Gas Flow-Rate Measurement Using a Transit-Time Multi-Path Ultrasonic Flow Meter Based on PSO-SVM.

Xiaoyu Tang,Qinmin Yang,Youxian Sun

DOI: https://doi.org/10.1109/i2mtc.2017.7969990

2017-01-01

Abstract:Transit-time multi-path ultrasonic flow meters (TM-UFM) have been widely employed for measuring flow rate of gas and liquid. However, its applicability is still dragged by the difficulties to describe and scale the systematic errors resulted from uncertainty of firmware size, delay of circuits, propagation of ultrasonic, and etc. Therefore, calibration of TM-UFM is one of the complications to obtain accurate measurement, except for precise transit-time detection and improved transducers and circuits. This paper proposes a novel data fusion method for TM-UFM calibration based on particle swarm optimized (PSO) support vector machines (SVM). The parameters of SVM are optimized via PSO method to dissect sources of systematic errors. Besides, extensive experiments have been conducted on a platform of TM-UFM, and the results have illustrated the effectiveness of this method. The PSO-SVM based model leads to a decrease of systematic deviations from ±2% to ±1% (full scale) and improvement of the precision of the TM-UFM, compared with the Gauss-Legendre weight integration method.