【作者】 徐正平；

【导师】 葛文奇；

【作者基本信息】 中国科学院研究生院（长春光学精密机械与物理研究所） ， 机械电子工程， 2011， 博士

【摘要】 高空飞行平台搭载的侦察仪器生存能力强、安全性高;与卫星平台相比,对同样焦距的侦察设备可获得更高分辨率的地面目标图像,且机动性较强。本文以某高空轻型可见光相机的像面扫描系统为研究对像,对其视场拼接方法、系统负载非平衡特性及其稳速控制策略进行了深入研究。首先给出了国内外高空飞行器及侦察相机的发展现状,指出了现有视场拼接方法中机械拼接和光学拼接的优缺点,提出了采用凸轮驱动的动态扫描拼接方法。凸轮结构的特殊性、平台姿态角和相机位角造成系统负载力矩的非平衡特性,对此进行了理论分析,结果表明:作用于电机轴上的负载力矩与平台横滚角和相机位角之和(记为ε角)及凸轮转速成正比。当ε角为零时,从动件匀速段负载力矩为零,而从动件变速段负载力矩发生变化;当ε角非零时,负载力矩在凸轮旋转一周范围内均非零。负载力矩特性的分析为后续控制策略的选取提供了理论依据。ε角为零时,为保证凸轮转速在从动件变速段平滑过渡,分别尝试PD控制、传统超前滞后稳速控制、超前滞后多模控制及神经网络多模控制,并进行了实验验证。实验结果表明:PD控制、超前滞后多模控制及神经网络多模控制均可有效抑制从动件变速段负载力矩对凸轮转速的影响,且从动件匀速段凸轮转速稳速精度优于0.5%(均方根值)。ε角非零时,负载力矩在凸轮升程段和回程段表现出不同的特性,从而造成凸轮转速在两阶段存在转速降。为解决此问题,提出三种控制策略:电流速度双环控制、改变凸轮升程段或回程段校正环节放大倍数或系统输入信号。从工程实现角度考虑,最终选取改变凸轮升程段或回程段系统输入信号的方法应用于实际工程。实验结果表明:当ε角非零时,在从动件匀速段凸轮转速稳速精度优于0.7%(均方根值)。在ε角为零及非零时,控制系统性能指标均能满足要求。最后对相机进行了成像实验,所得图像曝光均匀、层次分明、清晰无像移,符合预期要求。更多还原

【Abstract】 The reconnaissance instrument on the altitude-flight platform has strong viability and high security. Compared with secondary planet, the reconnaissance instrument with same focus could gain much higher resolution image, and it is also much more flexible. The image scanning system of one high altitude visible-light camera was chosen to be researching object, and the assembly method, the non-equilibrium load feature and relative control methods are discussed in detail.The actuality of the domestic or overseas high altitude aerocraft and reconnaissance camera was given first, and also the strongpoint and shortcoming of the mechanical assembly and optical assembly were pointed out. Then one dynamic scanning assembly actuated by cam was introduced. Some factors, such as the special structure of the cam, platform attitudes angles and camera azimuth cause the non-equilibrium feature of the system load torque, and it was analyzed in theory. In order to predigest depiction, the sums of platform roll angle and camera azimuth is marked to be angleε. The analysis result shows that load torque on the electromotor axis is directly proportional toεand cam speed. Whenεis zero, the load torque equals zero in the uniform section of the follower and it changes in the speed section of the follower. But whenεis not zero, the situation is different. The load torque doesn’t equal zero in the whole circle rotating track of the cam. The academic analysis of the load torque feature provides foundation for choosing control method.Whenεis zero, the methods, PD control, traditional lead and log correction, lead and log multi-model control and neural network multi-model control were used to ensure cam rotation speed transiting smoothly in the speed section of the follower. Some experiments were made to verify the methods stated above. Experimental result shows that the methods, PD control, lead and log multi-model control and neural network multi-model control could restrain the cam rotate fluctuating caused by the load torque in the speed section of the follower. At the same time, in the uniform section of the follower, the rotate speed precision of the cam is less than 0.5 %( r.m.s.).Whenεisn’t zero, the load torque has different feature in the lift section and return section of the cam. Those different features cause that the cam rotate speed isn’t same in the lift section and return section. In order to solve this problem, three control methods were induced, which are current-speed double loop control, changing the DC gain of the adjuster or the input signal in the cam lift section or return section. Considering the project realization, the third method was used in the project at last. The experimental result shows that whenεisn’t zero, the rotate speed precision of the cam is less than 0.7 %( r.m.s.) in the uniform section of the follower.Whetherεis or isn’t zero, the control system performance could satisfy the whole system. The imaging experiment was made to the camera and the gained image exposure evenly and has distinct gradations. At the same time, the image is clear and has no image motion. The image scanning system complies with expectative requirement.更多还原