Abstract: To ensure the stable metering performance of gas ultrasonic flowmeters and the fairness of trade transactions while reducing operational costs, this study investigates three in-service inspection methods: the standard flowmeter verification method, the speed of sound verification method, and the machine learning method. Taking the flowmeters at Xinjiang Gas Group's Xinhua Gate Station as the research subject, the first two methods were applied in case studies, and their effectiveness was evaluated through comparative analysis of test data. The results show that the standard flowmeter verification method can effectively assess flowmeter performance, while the speed of sound verification method evaluates metering performance through sound-speed-related parameters. Both methods demonstrated compliance in the case studies, though deviations in sound speed were found to potentially correlate with medium parameter acquisition. The study reveals that each method has its advantages and limitations. Regular in-service inspections and timely anomaly resolution can ensure measurement accuracy, providing practical guidance for industrial applications.