Research On ALS Noise Estimation And Its Application For Dynamic Measurement Of Accelerometer

Noise adaptive estimation algorithm is an effective way to achieve adaptive Kalman filter, which mainly includes Bayesian, maximum likelihood, covariance matching, correlation and Autocovariance Least Squares (ALS) methods. In which the ALS noise adaptive estimation algorithm has low computation, and whose covariance estimates are unbiased and has higher accuracy. But when the prior information of system noise or measurement noise is biased, the accuracy of ALS estimates will be declined, and in addition Gaussian white noise restricts the application field of ALS algorithm. In accelerometer dynamic measurement system, cascading dynamic compensator can extend acceleratormeter frequency band, and reduce the frequency errors caused by signal attenuation. But after adding dynamic compensator, the measurement system will have frequency noise disturbance whose prior statistics are unknown. How to choose an effective filtering algorithm is the key to filter noise disturbance. Adaptive Kalman filtering algorithm based on ALS noise estimation algorithm is an effective filtering method. Therefore, it has important significance and engineering practical value to research on ALS noise estimation algorithm its application for dynamic measurement of accelerometer.In this paper, an Autocovariance Total Least-squares (ATLS) noise estimation algorithm was proposed based on the study and analysis of ALS noise estimation algorithm, which uses Total Least-squares (TLS) to solve innovation autocovariance liner equation instead of Least-squares (LS); extend the application field of ALS noise estimation algorithm, propose a colored noise estimation algorithm based on ALS noise estimation algorithm, which achieves colored noise variances estimation without the non-Gaussian white noise model; The adaptive Kalman filter based on ALS noise estimation algorithm is used to filter the noise of compensator output in acceleratormeter dynamic measurement system.The experimental results verify the correctness and effectiveness of ATLS noise estimation algorithm. When the priori error of noise is larger, ATLS noise estimation algorithm has higher estimation accuracy than ALS noise estimation algorithm. The proposed colored noise estimation algorithm based on ALS extends the application field of ALS noise estimation algorithm, which can realize the estimation of colored noise variances without the non-Gaussian white noise model. The adaptive Kalman filter based on ALS noise estimation algorithm is applied to the dynamic compensator of accelerometer, which can effectively improve dynamic measurement accuracy of accelerometer.