基于神经网络算法的弹丸阻力系数辨识

为了解决传统辨识方法在应用于弹丸气动参数辨识时所存在的建模误差问题,基于极大似然准则,采用神经网络-牛顿法,对无控旋转弹丸的飞行状态数据进行处理,提取出其零升阻力系数。仿真结果表明,该算法具有较高的辨识精度和可靠性。基于实测数据,采用该算法和应用已成熟的扩展卡尔曼滤波算法进行辨识。结果表明,神经网络算法的前期辨识精度较高,但中期误差较大,而扩展卡尔曼滤波的前期精度较差,故可结合二者的优点有效地解决工程实际问题。

Identification of Projectile Drag Coefficient Based onNeural Network Algorithm

In order to solve the modeling error of the traditional identification method applied to the identification of the aerodynamic parameters of projectile,the neural network Newton method was used to deal with the flight state data of uncontrolled spinning projectile based on the maximum likelihood criterion,and the zero lift drag coefficient was extracted. Simulation results show that the algorithm has high identification accuracy and reliability. Based on the measured data,this algorithm and the mature extended Kalman filter algorithm were used for identification. The results show that the accuracy of the neural network algorithm is higher in the early stage,but the mid-term error is larger,while the extended Calman filter has poor accuracy in the earlier stage. The engineering problems can be solved effectively by combining the advantages of both.