Determination of Allan Variance Coefficients Using Genetic Algorithm

Anil Sami Önen and Bagis Altinöz

Peer Reviewed

Abstract: Inertial measurement units that are composed of gyroscopes and accelerometers are widely used in inertial navigation while determining the attitude and position of an aerial vehicle. The deterministic error parameters of the inertial sensors are calibrated prior to the mission in order not to have performance degradation. In addition to the deterministic errors, the stochastic errors are quite important for inertial sensors that are especially used for critically sensitive applications. In analyzing stochastic errors, the very well-known Allan variance method is used. Although the method is quite effective to find the error sources and plot the output variation in a log-log scale, while finding each error contributor an extra effort is needed. In this study, an alternative method for finding the error parameters is proposed; that is the genetic algorithm based error identification. Unlike the traditional method which generally depends on the slope matching (e.g. slope of -1/2 for angular random walk) this alternative gives the ease of analyzing Allan Variance result and finding the error parameters in a much shorter time. After verifying the proposed method with the outputs of a constructed stochastic error model, the outputs of various gyroscopes and accelerometers of different grades are analyzed, and associated stochastic error parameters are estimated. The results are compared with the traditional method and the specifications of the units that are supplied by the manufacturers.
Published in: 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS)
April 20 - 23, 2020
Hilton Portland Downtown
Portland, Oregon
Pages: 723 - 727
Cite this article: Önen, Anil Sami, Altinöz, Bagis, "Determination of Allan Variance Coefficients Using Genetic Algorithm," 2020 IEEE/ION Position, Location and Navigation Symposium (PLANS), Portland, Oregon, April 2020, pp. 723-727.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In