High-Resolution Radar Target Recognition Based On PSO-RBF Neural Network

Radar automatic target detection is an important development direction of modern radar, which is widely used in civil and military fields. High Resolution Range Profile of the target (HRRP) geometry information can provide the target along the radar ray direction, and compared with the two-dimensional imaging and3D imaging, HRRP is easier to access. In this paper, the radar automatic target detection technique based on HRRP is studied, the main contents are as follows:1. Research on HRRP and its sensitive characteristics, from the point of view of the target scattering center model. Scatter model by using four kinds of target, target High Resolution Range Profile is simulated (HRRP).2. Aiming at the question of the end of the PSO algorithm is easy to fall into local optimum and fixed weight in the algorithm to search for the loss of the diversity of particles leads to the convergence speed slow. An improved PSO algorithm has been proposed. The algorithm is based on the PSO algorithm, firstly, the adaptive strategy is adopted to adjust the inertia weight PSO algorithm to solve the nonlinear problem of the optimizing ability. Secondly, the local search operator a is added, which refers to the particles in the update location, in the speed range again random search. If the new position of search to the fitness location is better than the current fitness, the new position will be the next evolutionary origin. Otherwise, the original position is still at the next evolutionary origin. The performance and the improved PSO algorithm are validated through simulation experiments.3. A kind of improved PSO optimization algorithm based on RBF neural network is presented to avoid falling into local optimal. The algorithm used in the subtractive clustering algorithm based on RBF neural network to determine the number of hidden layer nodes. The PSO algorithm and the gradient descent method combined parameter optimization on RBF. And the performance of the algorithm is verified through simulation experiment of Hermit polynomial.4. A kind of radar target detection RBF neural network model based on improved PSO is presented. To validate the performance of the model, the measured results of simulation data of four targets and three targets as the experimental data on the performance of the model was trained and tested. The experimental data show that, the model is obviously improved in the detection rate, but also improve the convergence speed, anti-interference ability has been improved significantly.