Application of Stabilized Numerical Integration Method in Acceleration Sensor Data Processing

It is necessary to perform a numerical or a quadratic integral calculation on the acceleration sensor data by rectangular integration, trapezoidal integration or Simpson integration, etc. to obtain the velocity or the displacement result in engineering practice. Since poles of transfer functions of these numerical integration systems are on the unit circle, when acceleration signals containing interference factors, such as random noise, are integrated, integration results are often divergent. Therefore, velocity and displacement results will have serious drift, which affects the accuracy of the follow-up attitude calculation and target positioning. In order to overcome acceleration integration drift, a method based on the stabilized numerical integration to obtain velocity and displacement is proposed in this paper, which makes improvements of the traditional numerical integration. The proposed method obtains accurate velocity and displacement through the acceleration integration by constructing a stable system to modify the transfer function of the integration system, and utilizes the characteristics of the stability system to effectively suppress the drift of the acceleration data multiple integration results. Experimental results show that compared with traditional acceleration integration methods, the quadratic integral displacement peak error, difference error, and absolute error of the proposed stabilized numerical integration method are 0.059, 0.0619 and 0.0084, respectively, which demonstrate that the proposed method is feasible that it can reduce the integration results drift error and improve the accuracy of the integration results.

Automated summary

In this paper, it is discussed how to obtain accurate velocity and displacement results through numerical or quadratic integral calculation of acceleration sensor data, with a focus on overcoming the drift issue during integration by proposing a method based on stabilized numerical integration. This proposed method effectively suppresses the drift of acceleration data multiple integration results and improves the accuracy of the integration results.