空间斯特林制冷机系统模型辨识与预测控制

Model Identification and Predictive Control of the Space Stirling Cryocooler System

为解决空间用大功率机械式斯特林制冷机系统的高精度控温问题,针对这一类具有时延特性的大时间常数温度控制系统,首先,文章提出一种基于继电模型辨识的闭环辨识方法,通过设计合理的辨识参数,采集系统的输入输出响应数据并通过最小二乘法拟合获得了系统的辨识模型,克服了传统开环辨识方法无法获得固定控温点附近辨识模型的不足;其次,基于辨识模型的阶跃响应数据建立预测模型,采用动态矩阵预测控制,通过滚动优化和反馈校正机制,对温度控制系统性能指标实现了最优控制。最后,通过搭建模型辨识及实验平台,验证了系统模型辨识的准确性可达到94%,对长波红外探测器60K控温点的控温稳定性可达到±10mK(3σ)的水平,该模型辨识与控制方法已经成功应用于多台星载红外相机的焦面制冷机控温系统并获得在轨验证。

To solve the problem of high-precision temperature control of high-power stirling cryocooler system for space application,towards such a temperature control system with time delay and large time constant characteristics,firstly,a closed-loop identification method based on relay model identification is proposed,by selecting the proper parameters,the system model can be identified through the least square method based on the obtained input and output response data,which overcomes the shortcomings of the open-loop identification method that cannot obtain the model at the neighborhood of the fixed operation point;secondly,the dynamic matrix predictive control is employed based on the step response of the identified model,through the rolling optimization and feedback correction,the optimized control performance can be achieved for the temperature control system.Finally,through establishing the model identification and experimental test bench,it can be seen that the similarity of system model identification reached 94%and the stability of the temperature for the 60K operation point can achieve a high level at±10mK(3σ),and these achievements on the model identification and control method have been successfully applied to the temperature control system of focal plane refrigerator for several spaceborne infrared cameras and have been verified in orbit.