This paper discusses stability of a class of adaptive control systems. The system considered consists of an adaptive block and non-adaptive blocks. The adaptive block is constructed by the conventional model reference adaptive control (MRAC) approach. Assuming the general properties of MRAC, sufficient conditions for the whole system to be stable are obtained.
This paper deals with an application of analytic hierarchy process (AHP) to extract the user's purpose which must be input to expert systems. In the AHP, the process to obtain the weights for criteria by pairwise comparisons is interpreted as the extraction of the decision marker's idea about the importance of each criterion. With this interpretation, the analytic hierarchy process is applied to extract the analyst's idea of the weights of several factors to select constitutive equations suitable for a target analysis in a model selection support expert system under development for numerical simulation of nuclear thermal-hydraulics; the constitutive equations are ordinary introduced in a liquid-vapor two-phase flow analysis. Furthermore, an algorithm by applying the graph theory is shown to evaluate covering condition to obtain the weights from incomplete pairwise comparisons.
The rapid progress of micro-processing techniques enables us to use intelligent control algorithms of high level for complex systems by using micro-computers. The heating cylinder for plastic molding is one of those complex systems which is required to control the temperature as precise as possible. In this paper an adaptive temperature control method is proposed to improve performance of the temperature control system for the heating cylinder. This method is based on the multivariable self-tuning control (STC) theory. The temperature control system contains three heaters and three sensors. The unknown parameters in the equation are identified by the least squares estimation method. We obtain a minimum-phase model for the heating cylinder by tuning the value of the time delay contained in the identified model. The adaptive controller has been constructed for the identified system by STC theory. Some experimental results are illustrated to show that the STC method is effective in comparison with a conventional method by PID Controller and Ziegler-Nichols method.
This paper deals with a digital tracking control system subject to change in load condition. It is assumed that the plant is described by a discrete-time state-space model with a time delay and a direct forward link between control input and output. An optimal controller with state feedback plus integral and preview actions is derived by using the LQI technique. It is shown under the stabilizability and detectability conditions that the optimal closed-loop system is stable and achieves a complete output regulation. For a special case where the time delay is unilateral, we present an algorithm for computing feedback gain with less computational effort. The tracking performance of proposed optimal control law is evaluated through a numerical example for a heat exchanger model. Experimental studies using a pilot plant show that the controller gives acceptable tracking performance in the presence of load change.
In this paper, we consider the problem of continuous adaptive identification where a periodical 'test input signal can be used. First, we propose a design method of the gain matrix of the adjust rule for the Kreisselmeier's type adaptive identifier. With it, all of the parameters converge approximately, with similar exponential rate. Second, a new parameter adjust rule is presented for the Lüder-Narendra's type adaptive identifier. Using this rule, the parameters converge approximately with the same exponential rate as. Kreisselmeier's type adaptive identifier. Finally, the design of the input signal, the state filters and the gain matrix of adjust rule, etc. are discussed, and computer simulations are presented.