基于液态神经网络的超声波飞行时间动态校准算法

打开文本图片集

DOI：10.16652/j.issn.1004-373x.2026.10.002

中图分类号：TN911.7-34；TH814

文献标识码：A

文章编号：1004-373X（2026）10-0007-09

Ultrasonic time-of-flight dynamic calibration algorithm based on liquid neural network

Wang Jiangjie， Zhou Junyang， Qi Hao， Jing Shaojie， Zhu Binglei， Zhang Kai

（College of Metrology Measurement & Instrument， China Jiliang University， Hangzhou ， China）

Abstract： An ultrasonic flowmeter based on liquid neural network （LNN） is designed innovatively to solve the problem of insufficient accuracy and poor robustness of traditional flow measurement methods in complex fluid dynamics environment. With its dynamic time series modeling ability and adaptive learning of nonlinear features， LNN can effectively analyze the propagation time difference of ultrasonic signals in the fluid， and overcome the influence of noise interference， fluid turbulence and temperature changes. By constructing lightweight LNN model and combining with multi-path ultrasonic signal feature extraction and time series prediction， the system can realize high precision calculation for the flow velocity and flow rate. The experimental results show that， in comparison with the traditional static neural network model， the measurement error of the proposed method is reduced to less than ±0.5% in the dynamic fluid scene， and the response delay is less than 50 ms， which significantly enhances the environmental adaptability and real-time performance of the system. This research can provide a more reliable intelligent sensing solution for industrial process control， energy metering and other fields， and validate the potential application value of LNN in physical signal processing.

Keywords： liquid neural network; ultrasonic flowmeter; flow measurement; dynamic time series modeling; signal feature extraction; time series prediction

0 引言

超声波流量计凭借高精度、无接触式、耐腐蚀等优点，在石油、化工、制药、食品加工等工业领域得到广泛应用。（剩余11680字）