N.E. Wu

Robust feedback design with optimized diagnostic performance

This paper is concerned with the design of reconfigurable control laws that also effectively facilitate the acquisition of information used for diagnostic purposes in fault tolerant control systems. A control performance index and a diagnostic performance index are chosen to indicate the overall performance. A simple relationship between the two indexes is established that elucidates the control/diagnostic tradeoff. A control design method is developed that first finds a set of controllers achieving a certain control performance level and then uses the remaining degrees of freedom in the controller set to achieve an optimized diagnostic performance level.

Reliability of fault tolerant control systems: Part II

For Part I see ibid. 40th IEEE Conf. Proceedings on Decision and Control (2001). Reliability properties peculiar to fault-tolerant control systems are emphasized. The paper proposes to formulate the fault tolerant control problem as an optimization problem that maximizes coverage of failures through redundancy management. Coverage modeling is attempted in a way that captures its dependence on the control performance and on the diagnostic resolution. Under the proposed redundancy management policy, it is shown that an enhanced overall system reliability can be achieved with a control law of a superior robustness, with an estimator of a higher resolution, and with a control performance requirement of a lesser stringency.

Aperture error mitigation via local-state estimation for frequency-based emitter location

This paper considers the problem of locating a stationary coherent emitter via a single moving platform making frequency measurements in the presence of aperture state uncertainty. It is shown that the estimated emitter location is most sensitive to the receiving aperture velocity uncertainty. The required aperture velocity accuracy is determined through a noninfinitesimal perturbation analysis. A solution to location accuracy enhancement with a minimal hardware addition is attempted. It is shown that this can be achieved by mounting a high-resolution tri-axis microelectromechanical systems (MEMS) accelerometer at the aperture to measure its velocity, which can deviate significantly from that estimated by the on-board navigation system. The Doppler shifts of the GPS signal carrier frequency, whenever it can be acquired through the aperture, are also considered as a way to aid the aperture velocity measurement. A decentralized, federated processing method for the aperture velocity estimate referenced at the aperture, integrating all measurement data, is presented. An upper bound for the error of aperture velocity estimate is derived. The potential for significant accuracy enhancement for emitter location is demonstrated.

Reliability prediction for self-repairing flight control systems

By using aerodynamically redundant control surfaces, a flight control system can be reconfigured in the event of an actuator failure and/or surface damage. This reliability feature is usually implemented through the use of some impairment detection, isolation and accommodation (IDIA) schemes. A case study is presented for a self-repairing flight control system (SRFCS), for which the probability of loss of control is computed under varying levels of hardware redundancy and varying values of IDIA coverage. Given the conditions peculiar to surface damage and control reconfiguration requirements in our application, a simple formula for obtaining accurate reliability predictions without employing Markov chain models is derived. As a result of the study, we are able to recommend how much performance improvement is needed in designing the SRFCS, and where the hardware redundancy level can be further reduced.

Discrete Fourier transform and H/sub infinity / approximation

It is shown that uniform rational approximation of nonrational transfer functions can always be obtained by means of the discrete Fourier transform (DFT) as long as such approximants exist. Based on this fact, it is permissible to apply the fast Fourier transform (FFT) algorithm in carrying out rational approximations without being apprehensive of convergence. The DFT is used to obtain traditional approximations for transfer functions of infinite-dimensional systems. Justification is provided for using the DFT in such approximations. It is established that whenever a stable transfer function can be approximated uniformly on the right half-plane by a rational function, its approximants can always be recognized by means of a DFT. >

Failure sensitizing reconfigurable control design

The design of reconfigurable controls for fault-tolerant control systems is considered. The issue of performance robustness during the systems normal operation versus failure sensitivity at the time of a system component failure sensitivity at the time of a system component failure is addressed from the control design point of view. A criterion is proposed that balances the competing interests. The existence of an optimized solution under the criterion is proved. An example involving aircraft surface impairment is given to explain the procedure for obtaining the optimized design, and to demonstrate enhanced overall system performance achieved by the design under the proposed criterion.<<ETX>>

Reconfigurable control design: achieving stability robustness and failure tracking

This paper is concerned with the design of reconfigurable controls for fault tolerant control systems. The task is to seek for reconfigurable control laws that also effectively facilitate the acquisition of information used for diagnostic purposes. The overall control system performance is measured by a control performance index and a diagnostic performance index. A set of controllers is first designed to achieve a certain control performance level. Then an optimized diagnostic performance level is achieved by using the remaining degree of freedom in the controller. Simple relationships can be established between the two performance indices, which can be used as a guideline to reiterate the design. Technical discussion of the paper is focused on the selection of control criterion that guarantees robust stability in the presence of failures, and on the selection of locations where diagnostic signals are acquired.<<ETX>>

Feedback minimax synthesis for distributed systems

A procedure for feedback controller synthesis in a loop containing a distributed-parameter system under a frequency-domain minimax criterion is suggested. An explicit iterative algorithm allows the computation of the optimal controller. The controlled system is assumed to be possibly of a non-minimum-phase unstable distributed-parameter type. Through application of approximation theory in a frequency-domain setting and the theory of indirect coprime factorization, a practical procedure for detailing the optimal feedback controller is found.<<ETX>>

Supervisory Control of a Database Unit

To effectively enhance service availability, this paper proposes a redundancy configuration for a database unit residing in a command and control (C2) system that supports air operations. The results of modeling, supervisory control, and performance analysis of the database unit are presented. The unit is modeled as a closed Markovian queuing network. State variable feedback is used to implement the functions of restoration and routing upon the identification of the failure of one of the database servers in the unit. Several control policies are evaluated in terms of the resulting mean time to unit failure, the steady state availability, the expected response time, and the service overhead of the database unit.

Reliability of reconfigurable control systems: a fuzzy set theoretic perspective

From a fuzzy set theoretic perspective, the paper analyzes the relationship of overall system reliability to performance requirements on subsystem modules in a reconfigurable control system. A performance measure is introduced, accommodating the multiple impairment parameter situation, in the form of coverage that indicates the likelihood of success in a reconfiguration action. This definition of coverage can also be used as a fuzzy ranking criterion for making the decision in selecting the most reliable control law.