Research On Particle Filter Tracking Algorithm Based On Artificial Intelligence Algorithm

With the development of shipping technology and people's living standard,the navigation of ships at sea becomes more and more important.In order to ensure the safety of ships and reduce the probability of accidents,higher requirements are put forward for ship tracking.The traditional tracking algorithm has low detection accuracy in the case of non-linear and non-Gaussian.Particle filter tracking algorithm is based on Monte Carlo algorithm and recursive Bayesian theory,and has its unique advantages for target tracking in non-linear and non-Gaussian state.Therefore,it is widely used.However,the traditional particle filter algorithm still has some shortcomings.Some scholars have studied and used artificial intelligence algorithm to optimize the particle position and resampling stage to improve the accuracy of the algorithm.This paper mainly studies and discusses the application of two intelligent algorithms to different phases of particle filter algorithm,including particle swarm optimization algorithm applied to importance sampling to improve sampling particle accuracy and genetic algorithm used to replace resampling to improve particle diversity to improve tracking accuracy and efficiency.In addition,it is supplemented by adaptive adjustment to reduce the occurrence of local optimization through adaptive real-time adjustment of key parameters and thresholds in the algorithm to ensure algorithm accuracy and realize real-time monitoring and adjustment of tracking process to ensure tracking accuracy.The main work is as follows:Firstly,the basic concepts and principles of standard particle filter algorithm and artificial intelligence algorithm are introduced,and Bayesian estimation and Monte Carlo algorithm are discussed.The steps of standard particle filter algorithm,particle swarm optimization algorithm and genetic algorithm are explained,and the related contents of self-adaptation are analyzed,which lays a theoretical foundation for subsequent work.Secondly,particle swarm optimization is applied to the important sampling stage of particle filter algorithm,which optimizes the problem of insufficient accuracy of the sampled particle position.A particle filter tracking algorithm based on adaptive particle swarm optimization is proposed.Firstly,the adjustment method of the velocity and position of the primary particles is discussed.The orientation and velocity of the particle position adjustment are determined by the fitness value,that is,the particle weight.Keep particles closer to the real position to improve sampling accuracy.Secondly,the adaptive adjustment method is discussed.At the same time,the inertia weight and learning factor in particle swarm optimization are adjusted adaptively to balance the searching and moving ability of particles so as to reduce the occurrence of local optimization and improve the accuracy of particles.Simulation results show that the improved position of sampling particles improves the accuracy of particle filter algorithm.Finally,genetic algorithm is used to replace the resampling phase of particle filter algorithm to reduce the loss of particle diversity caused by this.An adaptive particle filter target tracking algorithm based on artificial intelligence algorithm is proposed.In the resampling stage of traditional particle filter,diversity is lost due to the elimination of small weight particles.Therefore,this chapter discusses how to improve the weight of particles through crossover and mutation in genetic algorithm to ensure the diversity of particles,and how to increase the diversity by keeping the particles whose weight increases after genetic algorithm to exceed the threshold so that the number of particles does not decrease too much.At the same time,the parameters are adjusted adaptively,and the resampling threshold and cross mutation probability are adjusted in real time.When the detection accuracy is abnormal,the adaptive threshold can also be adjusted to reduce the number of abandoned particles and ensure the diversity of particles..The simulation results show that the accuracy of the whole algorithm is improved and the effectiveness of the algorithm is verified.