火箭炮交流伺服系统模型参考模糊神经网络位置自适应控制

Adaptive Rocket Launcher Position Servo System Based on Model Reference Fuzzy Neural Network

针对多管火箭炮发射时恶劣的负载特性,设计了一种模型参考模糊神经网络自适应位置控制器。利用RBF网络作为辨识器,实现对被控对象的Jacobian信息辨识,用梯度下降法实时修正模糊控制器的输入输出隶属度参数,以使模型参考模糊神经网络能根据火箭炮跟踪发射过程中的负载特性实时调整速度给定值,从而减小火箭炮发射过程中系统参数变化和外部干扰的影响。仿真和实验结果表明,该方法可有效提高火箭炮位置伺服系统的动态响应能力和稳定性,并使系统具有很强的鲁棒性。

Considering the atrocious load property when the rocket is launched,an adaptive position controller based on model reference fuzzy neural network is designed.RBF network is used to recognize the Jacobian information of the object controlled.The parameters of input and output membership function are modified online by a gradient method to make the model reference fuzzy neural network adjust the given speed in real time according to the load property of rocket launcher.Thus the influence of the change of system parameters and external disturbance on the rocket launcher position servo system could be minified.Simulation results show that this method could improve the stability and anti-disturbance ability of rocket launcher position servo system effectively.