Research On Signal Processing Methods And High-precision Attitude And Velocity Algorithms Of Strapdown Inertial Navigation Systems Under Highly Dynamic Environments

Strapdown inertial navigation system(SINS) is an integrated science that spans several precision subject areas. Due to the inertial sensor error and the integration computing of the strapdown navigation algorithm, the small errors of navigation algorithm and sensor output are accumulated. The error accumulation will cause the navigation error increase with time. To satisfy the high-precision requirements in practice, the inertial sensor precision and measuring range should be improved, and the inertial sensor error should be reduced by modeling and compensation methods. On the other hand, the performance and precision of the SINS algorithms in highly dynamic environments, such as in coning, drastically vibration and sculling environments should also be improved to match the precision of the inertial sensor and satisfy the practical requirements for the SINS.The key techniques on high-precision SINS signal processing and attitude/velocity algorithms in highly dynamic environments are studied in this paper. The studies include three contents: 1) modeling and analyzing methods of the gyro’s random error; 2)high-precision attitude algorithm in a dynamic environment; 3)high-precision velocity algorithm in a dynamic environment. The studies are aimed at improving the accuracy of SINS by modeling and processing of inertial sensor error and the optimization of the navigation algorithms.The gyro performance has a significant influence on SINS. Modeling and compensating of the gyro error are always a key technique to improve the SINS precision. Time series modeling methods of gyro random noise are further studied in this paper. Adaptive/robust Kaman filter is used in the AR/ARMA modeling of the gyro random noise. Unknown steady estimator/unknown time-varying estimator of observation noise is used to achieve the statistical features estimates of the observation noise. The proposed methods can estimate the AR/ARMA model parameters at the most economical cost. So the sampling number and test time are reduced. The kalman filter also can correct the model parameters by the new samples. The estimated model parameters can follow the characteristic variation of the gyro random noise timely. To solve the problem of slow convergence speed in MA modeling method by using the traditional method of inverse correlation function, the Gevers-Wouters method and innovation recursive estimator method are used to model the FOG random noise. These two new modeling methods can effectively accelerate the convergence speed of the MA parameters estimation.In SINS algorithms, the performance of attitude algorithm is very important. The existing traditional strapdown attitude algorithms are all based on the first-order rotation vector equation. In general case these attitude algorithms can work satisfactorily. But in some highly dynamic environments, for example in coning, drastically vibration and sculling environments, the performance of the traditional attitude algorithms still needs to be improved. The paper first proves a proposition proposed in the derivation process of the traditional coning algorithm. The proof perfects the derivation of the coning algorithm. Then the error constitution of the coning algorithm is analyzed. The analysis results indicate that the coning algorithm error is composed of two parts: drift error and approximation error.The approximation error is comparable to the drift error for the most general case and can not be neglected. A second-order coning algorithm which based on second-order truth-value of the rotation vector is developed. The proposed coning algorithm supports two kind of gyro output: angular increment and angular rate. The proposed coning algorithm can reduce the approximation error effectively and improves the accuracy of the attitude algorithm greatly without increasing sample numbers. The performance of the rotation vector algorithms under nonconing environment(vibration and maneuvers) is analyzed and a second-order optimal rotation vector algorithm which has a best performance under vibration is developed.Besides attitude algorithm, velocity algorithm is another important factor of the SINS algorithms. Sculling motion is usually used as a standard input to evaluate the performance of the velocity algorithm in highly dynamic environments. Based on the various combinations of inertial sensors outputs in SINS, two new sculling algorithms which are based on incremental angular/specific force input and angular rate/incremental velocity input are proposed in this paper. The new sculling algorithms have best performance in a sculling environment. By the new algorithms, velocity can be calculated directly from the inertial sensor output. Hence they are easily applied to the SINS that have a gyro output of incremental angular and an accelerometer output of specific force, or a gyro output of angular rate and an accelerometer output of incremental velocity.The studies can be used to improve the precision of the SINS in highly dynamic environments. Experimental data have been demonstrated by the experimental results. Therefore these studies have important engineering values and widely application foreground.