of ultrasonic flow meters in the transportation of hydrogen blended natural gas and made corrections. In order to solve the error problem of traditional ultrasonic flowmeter under hydrogen mixing condition, a data fusion model based on the principle of time difference ultrasonic flowmeter and artificial neural network (ANN) is proposed. By inputting the simulation results of flowmeter, gas composition (hydrogen ratio) and flow field characteristics, ANN is used to predict and correct the error. The results show that the absolute value of the measurement error of the composite model is reduced to 0.2%, 0.5% and 0.6% at 10%, 20% and 30% hydrogen mixing ratios, respectively, which is significantly better than the error level of the traditional model. The research proves that the combination of machine learning technology can effectively improve the accuracy of complex parameter coupling analysis, and can provide theoretical support for safe transportation and accurate measurement of hydrogen-doped natural gas pipelines. In the future, it is necessary to further integrate multi-type flowmeters and experimental data to optimize the applicability of the model.