Smith predictor control system can not yield desired responses to disturbances. This paper presents a method which can decay the output error to disturbances as fast as desired without change of the input-output transfer characteristeristics. This method uses a dynamical compensator. The design of the compensator and the mechanism of the disturbance rejection are discussed in detail. Illustrative examples are given to demonstrate the availability.

View full abstract

Nowadays a chance using microcomputers for the design of digital feedback control system becomes increase. However existing digital control theory has many problems which must be solved before practical use. For example, in existing digital control theory, the time delay arising form the computation time is not properly accounted.
The authors have proposed a design method of the optimal state regulator accounting controlling time delay in the previous paper and have solved many other important problems.
The contents of this paper are the followings.
(1) Using the fundamental relation derived in the previous paper, a design method of the optimal digital servosystem is firstly proposed in which the optimal feedback gain matrices are derived from solving corresponding optimal output regulator problem.
(2) And another design method of the optimal digital servosystem accounting controlling delay is proposed. The feedback gain matrices of this servosystem are also derived form solving corresponding optimal output regulator problem.
The results are very simple and they seem to be suited for the practical use.

View full abstract

The algorithm of multiplier method is discussed from the view point of dual problem for nonlinear programming problem. We present a class of the updating algorithm of Lagrange multiplier which employs conjugate gradient method in order to accelerate the rate of convergence of multiplier method.
The rate of convergence of conjugate gradient multiplier method ranks next to that of quasi-Newton multiplier method. But, it is expected that the conjugate gradient multiplier method becomes a practical algorithm for nonlinear programming problems with many inequality constrains because there is no need of memory space to accumulate the information for Hessian inverse of dual objective function.

View full abstract

We describe an algorithm to determine the transfer function of a system.
The time constants of the system are included both in coefficients and exponential terms of the response. When the order of the system becomes high, it is difficult to evaluate them from the response. The exponential terms of the response can be replaced by a set of polynominals. They are derived from Euler's identity and maxima of output responses of Gaussian shape pulse signals. In addition, by using the cross-correlation functions between system output and Gaussian pulse signals, the method can easily be applied to the systems with white observation noise.
Our results demonstrate that it is possible to obtain a good evaluation of the system parameters by the simple procedures.

View full abstract

This paper deals with the problem to find the solutions of the convexly combined form of Sup·min and Inf·max composite fuzzy relations equation, λ·(x_??_R)+λ·(x_??_R)=y, and provides the method to solve it. First, the convexly combined form of Sup·min and Inf·max composite fuzzy relations equation is formulated by applying a notion of fuzzy convex combination. Furthermore, a mixed fuzzy input-output relational system of pessimistic and optimistic compositions is discussed, which is represented in the convexly combined form of Sup·min and Inf·max composite fuzzy relations equation. Second, we show the necessary and sufficient condition on which the set of solutions is not empty. Third, we provide the method to find the solution which satisfies an optimum condition. Finally, a numerical example is shown.

View full abstract

This paper describes a study on a new type adaptive noise canceller for periodic interference. The usual adaptive noise cancelling system is controlled by a clock pulse whose frequency is determined a priori considering the frequency band width of the reference input. On the other hand, the adaptive filter proposed in this paper is designed to operate synchronously with the reference input, that is, the frequency of its clock pulse varies in proportion to the fundamental frequency of the reference input. For the purpose, the phase-locked loop is used. It captures the fundamental frequency component of the reference input and generates synchronizing pulse. Synchronous operation yields such profit that the total computational cost decreases to about 1/2N times as compared with the usual nonrecursive adaptive filter, where N is the number of weight coefficient of the adaptive filter. It is a great help not only in simplifying the hardware construction but also in widening operational frequency range of the adaptive filter.
This paper presents theoretical investigations on the notch characteristics of the synchronous adaptive filter. And it is experimentally shown that the proposed adaptive filter has excellent performance.

View full abstract

In a servomechanism with high accuracy of operation, a feedback loop of high gain is required, whose realization is often difficult according to the parasitic mechanical resonance of actuators.
To avoid this difficulty, a method called “By-Pass Method” giving the construction and equalization of a feedback loop is proposed. In general, an existing actuator has a low pass characteristic, so that if we add a simple by-pass circuit of high-pass type in parallel, a flat frequency characteristic is obtained: in a high frequency region where parasitic resonances exist, the transmission through by-pass circuit is dominant. By this method it is not necessary to equalize one by one the peaks and bottoms in the frequency characteristic of an actuator.
As an example, an application to the tracking servomechanism in video-disk system is shown, and a loop gain over 80db is realized experimentally.

View full abstract

An large amount of leakage is one of the problems, for water distribution network management. The service pressure in the water distribution network should be set optimal so that the pressure is high enough to use water and is not higher than the required pressure, since the pressure cause the leakage. This is formalized as an nonlinear optimization problem constrained by water distribution network nonlinear equations.
This paper is concerned with a practical method for the pressure optimization. By representing the water distribution network by Loop Equation Formulation for purpose of describing it in a small number of equations, we formalize the constrainted optimization problem. And we transfer the constrained optimization problem which includes operational variables and flow variables into an unconstrained one which includes only operational variables by using Implicit Function Theory. To solve the unconstrained optimization efficiently, we adopt the variable metric method and some technique for speed-up.
The feature of the proposed method are as follows: (1) a small amount of computer memory, (2) short computation time, (3) evaluation by Kuhn-Tucker condition. (4) simple algorithm. The method was verified to be effective by experiments using the real water distribution network data.

View full abstract

Orange peel effect is the degree in which a painted surface resembles the surface of an orange peel. Both orange peel effect and gloss are very important factors which prescribes the quality of painted surfaces.
In this experiment, a series of six standard samples were prepared which possesed appropriate orange peel effect intervals scaled by the sensory evaluation using a modified paired comparison method and have various levels of gloss. These samples were coated with silver metallic paint and solid paint containing many kinds of pigments.
Substitute characteristics for orange peel effect were analyzed, using the samples and an apparatus which optically senses the distortion of a reflected square wave pattern image. Correlation was investigated between the orange peel effect intervals and the average distortion δ defined below; δ=(n-1)^-1n-1Σ_i=1(|l_i-l₀|/l₀), where n is number of lines, l₀ is line space of an image reflected at a plane mirror and l_i is the i th line space of an image reflected at a sample.
The results are as follows.
1) A linearity between the orange peel effect intervals and δ is better for l_i determined by dark-to-dark (trough-to-trough) than bright-to-bright (peak-to-peak) spacing.
2) A good linearity was obtained for the pattern pitch p=0.1&0.2mm.
3) These experimental results were explained by the theoretical analysis using a mirror formula derived by approximation of a painted surface to consist of a series of concave and convex mirrors.
4) It has been found from the above results, that the following condition should be satisfied to obtain a good linearity; 2βp_≅λ_w, where β is the magnification of projection and λ_w is the wavelength of waviness.

View full abstract

A flexible reconstitution distributed control system that has no redundant controller is proposed, and reliability of the system is evaluated.
(1) One controller is connected to a plant by process input/output modules (PMs) and a serial bus which is used to monitor the state of the other controller. A failed controller is disconnected from its serial bus by itself. At the same time, the serial bus of the failed controller is automatically connected to the serial bus of the normal controller which monitors the state of the failed controller. Therefore, normal controllers are able to backup the functions of any failed controllers, so that the system can continue to control the plant.
(2) The system has higher reliability than typical highly reliable distributed control systems that consist of triple redundant controllers or 2 out of 3 redundant controllers, in respect of System Function Reliability (SFR) defined as the rate of survival functions of control system in this paper, MTTF, and reliability.

View full abstract

Due to the wide variation of thermal constants such as specific heat, difference equations are used mostly to predict the heating temperature of slabs in the furnace. The difference equation approach entails time sharing, and thus vast amount of recursive computation is required to perform the prediction calculation of heating temperature of slabs by means of difference equation. As a result, the difference equation approach has a defect in long computation time. This drawback becomes a great obstacle when the difference equation approach is used in the application of on-line predictive control of heating temperature of slabs in the furnace.
In this paper, a new and fast algorithm is proposed for the prediction calculation of heating temperature. The algorithm uses analytical solution but is recursive in nature. First, the prediction of temperature is calculated through analytical solution derived in this paper. Every time when the calculated tempereture goes beyond the prespecified temperature a range, the calculation interrupts to correct the thermal conductivity approriate to the next range of temperature and to reset the initial distribution of temperature, and then the prediction calculation restarts.
The heat conduction equation which describes the slab considered in this paper is assumed to be one-dimensional. The temperature distribution of slabs governed by one-dimensional heat conduction equation can be approximated by quadratic function. Hence, in this paper, the analytical solution of one-dimensional heat conduction equation with quadratic function as its initial temperature distribution is derived using Laplace transform.
The proposed algorithm for the prediction of rising temperature has a great advantage in reduction of computation time over the difference equation aproach. As the proposed algorithm takes into account the nonlinearity of thermal conductivity, the computation accuracy is expected to be nearly equal to that of the difference equation approach.
Several numerical examples show the following results:
(i) the required computation time is reduced to 1/5 as compared with that of difference equation approach,
(ii) the accuracy of prediction remains within the allowable range.
These facts assure that the proposed algorithm is an effective way to the prediction computation of heating temperature and can be employed for the on-line control of slabs in the furnace.

View full abstract

Overall direction-of-action measurement is described. In the measurement, an opaque ball housed in a transparent spherical-vessel is used as a sensor equipment. Since there is sufficient room in the vessel, the ball can move freely under the influence of acceleration of gravity and motion. Specifically, in a static condition, it makes an equilibrium state by indicating the direction of the gravitational field. Under this condition, the ball becomes an optical target and is illuminated by the light from a point source of light. Projected image of the ball in a photosensitive area becomes nearly equal to an ellipse which distinguishes between interior darkness and exterior brightness. Relationships between the displacement of the target and the image data are geometrically analyzed to extract the directional information from the projected image. Flowchart to determine the exact direction resulted from the analysis.
In the experiment, the lens of ball-shaped is used to throw the projected image into the limited photosensitive area. It has been confirmed that the measurement error is within 12 degrees in all directions and the time required is within 500 milliseconds. This method would be applicable to realize direction sensors to prevent platforms, inverted pendulums, and leg systems from inclining, shaking, and falling down.

View full abstract