A data fusion approach based on sequential and strong tracking filters with nonlinear dynamic systems

For the nonlinear multisensor dynamic system, a new nonlinear data fusion algorithm is put forward based on filter step by step and strong tracking filter (STF) in the case of that there exist in mild transmission delay and the delay can happen while the measure data is propagated by local network from each local observation site to central processor. Suppose that the estimate and its covariance is obtained at a certain time, the predict estimate and its covariance at next time can be also acquired by use of all the observations with object state at this time. In the process of implementing we employ sequential (or step by step) filter to update step by step the state estimate by using these ordinal measures, the order may be randomly decided based on their delay. Synchronously, we also employ STF due to worrying about that extend Kalman filter (EKF) can have bad robustness when the theoretic model does not match actual systems and surroundings. Comparing the new algorithm with other similar algorithms, the new algorithm possesses of some larruping functions or capability, such as lower computer burden, preferable robustness, strong flexibility while some data be delayed or lost, and comparable estimate accuracy. Through comparison new algorithm using STF with the traditional fusion algorithms based on EKF via computer simulation, we illustrate to validate these outstanding performances arising from new algorithm