基于SRUKF-WN的传感器系统偏差估计算法

The Estimation of Sensor Systematic Biases Based on SRUKF-WN Algorithm

针对系统偏差模型难以构建和保证状态协方差半正定性的问题,提出了平方根无迹卡尔曼滤波(SRUKF)和权值网络(WN)相结合的系统偏差估计算法(SRUKF-WN算法)。首先,构造了包含WN的网络连接权值、系统偏差以及目标量测在内的WN系统偏差估计模型;其次,建立了SRUKF-WN算法状态空间模型,通过SRUKF算法估计最优的网络连接权值;最后,将最优网络连接权值估计值用于WN,进而得到最优系统偏差估计值,并用其配准传感器的量测。仿真结果表明,该SRUKF-WN算法具有较快收敛性、较短算法耗时以及较高的算法精度。

It is difficult to construct systematic biases model and to endure the semi-positive characterization of the system state variance.In order to solve above problems,this paper proposed a systematic biases estimation algorithm(SRUKF-WN algorithm)based on square root unscented Kalman filter(SRUKF)and weight-network(WN).Firstly,a state space model of weight-network(WN)was constructed which include weight network weights,systematic biases,and target measurements.Secondly,this paper utilized the square root unscented Kalman filter(SRUKF)to estimate the optimal weight of WN.Finally,the optimal weight was applied for the WN,and the systematic biases estimation was obtained and used to register the target's measurement.Simulation results showed that the SRUKKWN algorithm had faster convergence,shorter algorithm time and higher algorithm accuracy.