A New Norm-Observed Calibration Method Based on Improved Differential Evolution Algorithm for SINS

Abstract

It is vital for a strapdown inertial navigation system (SINS) to be calibrated before normal use. In this paper, a new kind of norm-observed calibration method is proposed. Considering that the norm of the output of accelerometers and gyroscopes can be exactly the norm of local acceleration of gravity and Earth rotation angular velocity, respectively, optimization function about all-parameter calibration and the corresponding 24-position calibration path is established. Differential evolutionary algorithm (DE) is supposed to be the best option in parameter identification due to its strong search and fast convergence abilities. However, the high-dimensional individual vector from calibration error equations restrains the algorithm’s optimum speed and accuracy. To overcome this drawback, improved DE (IDE) optimization is specially designed: First, current “DE/rand/1” and “DE/current-to-best/1” mutation strategies are combined as one with complementary advantages and overall balance during the whole optimization process. Next, with the increase of the evolutionary generation, the mutation factor can adjust itself according to the convergence situation. Multiple identification tests prove that our IDE optimization has rapid convergence and high repeatability. Besides, certain motivation of external angular velocity is added to the gyroscope calibration, and a series of dynamic observation paths is formed, further improving the optimization accuracy. The final static navigation experiment shows that SINS with calibration parameters solved by IDE has better performance over other identification methods, which further explains that our novel method is more accurate and reliable in parameter identification.