This proposal describes a project to explore a new area of adaptive
control and to address open problems of urgent relevance to theory as well as
applications: adaptive control of systems consisting of a linear part and a nonsmooth nonlinear input-output characteristic being in
either an actuator or a sensor. Typical examples of such nonlinear
characteristics are dead-zone, backlash and hysteresis. An adaptive inverse
approach is proposed to control such systems to meet desired performance
specifications, which exploits an adaptive inverse for an unknown nonlinearity
and a linear controller structure nonadaptive/adaptive
for a known/unknown linear part. Choices of nonlinear models, design of
adaptive inverse control algorithms, stability, convergence and robustness
analysis, and applications will be investigated. The results of this research
will provide new tools to handle unknown nonsmooth
nonlinearities which are common in practical control systems.
This proposal describes a university-industry collaborative research
project to explore a new area of adaptive control: adaptive control of
sandwich nonlinear systems ,
and to solve some long-standing and wide-range control problems of urgent
relevance to theory as well as applications. The proposed research will focus
on adaptive control of sandwich systems with linear and nonsmooth
nonlinear dynamics and on adaptive control of two-layer systems with smooth and
nonsmooth nonlinear dynamics. Typical nonsmooth nonlinear characteristics are dead-zone,
backlash, hysteresis, other piecewise-linearities as
well as frictions which are the main sources of component imperfections in
control systems. The proposed adaptive inverse control approach employs an
adaptive inverse to cancel the nonlinearity effects in order to achieve system
performance improvements. This approach points to a new direction to design
control systems using a new algorithm-based technology, which, after a period
of learning or adaptation, can recognize component imperfections and compensate
for their harmful effects. With such adaptive controllers, the component
specifications could be greatly relaxed, their cost reduced, and their
reliability increased. The results of this research will advance the knowledge
of adaptive control significantly, provide new tools to effectively handle
practical nonlinearities which have haunted the constructors of control systems
for many years, and have many applications in defense and civil industries in
which high-precision control systems are vital components.
This proposal describes a research project to develop new adaptive
failure compensation techniques for dynamic systems with uncertain failures.
The proposed research is focused on the development of a novel systematic theoretical
framework for adaptive failure compensation and specific solutions for several
synergic topics, to provide guidelines for designing control systems with
guaranteed stability and tracking performance in the presence of system
parameter, dynamics and failure uncertainties, with applications to
performance-critical control systems. New theories of nonlinear and
multivariable adaptive control, new approaches for system modeling in the
presence of system failures, and new methods of adaptive failure compensation
will be explored for new advances in this open area of research.
The first topic is the development of novel system modeling and adaptive
control approaches for systems with failures. For many applications, models of
systems with failure and without failures are essentially different (for
example, aircraft flight dynamics in an engine differential mode). We will
develop novel system models which capture the key features of dynamic systems
in the presence of failures, based which effective failure compensation schemes
can be designed. The second topic is the development of adaptive failure
compensation schemes for multivariable systems with space structure vibration
reduction control applications. The third topic is adaptive compensation of failures
in cooperating multiple manipulator systems. New
controller parametrization and adaptive laws are needed for intelligent
autonomous robot control systems which can adaptively compensate for uncertain
failures. The fourth topic is control of systems with MEM devices as actuators
which may fail during system operation. Effective compensation of failures of
MEM devices is a key component of successful MEMS technology and this research
is to develop such techniques illustrated by control of morphing actuators and
synthetic jet actuators applied to aircraft flight control. The unified theme
of these topics is failure compensation by direct adaptation of controller
parameters without explicit fault detection and diagnosis, aimed at achieving
fast response and effective compensation of uncertain failures. The unique
feature of adaptive failure compensation is that it ensures both stability
and asymptotic tracking, without the knowledge of when, how much and how
many failures appearing in the system. The importance of this research is its
potential for significantly improving control system performance in the
presence of uncertain failures for performance-critical applications.
Intellectual merit: The proposed activities has
high intellectual merit. Adaptive failure compensation has open issues such as
failure induced parameter/structure uncertainties, system failure
compensability, controller adaptivity to uncertain
failures, system stabilizability under multiple
failure patterns, and advanced applications, which are both important and
challenging in theory and practice as well. Those issues contribute to the
unique features of the control problems investigated in the project, and their
solutions will lead to creative concepts and effective methods for fields of
systems and control. This research will develop novel solutions to such issues,
which will advance the state-of-the-art in adaptive control theory and emerging
applications such as MEM technology, safe aircraft and intelligent robot
systems. Preliminary study has shown encouraging results of this promising
adaptive compensation approach.
Broader impacts: This research will have major impact on technology as
it will develop novel system modeling and adaptive control techniques for
aircraft flight systems, intelligent robot systems, active vibration control
systems, and for control of systems with MEM devices such as morphing actuators
and synthetic jet actuators, with uncertainty adaptation and failure
compensation capacities to improve system reliability, maintainability and
survivability. Impact on education will be strong as the research activities
and results will bring new concepts and theory of adaptive control into student
training and knowledge dissemination. Impact on outreach will be broad as the
proposed adaptive failure compensation techniques have attracted academic and
industrial/government researchers such as NASA and Air Force.
This research project is to solve open control
theoretical problems to build technical foundations for designing resilient
control systems which are capable of maintaining desired performance in the
presence of uncertain system faults such as actuator failures, structural
damage and sensor failures. This research is aimed at developing new fault
detection algorithms and new adaptive and robust control methods and criteria
to handle large and multiple system fault uncertainties. It will create new
resilient control theory and techniques applicable to performance-critical
systems such as aircraft, spacecraft, wind turbines, jet engines and
intelligent robots, to enhance their safety under uncertain fault conditions.
It will conduct new studies on modeling and control of systems under various
fault conditions for which existing control designs are not applicable. For
example, aircraft loss-of-control precursor conditions such as airframe damage,
component failures, icing and turbulence effects may cause large and variant
system uncertainties for which existing feedback controllers are not powerful
enough. The research is expected to advance feedback control theory and
technology for the need of emerging applications which require control systems
to be resilient to faults, that is, have desired capabilities to accommodate
uncertain and large system faults.
This research studies unique metrics of control system resilience, distinct
features of resilient control problems, and key control theory requirements of
performance-critical systems. It develops new control theory and design
techniques to ensure desired control system resilience for multivariable
nonlinear systems under uncertain multi-fault conditions. It solves new control
problems such as fault detection for unstable systems, control of systems with
uncertain structural characterizations or non-parametrizable
non-canonical form nonlinear dynamics, adaptive and fault-tolerant control of
systems with underactuation or nonminimum
phase. These problems have the key feature that the controlled systems have
fault-induced parametric, structural and functional uncertainties, difficult
for most existing control schemes to deal with (to ensure both system stability
and asymptotic tracking). For example, there can be uncertain failure patterns,
uncertain underactuation, uncertain system infinite
zero structures, uncertain dynamic variations, caused by uncertain faults.
A main goal of this research is to build a new control system design framework
with specific control schemes which are capable of dealing with such unsolved
system uncertainty problems, needed for resilient control systems technology.
Several new feedback control methods will be developed, including: adaptive
multi-layer multiple-model design (to ensure both fault handling and
performance improvement abilities), adaptive multi-design integration (to deal
with multiple faults), adaptive feedback-based fault detection (to have
self-stabilization capacity), adaptive structural uncertainty accommodation (to
deal large system structural damage), and characteristic parametrization based
adaptive approximation control for non-canonical form nonlinear systems (to
design an adaptable and stable controller structure). Direct control adaptation
techniques will be developed for fast and smooth compensation of large and
multiple fault uncertainties. New resilient control designs, for
performance-guarantee fault-tolerant control, will be tested on some benchmark
application system models. New concepts, theory and techniques for resilient
control systems will be used for student training and knowledge dissemination.
Novel outcomes expected from this research include: a new resilient control
theoretical framework with new solutions to some key technical problems; a new
direct adaptive multi-layer multiple-model control method for uncertain MIMO
systems; a new adaptive feedback-based fault detection scheme with guaranteed
stable detection conditions; a new adaptive multi-design integration based
method for multi-fault accommodation; new fault detection and fault-tolerant
control designs for aircraft, wind turbine, jet engine and robot system models;
and scalable fault accommodation designs applicable to multiple (combined and sequentially
or recurrently occurred) uncertain faults.