Transactions of the Society of Instrument and Control Engineers
Online ISSN : 1883-8189
Print ISSN : 0453-4654
ISSN-L : 0453-4654
Volume 10, Issue 5
Displaying 1-20 of 20 articles from this issue
  • HIDEO Hanafusa, Hiroshi KAWAGOE
    1974 Volume 10 Issue 5 Pages 521-526
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    A supersonic flow amplifler with a double nozzle was investigated. The supply nozzle consists of a main and a secondary nozzle. By this construction, the attainable Mach number is greatly increased due to entrainment by the main jet. A supersonic flow is easily obtained even if the supply pressure is low. Moreover, the control flow rate which is introduced to the inside of the secondary nozzle is greatly reduced. Having superior input and higher gain characteristics, this amplifier can be used in various fields.
    First a supersonic free jet flowing from a circular nozzle into wide space is experimentally studied. The diameter at the most expanded section and the wave length of the jet which are necessary for the design of the double nozzle were discussed. By using semiemprical formulae obtained in this study, the dimensions of the double nozzle were determined and verified by experiments.
    The position of the receiver port is determined by considering the position of the shock wave. As a resistance connected to the output port becomes lower, the shock wave approaches the receiver port, and finally it is absorbed into the receiver port. When the resistance is low, maximum flow recovery is obtained at a short lr, which is the distance between the secondary nozzle and the receiver port. But when the high resistance is connected, violent oscillations take place at this lr. The maximum output pressure is obtained when the lr is set a little longer. The maximum incremental pressure and flow gains are 4.1 and 8.1 respectively. Therefore this device is useful as an interface valve for controlling high power flow by low power flow.
    Download PDF (733K)
  • Yoshifumi OKUYAMA
    1974 Volume 10 Issue 5 Pages 527-532
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    In this paper, the stability problems, of linear dynamic systems with bounded uncertain coefficients or nonlinear time-varying elements are investigated in the time domain relating to the transition matrices of the nominal systems. Sufficient conditions for the relative stability of such systems are given in L1 and L spaces, respectively.
    It is shown that these two conditions are equivalent and described by a simpler form when the nominal systems are time-invariant.
    The method of analysis is as follows. First, integral inequality systems are given from the integral equations, which themselves are obtained from the state equations, and then the conditions of boundedness of norms are derived by using the properties of the M-matrix. The principle of this idea is quite simple. But the stability criteria developed here will be perhaps more favourable than any other results which have been already obtained for nonhomogeneous systems involving nonlinear time-invariant ones.
    Download PDF (703K)
  • Yoriaki BABA
    1974 Volume 10 Issue 5 Pages 533-539
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    This paper is concerned with a practical algorithm for the on-line computer control of a linear process with inequality constraints on the control forces. The control signals are determined so as to force the disturbed state to the desired state as soon as possible. The control concept is essentially based on a time-optimal control for a linear discrete-time system without inequality constraints on the control forces, and on Lyapunov's direct method. The control vector, which is suboptimal in the sense of minimum-time control, can be determined immediately by means of performing the product of the constant matrix and the state vector or by solving the linear optimization problem. Because the procedure of calculation is straight-forward and does not require much computer time or excessive computer memory, this algorithm seems to be a powerful tool for a computer control system.
    Download PDF (884K)
  • Kouyu MIYAZAKI, Masahiro SESAKI
    1974 Volume 10 Issue 5 Pages 540-545
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    A non-contact measuring method for testing the flatness of a mirror or mirror-like surface by reflection of a laser beam is mentioned. A laser beam breaks into several beams arranged at constant intervals and in the same direction, and these beams are projected on a mirror surface to be measured. The reflecting beams thus obtained are intercepted by a screen separately provided, and by measuring the coordinates of the spots of light thus obtained, the deviation compared with the ideal geometrical plane at each reflecting point is calculated. In this paper, the authors propose the measuring principle, measuring errors and experimental results of the above method, and make it clear that the flatness of the mirror can be obtained mechanically by this method.
    Download PDF (1251K)
  • Hideo HANAFUSA, Keiichiro MIYATA
    1974 Volume 10 Issue 5 Pages 546-551
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    The accuracy of the switching pressure of fluidic comparators is investigated by considering the statistical properties of the switching process. The switching process of wall-attachment amplifiers is classified into an initial switching process and a random switching process. The accuracy of the fluidic comparators is mainly concerned with the random switching process, where the main jet is switched by the random fluctuation superposed on the jet deflection after the full development of the attachment bubble.
    First, the probability distribution function of the switching delay for the random switching process is studied for step inputs. The assumption of the exponential distribution function for the switching delay was verified by experiments, and the statistical parameters in the mathematical model were determined.
    Next the accuracy of the switching pressure is discussed for rectangular pulse inputs and ramp inputs by referring to practical applications of the fluidic comparators. A typical application is seen in on-line measuring and sorting systems of mechanical parts, where the inputs to the comparators are approximated by rectangular pulses with short durations. In this case, it is desirable that the input pressure range of the uncertain switching be as narrow as possible. The uncertain switching pressure range decreases by increasing the input pulse duration. The switching probability for various input pressures and pulse durations was calculated by the mathematical model and verified by experiments.
    Another typical application of the fluidiccomparators is seen in process control systems such as pressure control or liquid level control, where the input pressure changes slowly. The swithcing of the comparator is desired to take place at a sepcified pressure level. As the estimation of the accuracy of the comparators, the average value and the variance of the switching pressure for ramp input were calculated by the mathematical model and were verified by experiments.
    Download PDF (1286K)
  • Hideo TAGUCHI, Katsuhiko FUJII
    1974 Volume 10 Issue 5 Pages 552-560
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    It is expected that a study on the human adaptive and learning process will provide some clues to the learning control system synthesis. After performing some learning experiments with a linear hand wheel on various controlled elements, human data are extracted in a gradual manner, and general characteristics about a gradual mode are discussed in this paper. Corresponding to this gradual mode, a sampleddata model for human behavior, say a linear reinforcement learning type model, is proposed. In order to bring the action of this model close to human behavior in a gradual manner, this model consists of four sub-systems-the probabilistic logic network, the performance evaluator, the memory and discriminator. The action principle of this model is as follows.
    First, in the probabilistic logic notwork, the most rewarding action for a given control feature is decided by reinforcement of the transition probability between elements of the control choice space. Second, the iterative control action of the probabilistic logic network and the performance evaluator is stored as control experience in the memory. And then a discriminant function of the control characteristics (i.e., a maneuver line) is learned on the basis of this experience.. This discriminant function is used to produce a deterministic control action.
    In the later section, the comparison between the subject's responses and their digital simulation results is discussed.
    Download PDF (1120K)
  • Akira INOUE
    1974 Volume 10 Issue 5 Pages 561-568
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    The observation problems for a wide class of discrete-time linear control systems are considered. The problems are to find observability conditions and to design state- and functional-observers. The class of control systems includes systems with multi-inputs and multi-outputs, and systems with singular transient matrices.
    First, the k-observability of the dual state of a control system and the observability of a control system are defined, and their necessary and sufficient conditions are obtained. Next, using the k-observability of a dual state, the minimal-time state-observer is defined and a construction of the observer is given. For a kind of nth-order control system with m independent outputs, the observer obtained by the construction has an order less than (n-m)-order. Finally a functional-observer estimating a linear functional of states is derived from the minimal-time state-observer. And it is shown that when k-times observations are necessary to know the value of the functional, then the minimal-order of the functional-observer is equal to k-1.
    Download PDF (940K)
  • Takashi KATO, Shuzo KAWAMOTO
    1974 Volume 10 Issue 5 Pages 569-574
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    The dynamic characteristics of cistern manometers are analyzed by the periods and logarithmic decrements of the free oscillation and by the resonance frequencies of the forced oscillation.
    The equation of motion of the cistern manometer is given in the form of a second-order nonlinear differential equation that is derived from the Navier-Stokes equation.
    The liquid of the manometer used is mercury. The cistern manometers are composed of a mono-tube and cisterns as follows:
    diameter of mono-tube:
    3, 5, 8mm
    diameter of cistern:
    3, 13, 20, 30, 44, 77, 106mm
    5, 13, 20, 30, 44, 77, 106mm
    8, 13, 20, 30, 44, 77, 106mm
    The following became clear after the comparison of the experimental values with the theoretical ones.
    1. On the periods, experimental values agree well with the theoretical ones. If the diameters of cisterns are more than 30mm, then the values of the period are found to be approximately 1.8sec.
    2. When the sizes of the mercurial column and cistern, are increased, the values of the logarithmic decrements increase in proportion.
    3. As the sizes of the cisterns increase, the resonance frequency decreases. If the diameters of the cisterns are more than 30mm, the values of the resonance frequencies are found to range from 32 to 34 cycle/min.
    Download PDF (700K)
  • Oscar A. NAWA, Jun IKEDE
    1974 Volume 10 Issue 5 Pages 575-580
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    In a previous paper, isocount scintiscanning was presented by the authors as a method for improving the statistical reliability. of the gathered data. It has been noted that the reduction of scanning time is very important for the sake, of the patient's comfort and also that the direct visualization of photographic film constitutes a considerable loss of information.
    A television display system that composes a scintigram from some fewer sampled data through two-dimensional interpolation is presented. It is demonstrated that the resulting picture has an improved quality and no loss of detail. Moreover, the scanning time may be reduced to less than that of the conventional constant speed method.
    The multilevel analysis is also presented as a technique that enables the distinction of a single bit by expanding those data between a reference level L and L+k(k=8 to 10) to the entire density curve of the TV screen.The reference L is sequentially changed, and this analysis has been demonstrated to be very useful for liver scanning. For the brain, with k=0 (binary picture), this expedient produces an animation-like sequence of silhouettes which facilitates the feature extraction, making the diagnosis more objective.
    Download PDF (2834K)
  • Tai Kyu KIM, Toshiro NOSE
    1974 Volume 10 Issue 5 Pages 581-588
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    A holder which converts the discrete input signal into continuous form by interpolation or extrapolation of the input pulse train is used frequently as an accessory of a digital computer.And this so called zero-order holder is commonly used.
    The purpose of this paper is to find a systematic procedure for obtaining the transfer function of higher-order holders from Lagrange's interpolation formula connected with directly sampled values and to show the characteristics of the systems thus found.
    A general method is made for the technique of realizing the higher-order holders. The Nthorder holders thus obtained are composed of a zeroorder holder, N integrators and several potentiometers. The frequency response and the error characteristic of the higher-order holder are given.
    The instantaneous error and the mean squared errer of each holder to the sinusoidal input have been measured by using higher-order holders constructed on a hybrid computer.Then the maximum instantaneous errors of the holders to the sinusoidal input are estimated theoretically and experimentally. The experimental results show that the instantaneous errors and signal to noise ratios are improved remarkably by increasing the order of the holders.
    Download PDF (2147K)
  • 1974 Volume 10 Issue 5 Pages 588
    Published: 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    Download PDF (90K)
  • Michio KONO, Ichiro SUGIURA
    1974 Volume 10 Issue 5 Pages 589-593
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    A non-interacting control of an r-input m-output multivariable control system via state variable feedback is considered. Since inputs of the control law correspond to reference inputs, a non-interacting control system is constructed by the control law with as many inputs as there are outputs of the plant. As a feedforward gain matrix of this control law is not square, this control law changes F-invariant properties. In this paper, sufficient conditions are derived for non-interaction by a control law with a rectangular feedforward gain matrix.
    An easily computable criterion for non-interacting is derived. Two examples are given illustrating the result.
    In a special case, the structure of the non-interacting system is shown.
    Download PDF (535K)
  • Bounded Connectedness and State Response Completeness-and Their Application to the Controllability Concepts
    Yasuhiko TAKAHARA
    1974 Volume 10 Issue 5 Pages 594-598
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    In this paper two state space related system properties, i.e., bounded connectedness and free state response completeness are introduced for linear dynamical systems, and a significance of the finiteness of a finite dimensional state space is explored by them. The results are applied to controllability concepts. The bounded connectedness requires of a linear system that every state reachable from the origin or able to be steered to the origin can be reached or steered uniformly within a time interval characteristic of the system. The free state response completeness requires that a free state response function should have the inverse. This paper shows that a time invariant linear dynamical system satisfies the bounded connectedness if its state space is finite and free state response completeness if its input-output response is analytic and its state space is finite. In particular, it is shown in the paper that a constant coefficient linear ordinary differential equation system possesses both properties.
    There have been defined three kinds of controllability, that is, controllability from zero, controllability to zero and complete control-lability. In general, they are different concepts. This paper however shows that they are equivalent for the class of time invariant linear dynamic systems satisfying both properties.
    This result imples that the three concepts of controllability are equivalent to each other for a system by a constant coefficient ordinary differential equation. This is well known but has not been proven by system theoretic arguments, as it is in this paper. Most of the results of linear control theory have not been sufficiently conceptualized so that the result can be helpful in understanding systems in an intuitive way. This paper, as well as another paper (Reference (4)), tries to make a contribution to the work of conceptualization of the linear control theory.
    Download PDF (737K)
  • Masaru ISHII, Tadashi NAGATA
    1974 Volume 10 Issue 5 Pages 599-605
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    One important aspect in research on an intelligent robot is visual information processing. A visual device is required for an intelligent robot to perform various complicated tasks.
    This paper describes the principle and the construction of a Laser Tracker as a visual device with an “active” faculty. The Laser Tracker consists of flying-spot equipment and a 2-dimensional measuring device. The flying-spot equipment moves a laser spot freely and the measuring device obtains DC voltages according to the 2-dimensional coordinates of the brightest point in the range of an image dissector.
    This robot's eye acquires 3-dimensional coordinates of a laser spot on an object on the basis of the triangulation and then extracts the feature of the object corresponding to the purpose of the intelligent robot.
    At the first part of this article, the basic experiment to obtain 3-dimensional coordinates of a laser spot is reported.
    As one of applications, experimental results of classification and feature extraction of several
    3-dimensional objects are followed.
    Download PDF (1932K)
  • Yoshimasa YOSHIDA, Kahei NAKAMURA
    1974 Volume 10 Issue 5 Pages 606-610
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    A dual control problem for linear, discrete-time, single-input single-output stochastic systems with unknown parameters is considered under the criterion of minimum variance. The dual control system is designed as a recursive learning process which collects information for control through an identifier by use of the Kalman filter, and simultaneously is controlled adaptively by the minimal variance control strategy which computes, the control signal required to minimize the variance of a linear function of state in the succeeding step.
    By using the observable state vector consisting of n past outputs and n-1 past inputs, the stochastic control problem can be solved analytically. The optimal regulator can be thus separated into two parts: one is a one-step predictor which predicts the distribution of the output caused by the disturbance and the parameter uncertainty, and the other is an adaptive minimum variance regulator.
    It is also shown that the control law is updated at each time when a new information received. The control error for the optimal strategy gradually approaches the part caused by the disturbance of the system as the process proceeds. This means that the part caused by the parameter uncertainty disappears gradually.
    Download PDF (704K)
  • Velocity Profile of Attached Jet
    Osamu KINOSHITA, Yasujiro OSHIMA
    1974 Volume 10 Issue 5 Pages 611-616
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    The velocity profile of the attached jet in a liquid fluid amplifier has been investigated by means of a constant temperature hot-film anemometer. The dimensions of the fluidic element used are: main nozzle width bm (=control nozzle width be): 25mm, offset D: 2.5 bm, side wall angle α: 15° and aspect ratio: 5.0.
    Using water of about 20°C as the operating fluid, the studies were performed in the conditions of main jet velocity vo=2.5m/s and dimensionless control flow rate Qc/Qm=0∼0.3. The data were analyzed referring to Goertler's model of a turbulent jet.
    The conclusions from the experimental study are:
    1) The velocity decay along the attached jet center line is remarkable at the region downstream from the vortex where it is affected by the side wall.
    2) The velocity profile perpendicular to the jet center line is closely approximated by Goertler's model in which a reasonable spread parameter iss evaluated separatel on the attached and unattached sides of the jet.
    3) On the unattached side of the attached jet the spread parameter is constant, but on the attached side the parameter is a function of s, which is the co-ordinate along the jet center line.
    Download PDF (873K)
  • Toshi TAKAMORI, Mikio ISHIBASHI, Masatada YONEMOCHI
    1974 Volume 10 Issue 5 Pages 617-624
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    In an analogy of electric circuit theory, a distributed constants pneumatic filter is designed for use as a transmission line matched with an nformation source and a load circuit in fluidics circuit systems. In this study, a low-pass filter is designed from the given approximating function of power attenuation by the following three different methods:
    -A method of design by an equivalent transformation from a given lumped constants filter circuit.
    -A method of design from an approximating function obtained by the Q-function.
    -A method of design by a digital computer compensating for the effect of a line loss.
    In the former two methods, the pneumatic circuit is treated as a loss-less one. But in the last method, the loss effect is considered.
    Three pneumatic filters are actually synthesized with each method mentioned above and their transmission characteristics are experimentally investigated.
    The strengths and weaknesses of these methods are discussed in detail in this paper.
    Download PDF (976K)
  • Saburo IKEDA, Masasuke SHIMA
    1974 Volume 10 Issue 5 Pages 625-630
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    In our previous papers, we have already shown that if “bilinear” distributed parameter system such as heat exchangers and tubular reactors are “reachable” and “controllable”, then the pattern of optimal control to minimize an integrated square error of the outlet state of the system is reduced to the type:
    Bang-Bang Control→Singular Control.
    This paper studies “reachability” and “control-lability” of the outlet state of a bilinear distributed system which is described by a partial differential equation of the first order. Transforming the partial differential equation into the ordinary differential equation with respect to the outlet state, we derive some theorems which give the necessary and sufficient conditions on “reachability” and “control-lability”, not only for the steady-state initial and boundary conditions but also for the non-steady-state ones. Moreover, the class of functions of initial and boundary conditions to satisfy “reachability” and “controllability” is specified.
    Download PDF (687K)
  • Juichi MIYAMICHI, Takeshi FUKAO
    1974 Volume 10 Issue 5 Pages 631-636
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    To embrace a broader class of input-output models in linear system theory, it is necessary to extend the notions devised in the context of stronger algebraic structures (vector spaces) to those in weaker structures.
    In this paper, the system theory-structure theory and realization theory-of linear dynamical systems, previously developed over a field, are extended to a principal ideal domain. It includes the system theory over integers as a special case.
    The systems under discussion here are linear, constant and discrete-time dynamical systems over a principal ideal domain A of the form x(t+1)=Fx(t)+Gu(t) y(t)=Hx(t) whereX(t)∈An, u(t)∈Am, y(t)∈Ap, and F, G, H are matrices over A.
    The concepts of reachability, observability, minimality, canonical system and realization are difined just as in the case when A is a field. Some of the results are analogous to those of linear systems over a field and others are not.
    Main differences:
    (1) The duality between reachability and observability does not hold.
    (2) The minimal system and the canonical system are not equivalent.
    (3) The state space can not be decomposed by reachability as a direct sum.
    Analogous results:
    (4) The state space decomposition by observability is possible.
    (5) For any system, we can find a canonical system which has the same input-output relations.
    (6) Ho's realization algorithm can be generalized (with minor modification) to a principal ideal domain.
    Download PDF (755K)
  • Mitsuo OHTA, Shizuma YAMAGUCHI
    1974 Volume 10 Issue 5 Pages 637-638
    Published: October 30, 1974
    Released on J-STAGE: March 27, 2009
    JOURNAL FREE ACCESS
    Download PDF (236K)
feedback
Top