Doppler Target Motion Analysis Based On Improved PSO

In practical application topics,passive target motion analysis(TMA)technology has important research significance in the fields of strategy and underwater sound.Traditional methods often have more stringent preconditions and are difficult to solve.Therefore,the target motion analysis can be easily and quickly solved by combining the acoustic features with the target motion analysis.The motion parameter solving method combined with the characteristics of acoustic information is mainly studied.This method does not require the observer to maneuvering.It is possible to passively estimate and track the target by relying on a single observation station.There is a Doppler shift phenomenon in the line spectrum of the radiated noise emitted by the target in motion,so the line spectrum information contains the motion parameter information of the target,using the line spectrum instantaneous frequency information of the extracted radiation noise to solve the target motion parameters is of great significance for passive tracking and analysis.The method of estimating the motion parameters of moving targets by using instantaneous frequency information and azimuth information is studied.The particle swarm optimization(PSO)optimization method is an intelligent optimization method,which has the characteristics of no need for complicated calculation,less parameters to be adjusted,outstanding search ability and simple structure,but there are also cases of premature phenomena.In order to solve these shortcomings of PSO method,based on the classical standard PSO method,the dynamic parameter improvement scheme is given,and a PSO method based on elite-random learning is proposed,which is ERPSO method.In the process of learning the natural biological group,the individual's learning style has randomness and concentration,and the learning model has anti-repetition periodicity.According to this,a periodic learning mode is proposed,and the elite group learning method and the random learning method are combined to make the optimization method can converge quickly,and also in the solving process,diversity of solutionis guaranteed,and reduced the emergence of premature phenomena.Elite group learning method(EPSO)and random learning method(RPSO)are proposed.The EPSO method shares information and guides individual particles by grouping outstanding elite individuals into a group.This avoids the fact that it is easy to be trapped in the local optimum when only the best individual particles are used as a reference,reflecting the learning form between groups.In the RPSO method,in addition to selfexploration to obtain information and experience,particles can also obtain more information and experience through information exchange with other particles,reflecting the form of learning between individuals.Because of the random distribution of learning objectives,the diversity of each particle in the learning process is improved.The numerical simulation experiment analyzes the feasibility and comprehensive performance of the ERPSO method.The ERPSO method can fully utilize the advantages of the local high search ability of the elite group learning method and the global large-scale random search ability of the random learning method,so that the convergence result is more excellent,and in the single-peak and multi-peak objective function optimization problems have satisfactory performance.In the end,the parameter estimation test of the target motion analysis problem under the uniform linear motion model is carried out by numerical simulation experiment,and the Unscented Kalman filter(UKF)technology is used to realize the restoration and prediction of the target position.Through numerical experiments,the effects of measurement error and motion parameters on the solving process are analyzed.The simulation results illustrate the feasibility of the proposed ERPSO method and the motion parameter solving method of moving targets.