A novel method for improving the long-term stability of inertial devices based on model prediction

Abstract

The long-term stability (LTS) of inertial devices refers to their ability to maintain consistent performance parameters over a long period. For inertial devices with poor LTS, there is a significant difference between the actual values of their performance parameters and the originally calibrated values, which severely restricts their measurement accuracy. The LTS improvement methods based on hardware structure are technically difficult, time-consuming, and costly; while those based on accelerated test are prone to device damage and have high testing costs. Therefore, a model prediction-based method for improving the LTS of inertial devices is proposed. By analyzing the internal and external factors that affect the LTS of inertial devices, the time-varying mechanism of their performance parameters is revealed, and the Wiener process suitable for describing the time-varying characteristics of performance parameters is obtained. Furthermore, a Wiener process model identification method is proposed, and a novel online prediction scheme for inertial device performance parameters is designed. Experimental results show that the proposed method reduces the error of inertial device performance parameters by 70.00 %–88.42 %, and shortens the stability period of inertial measurement unit (IMU) from 6-9 months to 2–3 months, significantly improving the accuracy and LTS of IMU.