Target state estimation method for single radar linear track line

Abstract

The invention belongs to the technical field of sensor target tracking and data fusion, and specifically relates to a target state estimation method for a single radar linear track line in a complex data environment with unknown system noise and observation noise. According to the method, weighted estimation of the vertical distance relative to the measurement time is carried out on X, Y components of the current limited measurement points by making full use of a characteristic that the target is in the state of uniform linear motion, and then the final target state estimation (position, speedand course) result is determined according to the obtained component linear track line model parameters. The comparative experiment shows that the method not only is easy to implement in engineering,but also has a better target state estimation effect than the traditional Kalman filtering method.

Description

technical field [0001] The invention belongs to the technical field of sensor target tracking and data fusion, and in particular relates to a method for estimating the state of a target for a single radar straight-line track in a complex data environment with unknown system noise and observation noise. Background technique [0002] Target tracking refers to the time-series discrete observation data of moving targets (aircraft, tanks, ships, etc.) obtained by sensors (radar, sonar, infrared, etc.), Methods for estimation and tracking. Kalman filtering is a classic optimal linear unbiased estimation method in the application of target tracking and data fusion. When the target state space and measurement space are both linear Gaussian systems, it can minimize the variance of the target motion state in the background of interference noise. Estimation and tracking, and it is a recursive estimation method, which performs filter update according to the observation data at each mom...

AI Technical Summary

Problems solved by technology

[0004] 1. The application conditions of the traditional Kalman filter are relatively harsh, and there is a contradiction between the complex application data environment in the actual system

In the actual system, the measurement data sampling interval is relatively large, measurement errors, coordinate conversion errors, and data calibration and transmission errors exist at the same time

The Kalman filter requires the system model to be accurate and the system error model and the observation error model to be known, which is difficult to meet in practical applications, or in the process of system work, any error component model changes, which will lead to traditional Karman Mann filtering method diverges or reduces accuracy

[0005] 2. The process noise covariance Q and the measurement noise covariance R have difficulty in selecting parameter values, and cannot take into account both steady state and transient

In the traditional Kalman filtering algorithm, the process noise covariance Q and the measurement noise covariance R need to be estimated in advance, and they remain unchanged throughout the filtering process. Therefore, when the two estimated noise covariance matrix parameters are inaccurate, during the filtering process There will be accumulated errors in the filter, causing the filter to diverge

At the same time, when the value of Q is small and the value of R is large, the target state estimation formed after filtering has better stability, but the transient performance becomes worse, which shows that the target position estimation is the time registration point (the target at the same moment The distance between the real position point and the estimated position point) is relatively large, and sometimes it may lag behind a measurement cycle

Conversely, when the value of Q is larger and the value of R is smaller, the estimated state of the target formed after filtering has better transient performance, but poorer steady-state performance. The heading and speed estimates vary greatly, so filtering is meaningless

[0006] 3. Inevitable outliers in system measurement have a serious impact on the work of traditional Kalman filters

Outliers may cause filter divergence or target state estimation to fail in a short period of time, manifested as large sawtooth on the estimated track line, sudden changes in target heading and speed estimation

Images