Application Of PSO Optimizing WNN In Target Tracking

With the development of military technology development,the target itself movement and the complexity of the environment bring an enormous challenge to the target tracking.In order to solve the problem of target stability and precision tracking in motor cases.In this paper, particle swarm optimization to wavelet neural network used in target tracking the "current" statistical model,by the Monte Carlo simulation show that the algorithm can improve the tracking performance.Based on the study of the particle swarm algorithm, an improvement aimed for dealing with the shortcoming of standard particle swarm algorithm as the later with the loss of population diversity, slow convergence speed,extremely premature convergence are proposed.When the particles get into premature convergence,give the particles that can't jump out of the local optimum a certain disturbance,which can effectively avoid the premature convergence. Simulation results show that the improved particle swarm algorithm reduces the number of interations and effectively avoid premature convergence problem. Secondly, the wavelet neural network is analyzed in details, wavelet neural network has better convergence speed and approximation accuracy than BP neural network.However, neural network gradient descent method is often used to train network parameters which is easy to fall into the local minimum values and has slowly convergence speed.And the rest of the particle remain unchanged,in subsequent iteration such particles are divided into two parts,which not only remain the global optimal particle of each generation,but also maintain the diversity of the particle population. In this paper, using the improved particle swarm optimization to train the wavelet neural network parameters. Simulation experiment show that the improved wavelet neural network reduces the number of iterations, improves the convergence precision and has better prediction performance.The selection of maneuvering target model is a key issue in target tracking.In establishing motor model,it is necessary to comply with the actual situation of maneuvering and to facilitate the mathematical processing. Based on the analysis of the CV model, CA model, Singer model, the "current" statistical model, and based on the MATLAB simulation environment, using Monte Carlo simulation to compare the tracking accuracy of each model. Do a more in-depth research about the the "current" statistical model, analyse the influence of the motor frequency on the "current" statistical model and the principle of adaptive adjustment the frequency of motor maneuver. Using the particle swarm algorithm to optimize the wavelet neural network for the "current" statistical model,and adaptive adjustment motor frequency,the simulation results show that this method can not only improve the network training speed, but also improve the tracking accuracy,enhance the stability of the system.