基于神经网络逆模型辨识的舵机云台控制  

Steering gear control based on neural network inverse model identification

作　　者：吴岸平 侯中喜[1] 王鹏[1] 郭天豪 

机构地区：[1]国防科学技术大学航天科学与工程学院,湖南长沙410073

出　　处：《华中科技大学学报（自然科学版）》2013年第S1期13-16,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

摘　　要：设计了一种小型舵机云台的控制方法,它利用无人机上已有的传感器信息,结合目标点的GPS坐标,计算得到云台指向所需的摄像机俯仰角和方位角,采用舵机作为执行机构,完成对目标点的跟踪观测.针对舵机执行控制指令时的延时问题,提出了基于神经网络的逆模型辨识方法,利用通过外围测量设备得到的系统输入/输出时间序列,采用离线方法对神经网络进行训练,得到系统的逆模型,从而根据期望输出修正系统的控制输入.仿真结果表明:该方法能有效辨识出系统的逆模型,对控制输入修正后能基本消除舵机环节的延时,从而提高云台指向精度.A control method for small steering gear was designed,sensor information available on unmanned aerial vehicle(UAV)and global position system(GPS position of the object point was used to calculate the pitch angle and yaw angle of the camera,steering engine was used as the actuator,target tracking and observation was then completed.An inverse model identification method based on neural network was proposed to solve the delay problem which occurs when the steering gear was executing the control instruction,the system input and output time series which were obtained by the external measurement equipment were used for the off-line training of the neural network,then the inverse model of the system was got to revise the control input according to the desired output.Simulation results show that this method can identify the inverse model effectively,the delay of the steering gear can be eliminated with the revising of the control input,which can raise the pointing accuracy of the steering gear.

关 键 词：无人机 舵机云台 目标跟踪 指向控制 神经网络 逆模型辨识 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] V227[自动化与计算机技术—控制科学与工程]