Improved Algorithm of Autocovariance Least-Squares Noise Estimation

Noise estimation is the foundation for the application of Kalman filtering theory.The noise estimation results from the conventional autocovariance least-squares(ALS) are usually non-positive definite.For this purpose,an improved algorithm of ALS(IALS) is proposed to overcome effectively the problems of insufficient data and inaccurate priori information,and consequently to get positive definite noise estimates with better accuracy.Numerical simulation results validate the correctness of IALS.