摘要
为提高雷达系统目标识别能力,对粒子群算法及RBF神经网络进行了分析。针对离子群算法(PSO)易陷入局部极小的缺陷,提出了基于自适应时变权重和局部搜索算子的改进PSO算法,并将该算法应用到RBF神经网络核函数参数的优化学习中,进行了雷达目标识别仿真实验。仿真结果表明,相对于标准PSO-RBF神经网络,改进算法不仅收敛速度快,且误差精度高,特别在干扰较强时,目标的识别率有较大提高。
In order to improve the ability of radar target recognition, particle swarm algorithm and ra- dial basis function（RBF） neural network are analyzed. As particle swarm optimization （PSO） algorithm is liable to trap in local minimum, an improved PSO neural network parameter selection of kernel function, algorithm is presented, which is applied to RBF and the radar target recognition simulation experi- merit is conducted. The Matlab simulation results show that, compared to the standard PSO-RBF neural network, the improved algorithm converges faster, has higher error precision, and target recognition rate is improved, especially in the strong disturbance
出处
《现代防御技术》
北大核心
2014年第5期115-120,共6页
Modern Defence Technology
关键词
一维距离像
粒子群算法
RBF神经网络
雷达目标识别
range profile
particle swarm optimization （ PSO ）
radial basis function （ RBF ） neuralnetwork
radar target recognition
