作者：
P. LöhnbergB. FrankeControl Laboratory
Control Systems and Computer Engineering Group Department of Electrical Engineering University of Twente Enschede The Netherlands
Optimal parameter estimation and experiment design requires a weighting between the cost of parameter errors and that of identification itself. Having identified this need, the general principle to obtain both costs f...
详细信息
Optimal parameter estimation and experiment design requires a weighting between the cost of parameter errors and that of identification itself. Having identified this need, the general principle to obtain both costs for using the model in control design is described. Fast experiment design is achieved by using analytic expressions to calculate the optimal experiment parameters. This is illustrated by a simple example. It is shown that the use of the resulting model criterion in optimal identification is equivalent to cautious stochastic control. A more realistic example is illustrated by simulation.
This paper presents a foundation for an integrated approach to the design of controls and diagnostics in reliable controlsystems. In this approach the control module and diagnostic module of the control system are de...
详细信息
The siandard H ∞ control problems has now been solved, in which some rank conditions are imposed on the plant data. These conditions are sometimes not satisfied in practical applications, and it is desirable to obtai...
详细信息
The siandard H ∞ control problems has now been solved, in which some rank conditions are imposed on the plant data. These conditions are sometimes not satisfied in practical applications, and it is desirable to obtain general solutions to “non-siandard” problems which do not require these conditions. In this paper, we shall derive a parameterization of the H ∞ controller for a non-standard two-block case which include the state feedback control law as a speial case.
An on-line scheme to failure diagnosis is proposed for dynamic systems under adaptive control, which is designed based on a direct approach to self-tuning regulator. Failure modes occurred in the system are assumed to...
详细信息
An on-line scheme to failure diagnosis is proposed for dynamic systems under adaptive control, which is designed based on a direct approach to self-tuning regulator. Failure modes occurred in the system are assumed to be described by unexpected changes in physical parameters of the system. The parameter changes in the controlled system can effectively be detected by using Kullback Discrimination Information (KDI) as an index for model discrimination. In order to decide whether the detected system parameter change is caused by a failure or not, a fuzzy inference approach to failure decision is considered. Some appropriate membership functions which describe fuzzy events of failures are constructed to perform the fuzzy inference. In this way, useful knowledge about failure modes which is available from, e.g., experts can be introduced into the model- based diagnosis technique. Simulation studies of a second-order damped oscillator have been carried out to demonstrate the effectiveness of the method.
Some fundamental results related to binary and multi-level decision logic distributed decision (evidence) fusion (DD(E)F) are presented. New asymptotic performance results for binary and multi-level Neyman-Pearson (N-...
详细信息
Some fundamental results related to binary and multi-level decision logic distributed decision (evidence) fusion (DD(E)F) are presented. New asymptotic performance results for binary and multi-level Neyman-Pearson (N-P) DD Farederived. Acomparative resultbetween N-P DDF and Dempster-Shafer DEF in the framework of the Generalized Evidence Processing (GEP) theory is presented under a specific decision rule at the fusion.
In this paper a new method for the computation of the optimal step in gradient algorithms is presented. This method improves the convergence of the gradient algorithms and outperforms any other suboptimal scheme on li...
详细信息
In this paper a new method for the computation of the optimal step in gradient algorithms is presented. This method improves the convergence of the gradient algorithms and outperforms any other suboptimal scheme on linear problems while it does not require any additional storage. The method may be also applied to problems with state or control constraints, linear time varying systems and, via linearization, to nonlinear systems as well.
The dual relation between the Model Reference Adaptive control (MRAC) and Identification (MRAI) problems is discussed. A common framework for both problems is established.
The dual relation between the Model Reference Adaptive control (MRAC) and Identification (MRAI) problems is discussed. A common framework for both problems is established.
A solution to the H/sub infinity / mixed sensitivity problem for the SISO (single-input single-output) case is obtained using a Wiener approach to parameterize all equalizing and stabilizing controllers. The controlle...
详细信息
A solution to the H/sub infinity / mixed sensitivity problem for the SISO (single-input single-output) case is obtained using a Wiener approach to parameterize all equalizing and stabilizing controllers. The controller which incorporates the LQG (linear quadratic Gaussian) solution has a structure similar to that of D.C. Youla et al. (1976). The system of equations thus obtained is square and has some degree of advantage over previous solutions.< >
A common framework for model reference adaptive identification and control (MRAI and MRAC) is established. The key idea for this common framework is to maintain the same set of equations for describing the parameteriz...
详细信息
A common framework for model reference adaptive identification and control (MRAI and MRAC) is established. The key idea for this common framework is to maintain the same set of equations for describing the parameterizations of the plant and the model and to solve the control equation properly for each case (identification or control) for the corresponding tuned system, i.e. the model (identification) or the plant (control). Within this framework, two methods, the MOEM (modified output error method) and the IEM (input error method), for the MRAI and MRAC problems are studied, and global asymptotic stability properties are established.< >
A modified output error method (MOEM) for model reference adaptive control (MRAC) and identification (MRAI) is introduced. The regressors are properly chosen so that the open-loop system can be compactly expressed as ...
详细信息
A modified output error method (MOEM) for model reference adaptive control (MRAC) and identification (MRAI) is introduced. The regressors are properly chosen so that the open-loop system can be compactly expressed as a stable system with an input linear with respect to the unknown plant parameters. The output error satisfies a constructible stable filtered equation. The method does not require any strictly positive real or arbitrary stable filterings, and the uncertainty on the magnitude of the high-frequency gain of the plant does not result in an overparameterization of the identifier. Sufficient conditions on the reference input under which the parameter and output errors converge exponentially to zero are also given.< >
暂无评论