Fixed-Structure Methods in Robust Estimation and the Control of Nonlinear and Parameter Varying Systems

本IIA项目研究固定结构方法在非线性和线性参数变化系统的鲁棒估计和控制中的应用。卡尔曼滤波等估计技术在导航、工业过程、电力系统和故障检测与隔离等领域有着广泛的应用。然而，由于缺乏对建模误差的稳健性，标准估计器的性能受到限制，尽管稳健估计理论的某些方面已经得到了发展，但混合结构奇异值(MSSV)理论的能力尚未得到应用。因此，发展基于MSSV的稳健估计技术成为研究热点。考虑了三个性能标准：H2、HoO和L1。这项研究还考虑了非线性系统的固定结构控制设计问题。特别是，发展了用于设计输出反馈控制律和Lyapunov函数的自动化技术，这些函数在平衡点附近产生大的吸引域。将固定结构技术应用于线性变参数系统的控制设计。LPV控制理论的发展为设计非线性系统的增益预定控制律提供了一种更为严密和有效的手段。本研究致力于发展与LPV系统控制设计的两种主要方法相对应的固定结构方法。LPV设计技术被应用于休斯机构和低温中正在考虑的指向问题。

This IIA project deals with fixed-structure methods in robust estimation and control of nonlinear and linear parameter varying systems. Estimation techniques such as the Kalman filter have application in applications such as navigation, industrial processes, power systems and fault detection and isolation. However, the performance of standard estimators is limited due to the lack of robustness to modeling errors, and although some aspects of robust estimation theory have been developed, the power of mixed structured singular value (MSSV) theory has not yet been applied. Hence, research in the development of robust estimation techniques using the MSSV is pursued. Three performance criteria are considered:H2, Hoo, and l1. This research also considers issues in fixed-structure control design for nonlinear systems. In particular, automated techniques for the design of output feedback control laws and Lyapunov functions that yield large domains of attraction about an equilibrium point are developed. Fixed-structure techniques are applied to control design for linear parameter varying (LPV) systems. LPV control theory was developed to provide a more rigorous and effective means of designing gain scheduled control laws for nonlinear systems. This research focuses on the development of fixed- structure methodologies corresponding to the two main approaches to control design for LPV systems. The LPV design techniques are applied to pointing problems which are being considered at the Hughes Mechanisms, and Cryogenics.