Search

Application of Model Reference Adaptive Control (MRAC) to vehicle active suspension subsystem is investigated. The MRAC objective is to automatically self-tune the controller of the active suspension such that disturbance and vibration of a vehicle is reduced to a level which is determined by an "ideal" conceptual suspension system (a reference model). An MRAC vehicle suspension can thus accommodate for variances in sprung load and suspension component characteristics. An MRAC was designed using Lyapunov stability approach and was tested using different scenarios such as ±10% to ±30% variances of suspension components and sprung load. It is shown that the MRAC suspension closely followed the performance of the prespecified reference model and achieve significant improvements over passive suspension. Real-Time Simulation and Animation (RTSA) software was developed to provide an excellent visual aid for understanding and interpreting the better ride performance achieved by the MRAC suspension.

The problem of vibration isolation is investigated from the standpoint of modern control and optimization theory. The proposed optimal model-following suspension design was verified experimentally by means of a microcomputerized suspension model. The experimental results are very encouraging and indicate promising potential application of the proposed scheme to real-world systems. This paper describes the modeling and formulation of an optimal control suspension reference model, and the implementation algorithm of the proposed microcomputerized optimal model-following suspension scheme.

This paper presents a unified treatment to the problem of optimal feedback control of discrete-time linear shift-invariant systems using state feedback, output feedback, dynamical output feedback, and observer-controller feedback. A class of generalized optimal feedback controllers is proposed as a basis for obtaining existing results as well as for deriving new results such as the interesting min-max feedback controllers. The essence of the generalized results lies in the minimization of a class of generalized performance measures subject to a flexible controller configuration. The stability aspect of the generalized optimal feedback controlled system is addressed.

The theory of optimal output feedback control is extended to a class of quadratic performance measures whose weighting matrices are functions of frequency. The incorporation of frequency-dependent weighting matrices allows one to emphasise or de-emphasise the importance of the variables being penalised over specific bands of frequencies. Optimization of the frequency-shaped performance measure yields optimal compensators that can be more meaningful for certain minimisation problems. Results are presented for different configurations of frequency-shaping feedback control. An example is presented to illustrate the usefulness of the design technique.

This paper describes a technique for modeling and identifying a class of nonlinear servomechanism systems with stick-slip friction. The physics of the stick-slip friction is considered in modeling the process. Identification of the system parameters is formulated as a nonlinear optimization problem. A modified simplex algorithm is proposed as the optimization procedure. The difficulties encountered in choosing identification algorithm and input signals for the problem are discussed. A simulation example of a servomotor system is provided.

The linear quadratic design of an optimal robust servo–controller for a continuous–time control system is described. It introduces a servo–control performance measure which accommodates the internal model principle. The measure selectively discounts penalties on control effort of desirable frequencies and provides complete flexibility in the selection of weighting matrices. The proposed servo–controller uses plant state feedback, signal state feedforward and a servo–compensator for ensuring robust asymptotic command tracking and disturbance rejection. An informal alternate proof of the internal model principle in the state variable domain is presented using an operator-transformation technique. Where necessary, observers are used to complement the servo-controller. The proposed optimal robust servo–controller yields the expected superior performance in terms of response and error minimization. An illustrative example is given.

The reaction mechanism of DNA topoisomerase II (TOP2) involves a covalent double-strand break intermediate in which the enzyme is coupled to DNA via a 5′-phosphotyrosyl bond. This normally transient enzyme-bridged break is stabilised by drugs such as mitoxantrone, mAMSA, etoposide, doxorubicin, epirubicin and idarubicin, which are referred to as TOP2 poisons. Removal of topoisomerase II by the proteasome is involved in the repair of these lesions. In K562 cells, inhibiting the proteasome with MG132 significantly potentiated the growth inhibition by these six drugs that target topoisomerase II, and the highest level of potentiation was observed with mitoxantrone. Mitoxantrone also showed the greatest potentiation by MG132 in three Nalm 6 cell lines with differing levels of TOP2A or TOP2B. Mitoxantrone was also potentiated by the clinically used proteasome inhibitor PS341 (Velcade). We have also shown that proteasome inhibition with MG132 in K562 cells reduces the rate of removal of mitoxantrone or etoposide stabilised topoisomerase complexes from DNA, suggesting a possible mechanism for the potentiation of topoisomerase II drugs by proteasomal inhibition.

A class of general structured discrete-time deterministic observers is developed. The one-step predicting or Luenberger observers and the current-update observers may be obtained from this class of general structured observers. Interesting relationships and important properties among various full-order observers are established. Optimal discrete-time observer gains are then derived, and it is shown that the optimal observers can be structually and numerically equivalent to various forms of Kalman filters.

This paper extends the familiar 1-D concepts of observer designs to the design of observers for 2-D systems described by Roesser's model. Both full-order and reduced-order observers are considered. Extensions of 1-D concepts to 2-D is non-trivial in view of the requirement that state transformations for 2-D systems have to be block diagonal in order to preserve their input-output properties. Another issue addressed in this paper is the required asymptotic stability property of the 2-D observers. To maintain tractability in asymptotic stability analysis, we consider the class of 2-D observers with separable characteristic polynomials. Illustrative examples are provided.

The modeling, analysis and design of a ball-balancing control system is considered. The control system is constructed and an optimal hybrid controller capable of absorbing external disturbance is fabricated. The controller design is based on the theory of disturbance-accommodating control, optimal control and digital delayed-data observer. Experimental results presented are useful for demonstrating the practicalilty aspects of the above theory.

The design and implementation of a controller for a six-degree-off-freedom robot manipulator is presented. Two IBM Personal Computers were used to implement the preliminary scheme reported here where execution of programmed motion can be carried out in either point-to-point or straight line mode.

The purpose of this paper is to utilize a class of compensated parameter identifiers for generating bias-free parameter estimates, in the presence of noise disturbance which accounts for the uncertainties in the system. A forward feedback is used to minimize the effect of noise. The observer form of state variable modeling is used to synthesize a ladder circuit whereby the compensated parameter identifiers are asymptotically stable, with added rapid convergence. A new technique is applied to derive state feedback that assures the assignment of closed-loop eigenvalues in arbitrary sectors within the unit circle.

The potential of a new adaptive control for reactive power sources is investigated. In the large disturbance regimes, the adaptive signals supplement the nominal, thyristor activated, reactors and capacitors in manipulating the network voltage. Convergence of the scheme, and its relationship to the inherent system nonlinearities is considered.