改进的粒子群算法在弹体分类中的应用

以沈阳军区某部“超声波弹体内部成份探测系统”为课题背景，利用低能量超声对弹体进行探测，并基于神经网络对捕获数据进行分析，以此来确定废弹的具体类型，提出一种混沌变异粒子群优化算法（CMPSO）。该方法将神经网络的参数优化和权重优化融合入一个统一的框架之中，充分利用粒子群算法寻优能力强、收敛速度快的特点。相对于一般的神经网络结构优化算法，具有设置参数少、计算复杂度低的特点，最后将提出的算法应用于弹体分类问题之中取得了十分显著的效果。

Improved particle swarm algorithm in application of bullet classification

Shenyang military area “ultrasonic projectile interior component detection system” as the subject background, low energy ultrasound is used to detect the bullet. Based on neural network to capture data, so as to determine the specific types of waste, a kind of Chaotic Mutation Particle Swarm Optimization algorithm（CMPSO） is put forward. The parameters of the neural network optimization and weight optimization are made into a unified framework, making full use of particle swarm algorithm optimization ability and fast convergence rate of characteristics. Relative to the general neural network structure optimization algorithm, parameters are less and computation complexity is easier. Finally the algorithm is applied to the bullet classification problems and gets good effect.