Research On Prediction Of Filtering Algorithm For Systems With Delayed Measurements

In the actual target tracking applications, time-delay phenomena caused by thecharacteristics of the sensor often exists. However, time delay is the main reasonleading the stability of the system to be worse. The time delay effects on the targettracking precision obviously, it often leads to poor accuracy of target tracking. Atpresent, there are increasingly high requirements on the target tracking accuracy ofproduction, life and even military. Therefore, the research on time delay has a greatvalue.Filtering algorithm is an important part of target tracking which has greatinfluence on the tracking accuracy. Kalman filter is a mature and effective filteringalgorithm. But for systems with delayed measurements, the traditional Kalman filteralgorithms cannot be applied directly. The key problem is the observation equationsin the form of time-delay system model does not meet the realization form of Kalmanfilter algorithm.Therefore, basing on the system state augmentation methods, this paper proposesa method which converts the observation equation to an observation equation with nodelay. At the same time, according to the principle of Kalman filter algorithm,aiming at the model equation of no delay system after conversion, the observed delayKalman filter algorithm steps are given. The proposed observed delay Kalmanfiltering algorithm can be able to solve the problem of prediction filtering for linearsystems with delayed measurements, and the proposed EKF and UKF algorithm canbe used for nonlinear systems with delayed measurements.The algorithm proposed in this paper are applied to the actual nonlinearmeasure electro-optical tracking system, comparing the performance of algorithm.The results of simulation experiment prove that the method of converting delayobservation equation to the observation equation without delay is feasible, and the accuracy of algorithm is relatively high.