论文部分内容阅读
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 long forecasting accuracy and low precision of projectile falling point forecasting method, this paper proposes a support vector machine (SVM) neural network forecasting method based on Particle Swarm Optimization (PSO). The method uses PSO optimization algorithm to optimize SVM training parameters to get the optimal SVM neural network landing prediction model. On this basis, the Kalman filter is used to process the data of the outer ballistic to form the neural network training data, and the simulation test of the falling point is carried out. The simulation results show that the maximum range error is 7.371m and the maximum range error is 0.886m. The fall time is within 35ms, an order of magnitude faster than the numerical integration method, which provides a way for the practical application of projectile falling forecast .