# Compensation of Temperature-Induced Bias Drift in Hemispherical Resonator Gyroscopes: An Inherent Data-Driven Architecture

## TL;DR

This paper proposes a data-driven method to reduce temperature-induced bias drift in hemispherical resonator gyroscopes, improving their stability and reliability.

## Contribution

A novel temperature drift compensation method using KAN neural networks and internal operational parameters is introduced.

## Key findings

- The proposed model reduces HRG output stability from 0.022°/h to 0.013°/h.
- Stability improves further, decreasing from 1.1392°/h to 0.0651°/h.
- The method effectively predicts and compensates for hysteresis phenomena.

## Abstract

To address the bias drift problem and hysteresis phenomenon of hemispherical resonator gyroscope (HRG) under temperature change, a temperature drift compensation method based on internal parameters is proposed. The influence model of zero-rate output bias is established with the parameters such as resonance frequency, driving signal amplitude and quadrature suppression voltage amplitude during HRG operation. The temperature cycle experiment is carried out in the range of −20 to 60 °C, and the relationship between internal parameters and working temperature is revealed. Using KAN neural network combined with time series data as input features, a real-time compensation model is designed to effectively improve the prediction accuracy of hysteresis phenomenon. The experimental results show that the model significantly reduces the output stability performance of HRG, from 0.022°/h to 0.013°/h, and the stability decreases from 1.1392°/h to 0.0651°/h, which improves the stability and reliability of HRG.