基于多维泰勒网的运载火箭鲁棒自适应控制

Robust Adaptive Control of Launch Vehicle Based on Multi-dimensional Taylor Network

针对运载火箭主动段飞行过程中的参数时变及存在弹性振动、液体推进剂晃动和风干扰等不确定性和非线性问题,提出了一种基于多维泰勒网(MTN)的直接鲁棒自适应控制方案。基于鲁棒自适应控制技术,通过多维泰勒网可以逼近任一非线性函数的性质,实时逼近时变参数模型的理想控制输出。对比通过实验的方式获得某些特征时刻运行参数,再根据稳定裕度设计控制器的传统方案,该MTN方案具有成本小、控制精度高等优点。采用MATLAB/Simulink仿真技术搭建运载火箭全量仿真模型,通过额定实验和鲁棒实验,综合分析、比较了传统PD控制器和多维泰勒网控制器的控制性能。仿真结果表明,多维泰勒网鲁棒自适应控制器具有更好的控制效果。

Aiming at the time-varying parameters and the existence of such uncertainty as elastic vibration,liquid propellant sloshing and wind interference during the active flight phase of the launch vehicle,a direct robust adaptive control scheme based on multi-dimensional Taylor network(MTN) is proposed in this paper.Based on the robust adaptive control technology,the ideal control output of the time-varying parameter model can be approximated in real time by the properties that any nonlinear function can be approximated through MTN.Compared with the traditional scheme that operating parameters at some characteristic time are obtained through experiments and then the controller is designed according to the stability margin,this MTN scheme has the advantages of low cost and high control accuracy.MATLAB/Simulink simulation technology is used to build the full launch vehicle model,and the control performances of the traditional PD controller and MTN one are comprehensively analyzed and compared through rated and robust experiments.