In this paper, the ranking methods of basic probability assignments (b. p. a.) in Dempster-Shafer theory are discussed and decision making methods based on the ranking indices are proposed. First, it is emphasized that a b. p. a. has two interpretations, i. e., measure-theoretic and set-theoretic interpretations. On the basis of these two interpretations, various ranking indices for b. p. a. are defined and classified into six groups. Next, using Orlovsky's fuzzy decision making method, we apply the six indices to two types of decision making problems, i. e., the problem in which a probability is assigned not only to a state of nature but also to a set of states of nature and the problem with interval utilities. Six Kinds of decision making methods are proposed.
In general, it is difficult to design a control system which satisfies completely all demanded design specifications for the control system, because these design specifications are ambiguous and contradict each other. in this paper, to make the design specifications clear, we apply the fuzzy division which has been used in the fuzzy control theory in the treatment of design specifications for linear control systems. That is, we choose the phaseplane as the appropriate designing space which has relation with these specifications and fuzzily divide this phaseplane according to these specifications. In the each fuzzily divided region, we decide the proper control which satisfies the cleared design specifications. And then, the total control for the control system is constructed by the weighted control which is decided properly in the each fuzzily divided region. Lastly, we apply this design method to the linear regulator in problems.
This paper describes a modeling with unmodeled uncertainties and control configuration for a magnetic suspension system of a double flexible beam. It simplifies both an elastic rotor and a body in a magnetic bearing system. At first, the magnetic suspension system is introduced. Next, dynamical equations of two beam motions as a concentrated state system and an electrical equation of an electromagnet are derived. Then its linearized model with disturbance terms as unmodeled uncertainties is formulated in a state space form. Finally, in order to deal with the uncertainties as a control problem, a standard H∞ control framework is set up. There the generalized plant is constructed from LQ differential game theoretical point of view, which involves weighting matrices for state variables and control input and frequency weighting functions for uncertainty models.
We propose two unsupervised learning control schemes for a nonlinear system based on a performance function, which is given by the total cost without presenting a target signal. A feedback neurocontroller self-optimizes the system through minimizing the performance function. The first uses a generalized back-propagation method which calculates the sensitivity of the performance function with respect to the cennection weights. The second uses difference approximation of the sensitivity which is given by suitable perturvations for the connection weights. We apply these two schemes for controlling an inverted pendulum whose initial position is hanging. Each scheme accomplishes stabilization of the pendulum at the center of the rail. The neurocontroller acquires a strategy of accerative oscillation round the hanging position to raise the pendulum within a short rail, and it damps the oscillation and leads the pendulum to the center of the rail by a skillful balance about the standing position.
This paper compares the time and memory requirements of an approximately linear phase IIR digital filter with those of a linear phase FIR digital filter, and shows that, only if a substantial difference between the orders of the filters exists, the former can be considered as an attractive substitute for the latter. Furthermore, it presents a method which makes feasible the design of reduced-order IIR digital filters that resemble the characteristics of a linear plase FIR digital filter. Some design examples are shown, illustrating how the proposed method can be used to design approximately linear phase IIR digital filters that are faster and require less memory than their equivalent linear plase FIR counterparts.