摘要:

The special stability properties of SISO systems incorporating Smith dead-time compensator controllers are investigated. It is shown that the conventional approach to the design of such systems may lead to practical instabilities. Necessary conditions for practical stability and sufficient conditions guaranteeing certain sensitivity and relative stability properties are stated and proven. The inaccuracies in the models are measured by means of an ' ignorance function '. Sufficient conditions expressed in terms of this function, from which conservative estimates on allowed mismatches in the models may be found, are stated. Illustrative examples demonstrating the main results are presented.

ISSN:0020-7179    

DOI:10.1080/00207178008922900

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

A finite-dimensional asymptotic observer is derived on the basis of a Calcrkin approximation for a class of distributed parameter systems. The systems are described by a partial differential equation of parabolic type. The measured outputs are assumed to be obtained through a finite number of sensors located in the interior. The sensor influence functions are added to the usual basis for the Galerkin approximation and Schmidt's orthogonalization is performed to yield a new basis. The Galerkin approximate solution is sought in terms of this basis. By this procedure the observation spillover problem is overcome. Moreover, in view of the fact that it is generally difficult to obtain the closed-form expressions for the eigenfunctions of the equation, the method is useful for practical substantiation of the observer. The uniform convergence of the Galerkin approximate sequence for the partial differential equation is proved and used to ensure the convergence of the estimated state in a somewhat stronger sense. A numerical example is given which illustrates the power of the method.

ISSN:0020-7179    

DOI:10.1080/00207178008922901

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

The describing function method is commonly used for the prediction of limit cycles in non-linear systems. This method is based upon the representation of signals by a Fourier series, and assigning a gain to the non-linear component derived from its response to a sinusoidal input, In this paper we show that a different approach is feasible which in some respects is the dual of the describing function method. The signals are analysed into a Walsh scries, and a gain is assigned to the linear component of the system derived from its response to a square, wave. In many instances the method is likely to give more accurate predictions of limit cycling.

ISSN:0020-7179    

DOI:10.1080/00207178008922902

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

Active control of large flexible structures in space requires the control of a very large dimensional system with one of substantially lower dimension. Modern modal control (MMC) has been developed previously to deal with this active structure problem ; however, control and observation spillover (as well as modelling error) can seriously degrade the controller performance by causing pole-shifts from the desired closed-loop locations. In this paper, the original MMC is enhanced by the addition of an innovations feedthrough term to the control law and a residual mode aggregation term to the state estimator. Under conditions presented the enhanced MMC can eliminate the observation spillover from critical residual modes ; hence, the spillover-induced pole shifting in these modes can be mitigated without changing the design of the original MMC. As shown, the trade-off is that the number of residual modes that can bo so enhanced is directly related to the number of available sensors

ISSN:0020-7179    

DOI:10.1080/00207178008922903

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

This paper describes an approach to maximum likelihood estimation which is based upon recursive hypothesis testing. The advantage of the method over existing hill-climbing and approximate recursive methods lies in the degree of information which can be obtained from the algorithm. In particular, cross-sections through the loss function are provided in a graphical form. This allows the analyst to see the likelihood function evolve, and make subjective decisions based upon this information. Effectively, the method computes all the relevant points on the loss function surface, making the technique ideal for short data runs where conditioning of the likelihood function is a problem. The method compares favourably with alternative maximum Likelihood algorithms in terms of computational effort and has none of their convergence problems.

ISSN:0020-7179    

DOI:10.1080/00207178008922904

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

A means is provided to assess the value or ‘ cost& rsqquo; of each component of a. large scale system, when the total cost is a quadratic function. Such a ‘ cost decomposition ’ of the system has several important uses. When the components represent physical subsystems which can fail, the ‘ component cost ’ is useful in failure mode analysis. When tiie components represent mathematical equations which may be truncated, the ‘ component cost ’ becomes a criterion for model truncation. In this latter event component costs provide a mechanism by which the specific quadratic control objectives dictate which components should bo retained in the model reduction process. This information can be valuable in model reduction and decentralized control problems.

ISSN:0020-7179    

DOI:10.1080/00207178008910039

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

Applications of the matrix sign function to solving Lyapunov and Riccati equations and to other systems theory calculations are reviewed and the spectral implications of its definition discussed, Tho disadvantages of reduced order formulations are presented. Several theorems related to the mapping of eigenvalues under accelerated sign function algorithms are developed together with a consistent metric for measuring the distance of eigenvalues from their ultimate destination. After listing the desirable properties of an accelerated map, several previous attempts at acceleration are analysed and the fundamental reasons for their degree of success or otherwise determined. An optimally accelerated method and another rapidly convergent one are then developed. All of the methods are compared on two 28 x 28 examples arising from a Lyapunov and a Riccati equation. Finally, global and higher order convergence of the three successfully accelorated algorithms are proved.

ISSN:0020-7179    

DOI:10.1080/00207178008910040

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

Output feedback with a series compensator is considered for controlling SISO discrete systems. The MSE criterion of design is used in tracking a given input sequence. The plant can be unstable and/or non-minimum phase, while the compensator is constrained to be physically realizable and the closed-loop configuration is both stable and physically realizable. Using variational methods the required compensator is found, and shown to be the necessary and sufficient optimal solution.

ISSN:0020-7179    

DOI:10.1080/00207178008910041

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

The paper deals with the problem of minimizing the gain-bandwidth of a class of sampled systems. The sampling process of a continuous linear system yields in general a sampled linear system having non-minimum phase features, Such systems have inherently limited gain-band widths The maximum gain-bandwidth is found for a class of sampled systems.

ISSN:0020-7179    

DOI:10.1080/00207178008910042

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor

摘要:

A simple algorithm for dual control using the SAFER concept has been developed for a single-input single-output, linear discrete system with random parameters. A critical feature of this SAFER algorithm is the use of a minimal number of sensitivity functions which captures the probing properties of all the sensitivity functions required for estimation purposes. This results in significant reduction of computation and a feasible and implementable estimator controller with dual properties. Numerical simulations wore carried out to demonstrate the superiority of this new algorithm with dual properties.

ISSN:0020-7179    

DOI:10.1080/00207178008910043

出版商:Taylor & Francis Group    

年代:1980

数据来源: Taylor