基于粒子群优化支持向量机神经网络的弹丸落点预报

Impact-point Prediction Based on Particle Swarm OptimizationSVM Neural Network

针对目前弹丸落点预报方法预报时间较长和精度不高的问题,提出了基于粒子群(PSO)优化的支持向量机(SVM)神经网络预测方法。该方法采用PSO优化算法优化SVM训练参数,以获得最优SVM神经网络落点预测模型。在此基础上,使用卡尔曼滤波处理外弹道数据形成神经网络训练数据,进行落点预报仿真测试。仿真结果表明,射程最大误差为7.371m,横偏最大误差为0.886m;落点预报时间在35ms之内,比数值积分法快了一个数量级,为弹丸落点预报的实际应用提供了一种途径。

Aiming at the problem of the shortage of predicting a projectile impact-point quickly and precisely , this paper introduced the forecasting method based on Particle Swarm Optimization support vector machine neu-ral network. In order to obtain the optimal support vector machine neural network prediction model of impact- point, particle swarm optimization algorithm is used to optimize the training parameters of support vector ma-chine. And then,integrated the exterior ballistic data using the Kalman filtering into the training data of neural network for the impact-point prediction simulation test. Simulation results show that the maximum range error of the method is 7. 371m, and the maximum partial navigation error is 0. 886m; The forecast time of impact-point is within 35ms which is faster than the numerical integration method in an order of magnitude. Therefore, this method provides a road for the practical application of the projectile impact point prediction.