基于模型辨识的CMOS图像传感器热电制冷系统滑模控制

Model Recognition Based CMOS Image Sensor Sliding Mode Control of Thermoelectric Refrigeration System

为了提高CMOS图像传感器热电制冷系统的工作性能,解决CMOS图像传感器工作时由于积热导致的暗电流问题,提出了基于模型辨识的CMOS图像传感器热电制冷系统滑模控制方法。分析了CMOS图像传感器热电制冷系统的传热机理,进而建立了CMOS图像传感器热电制冷系统在工作点附近的控制模型。针对辨识出的模型在不同工作点参数变化较大的问题,设计了对CMOS图像传感器热电制冷系统参数变化不敏感的基于指数趋近律的滑模控制器,以提高系统的鲁棒性。利用Matlab/Simulink对系统进行仿真分析,仿真结果表明:相比于传统的PID控制算法,滑模控制方法的动态响应和稳态精度均更优。最后,在实际的CMOS图像传感器热电制冷控制系统中进行了实验测试,结果表明:室温20℃的条件下,输入不同的温度指令滑模控制算法的稳态误差均小于PID控制算法,为±0. 1℃,满足CMOS图像传感器热电制冷控制系统的温控精度的要求。

In order to improve the performance of CMOS image sensor thermoelectric refrigeration system and solve the dark current problem caused by accumulated heat when CMOS image sensor is working,a sliding mode control method for CMOS image sensor thermoelectric refrigeration system based on model identification is proposed. The heat transfer mechanism of the CMOS image sensor thermoelectric refrigeration system is analyzed, and the control model of the thermoelectric refrigeration system near the working point is established. Aiming at the problem that the identified model has large changes in parameters at different working points,a sliding mode controller based on exponential approach law which is insensitive to system parameter changes is designed to improve the robustness of the system. The simulation analysis of the system is carried out by using Matlab/Simulink. The simulation results show that the dynamic response and steady-state accuracy of the sliding mode control method are better than the traditional PID control algorithm. Finally,the experimental tests were carried out in the actual thermoelectric refrigeration system. The results show that the steady-state error of the input temperature command sliding mode control algorithm is less than the PID control algorithm at room temperature 20 ℃,which is ± 0. 1 ℃,which satisfies the image sensor. The temperature control accuracy requirements of the refrigeration system.