Signal Processing Method Based on First-Order Derivative and Multifeature Parameters Combined With Reference Curve for GWRLG

Abstract

Guided wave radar level gauge (GWRLG) is widely used to measure level in industrial production. For the signal processing methods commonly adopted of GWRLG, there are some defects in the identification of the surface echo and determination of the positioning point for timing, which results in the measurement being susceptible to noise, having a large blind area and low accuracy. Therefore, a novel signal processing method is proposed to deal with the echo profile. First, the echo profile is calculated by the first-order derivative to highlight its features. Accordingly, the feature parameters are computed and assembled in different ways to constitute judgment conditions for the surface echo. The reference curve is then combined to remove spurious echoes and obtain the surface echo. The point with the minimum slope is selected on the surface echo as the positioning point for timing so as to achieve good linearity. Moving average filtering, interpolation on the positioning point, and median average filtering are utilized to improve the measurement accuracy. A signal processing system is developed to realize the real-time measurement, and the water level calibration and measurement experiments are performed. In the experiments, a three-point calibration scheme is proposed. The developed system reduces the blind area from 25–30 to 10–15 cm and solves the problem that the surface echo cannot be identified due to its small amplitude for common GWRLG. The actual measurement error is less than 1 cm in the case that the theoretical error is 2.790–4.185 cm, and the measurement results are satisfactory for industrial applications.