A Double Channel Frequency Counter Based on Kalman Filtering and Neural Network Compensation

Abstract

The paper presents the design of a lightweight, wide-range, high-frequency measurement counter. This counter provides a measurement range of 1 Hz–250 MHz on one channel and 200 MHz–8 GHz on the other, enabling broad coverage across high-frequency applications. Measurement data is processed using a Kalman filtering algorithm in conjunction with a back propagation (BP) neural network compensation algorithm, which together enhance the stability and accuracy of the system. Within the input power range of ±12 dBm, the high-frequency channel achieves average relative error of 2.30 × 10^–8, while the low-frequency channel maintains average relative error of 2.22 × 10^–8. Compared to general-purpose frequency counters, the proposed system offers advantages in terms of low cost, compact size, reduced complexity, high accuracy, good stability, and practical usability, effectively meeting actual measurement needs. It has significant potential for applications in high-precision GNSS timing systems.