Temperature Compensation and Amplitude Prediction in Ultrasonic Measurement Based on BP Neural Network Mode

In the system that uses ultrasonic to measure the liquid level, the sound velocity of the ultrasonic will be changed due to the difference in temperature, which will affect the reliability of the experimental results. This paper explores the relationship between temperature and sound velocity, and acoustic impedance. Based on that, the temperature effects on the ultrasonic signals are explored. And this paper fits the evaluations to the temperature compensation formula with the ordinary least squares. Then this paper builds a temperature-amplitude prediction model based on the BP neural network using the received ultrasonic echo data. The evaluations show that as the temperature increases, the received echo amplitude decreases. The voltage drops by 10 mV for every 5°C increase in temperature. The measurement accuracy is ±1 °C. Its linear fitting r
²
is 0.99326. The amplitude prediction model built by BP neural network has a prediction accuracy R
²
of 0.96887. The error between the predicted value and the ground truth is less than 4.09%. The research can effectively predict the amplitude of data collected at different temperatures. It provides an effective temperature compensation reference in the experiments based on ultrasonic.