On-line Learning and Control of Partially-Known and Unknown Nonlinear Systems

当线性时不变（LTI）系统的控制已经达到很高的成熟度时，非线性系统的控制仍然非常未被探索。非线性设计和分析技术的最新进展，以及快速低成本计算机和复杂软件工具的广泛使用，为非线性系统控制设计的成功发展提供了强大的潜力。在本提案中，由主要研究人员为一类非线性系统开发的初步控制设计用于激励一系列研究任务，这些研究任务将导致控制未知非线性系统的一般方法的发展。所提出的控制设计采用鲁棒自适应和非线性控制的思想以及神经系统和模糊系统的函数逼近结果来满足当前控制方法无法满足的非线性对象的控制目标。鲁棒自适应控制修改。退步、调谐函数、非线性阻尼和控制李雅普诺夫函数是一些设计技术要考虑在拟议的工作。分析工具，如李亚普诺夫理论，泛函分析和函数逼近将与计算机模拟一起使用，以建立和演示所开发的控制方法的特性。所提出的研究的预期结果是开发一种方法来控制大量未知或部分已知的非线性植物，这些植物无法使用当前的控制方法进行控制。

ECS-9877193IoannouWhile the control of linear time invariant (LTI) systems has reached a high level of maturity the control of nonlinear systems is still very much unexplored. Recent advances in nonlinear design and analysis techniques together with the widespread use of fast low cost computers and sophisticated software tools offer a strong potential for the development of successful control designs for nonlinear systems.In this proposal, a preliminary control design developed by the Principal Investigators for a class of nonlinear systems is used to motivate a series of research tasks that will lead to the development of a general methodology for controlling unknown nonlinear systems. The proposed control design employs ideas from robust adaptive and nonlinear control as well as function approximation results from neural and fuzzy systems to meet control objectives for a nonlinear plant that cannot be met by the current control methods. Robust adaptive control modifications. backstepping, tuning functions, nonlinear damping and control Lyapunov functions are some of the design techniques to be considered in the proposed work. Analytical tools such as Lyapunov theory, functional analysis and function approximations will be used together with computer simulations to establish and demonstrate the properties of the developed control methodologies.The expected outcome of the proposed research is the development of a methodology for controlling a wide class of unknown or partially known nonlinear plants that cannot be controlled using the current control methods.