Model reference adaptive control system (MRACS) theory is an effective method for dealing with unknown systems. When an MRACS is constituted, current parameters and the state variables of the unknown plant are usually estimated to adjust control parameters. But there are some problems when they are estimated. This paper presents a design method of an MRACS by an approximately inverse functional compensator. The system is constructed from a viewpoint of a learning control method and is, based on an exact model matching (EMM) technique. The system designed by the proposed method can decrease the control error between the output of the plant and the reference model without estimation of the parameters and the state variables. The, learning control method is used for the plant which can be controlled repeatedly. In the frequency zone of signals passing through the plant and the reference model, the approximately inverse functional compensator has inverse properties of the reference model.
This paper deals with the following two design methods of finite-time settling control system (FTSCS) with softening filter for general one-input and one-output linear controlled object. One is the design of FTSCS under the input saturation, and another is that minimizing the quadratic performance index. The structures of the systems designed by proposed methods are more simple than those by conventional methods. And it is shown that the design parameters are easily and briefly specified by using some polynomials describing the filter and the state feedback elements. To demonstrate the effectiveness of proposed methods, some numerical examples are also presented.
This paper proposes a new knowledge representation and a new reasoning for an expert system in assisting production line design which is the planning process used when introducing new machine tools into a factory. In this system hybrid knowledge representations are used. These are the hierarchical knowledge representing the structure in three layers and the knowledge representation using semantic network. In semantic network a machine tool type is divided by its machine mechanism and function in detail. In regard to the reasoning, we propose the multiobjective evaluation reasoning which has a two stage reasoning process. These processes use the elimination reasoning method which can examine a machine structural characteristic and compare the multiplicative utility function values. Because of this comparison capability it is able to evaluate many objectives simultaneously when purchasing machine tools.
The purpose of this study is to estimate dynamics of the articulatory organs (jaw and tongue) by voice analysis. Relation between control signals to the articulatory organs and PARCOR coefficients is simulated by a transfer function model of second order systems. Then, parameters of the model (time constants and damping rations) and control signals (control timings and amplitudes) are estimated by applying least mean square method to the error between PARCOR coefficients and responses of the model for sequential vowel/aiue/. Synthesized voice using the. estimeted model has the same characteristics as the original voices in voice analysis and hearing test. Also, the dynamics of the model estimated are almost equivalent to five male subjects, while the sequential patterns of the control signals show the personal differences of the subjects. It could be concluded from the estimation that personality of utterance mainly depends on the control signal to the articulatory organs.
This paper deals with fuzzy control of state variables in an interconnected multi-reservoir power system combined with real-time prediction. A mathematical model of the system in described as a discrete-time stochastic state-space model, including conditions which restrict the amount of water released from the upper stream reservoir and also restrict the method of discharging the water into the lower stream. In practice, it is required to notify few hours in advance the change of discharge into the lower stream. This is the reason why we need real-time prediction of the states for determining the amount of control well in advance. As the results of simulation studies we point out the effectiveness of fuzzy control compared with optimal regulator from the point of view of practical control of extending the durability of water gates and power generators.