计量学报2025，Vol.46Issue(12)：1762-1767,6.DOI:10.3969/j.issn.1000-1158.2025.12.08
基于多层感知器的超声流量计在线检验系统
Multilayer Perceptron based Online Inspection System for Ultrasonic Flowmeter
摘要
Abstract
Ultrasonic flowmeter is the most important meter in China's natural gas trade transfer.In view of the high-dimensional and multi-variable working data of ultrasonic flowmeter,as well as the complex working conditions of natural gas station,an online inspection method of ultrasonic flowmeter based on Multilayer Perceptron(MLP)was proposed.The method includes two stages:real flow modeling and online testing.Firstly,the important correlation between the target quantity and the related input variables in the spatiotemporal dimension is fully analyzed.According to the test data of the gas flow standard device in the real flow verification of the ultrasonic flowmeter,the model of the online inspection for ultrasonic flowmeter was established,and the corresponding inspection threshold was set through the model performance evaluation and optimization.Secondly,the established model was used for online testing,and real-time online prediction was obtained through the interaction between the model and data control system of natural gas station.Based on analysis of the model and the comparison of the threshold,the online inspection of the ultrasonic flowmeter is realized.The R2 of the model is 0.45～0.67,and the root mean squared erro is in the range of 0.058～0.138,the deviation between the predicted value and real value is within±0.5%.The result shows that the proposed method can meet the accuracy degree requirements of the ultrasonic flowmeters.关键词
流量计量/超声流量计/机器学习/多层感知器/在线检验Key words
flow measurement/ultrasonic flowmeters/machine learning/multi-linear perception machine/on-line measurement分类
通用工业技术引用本文复制引用
苏祥伟,王永峰,俞宽,周婷..基于多层感知器的超声流量计在线检验系统[J].计量学报,2025,46(12):1762-1767,6.基金项目
国家管网集团浙江省天然气管网有限公司技术服务项目(GWHT20230040461) （GWHT20230040461）
