Method And Application Of State Estimation Of Trajectories Based On Probability Model

With the rapid development of science and technology and the continuous improvement of military,civil,medical and other needs,combine new technologies effectively,develop and improve the state estimation theory,improve the state estimation accuracy and efficiency of moving objects,Which is of great significance to the development and application of state estimation theory.However,classical state estimation algorithms,such as Kalman filter algorithm and some of its improved algorithms,applicable to linear or weakly nonlinear system,which greatly limits the application of these algorithms in practice.Therefore,this paper mainly based on the probability model,improved the original algorithm,proposed new state estimation method for complex nonlinear systems.For the classical estimation algorithms,the convergence and the estimation accuracy are proved in detail respectively.Through the simulation experiments in different systems,the comparison of the advantages and disadvantages of various methods as well as demonstration of the applicability of each method.It is found from the experimental results that the estimation accuracy of the extended Kalman filter in the strong nonlinear system and high-dimensional system is obviously reduced,and because the is an uncertain sampling algorithm,it also applies to strong non-linear systems,but the running time of the particle filter is too long,which reduces the estimation efficiency of the algorithm.As a deterministic sampling algorithm,the unscented Kalman filtering algorithm mainly uses a series of sigma points with weight to approximate the posterior distribution of the state vector of nonlinear system.This method only captures the first two moments of the posterior distribution,based on this,the third-order moments are also captured by improving the sampling method.The experimental results show that the estimation accuracy of this method is obviously improved compared with the standard unscented Kalman filter,and the same is true for complex multimodal systems.The essence of state estimation is a method which use observations,the state equation and the observation equation of the system to get the state value.Based on this,in this paper,we take the observation as the initial sequence,and use the gradient descent method to obtain the state value which satisfies the equation of state by iteration.Experiments show that the method has good estimation accuracy.In Bayesian filtering,when the priori distribution is known and determined,a simple algorithm is proposed,that is,based on the idea of conjugate distribution,the closed-form solution of the posterior distribution can be obtained directly.In addition,several maneuver models are introduced,combine the practical problems and algorithms effectively and illustrate the application of the state estimation algorithm in target tracking by experiments.