巻号一覧
23 巻, 7 号
選択された号の論文の18件中1～18を表示しています
- |<
- <
- 1
- >
- >|
-
盧 度煥, 山浦 富雄, 小林 彬1987 年 23 巻 7 号 p. 659-664
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThe scales of the Vernier calipers widely used in precise length measurement are a combination of a main scale and a Vernier scale, which are two kinds of scales with wide different pitch scale lines. The scale combination makes it possible to measure the length in higher resolution than the original scale line pitch.
In this paper, the above concept of the Vernier calipers is introduced to a new displacement measuring system for high resolution and high accuracy. In this system, the main and Vernier scales are replaced with light sources and photosensors to make automatic scale reading (this is called automated Vernier scale method). The measured displacement is given in the discrete interval, so the measuring resolution is limited with respect to the width of the discrete interval. To improve this resolution limit, we propose two algorithms utilizing analog data processing without any change of the hardware construction. They are the first order moment algorithm and the phase discrimination algorithm.
To examine the improved automated Vernier scale method, we construct an experimental measuring system. The main scales are replaced with photosensor array of twenty sensors at intervals of 0.5mm. The Vernier scales are replaced with light beams of a laser diode scanned by a polygon mirror. As the experimental results, the repeatable accuracy of the measurement was ±3∼4μm within 5mm of measuring range.抄録全体を表示PDF形式でダウンロード (1784K) -
三橋 渉, 望月 仁, 田村 安孝, 山崎 弘郎1987 年 23 巻 7 号 p. 665-670
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーA linear-period-modulated (LPM) signal is a well known means used by some species of echolocating mammalian such as bats and cetaceans.
In an acoustical measurement method detecting many moving objects, the LPM signal is free from the range-Doppler coupling if constructed symmetrically with respect to time reversal. Therefore, transmitting the LPM signal and recording the time variations of echoes by a sensor array, both objects positions and velocity vectors can be reconstructed in good accuracy.
In this paper, the theory of velocity estimation by means of the LPM signal is described and is extended to a two-dimensional case with a Linear sensor array. Furthermore a new method to reconstruct objects velocity vectors by using orthogonal correlation detectors is proposed. The validity of this method is confirmed by the experiment in air with the linear sensor array composed of a transmitter and 8 receivers. The estimated velocity vectors agree well with those of actual object motions.抄録全体を表示PDF形式でダウンロード (913K) -
鳥越 一平1987 年 23 巻 7 号 p. 671-676
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーA new method for measuring the mass flow rate through a pipe is described. A pulsating source of known strength is placed in the flow. This source produces an axial flow pulsation superimposed on the otherwise steady flow. And the flow pulsation causes an alternating pressure as a direct result. The pressure difference across the source is picked up by a manometer of high sensitivity. A phasesensitive detector being employed, the output of the manometer is converted to a DC voltage which is proportional to the mass flow rate. An experimental device which employed condenser microphones for picking up the pressure difference has been built on a trial basis. Several measurements were performed for air flow. The mass flow rates measured with this device were in good agreement with the ones which were determined from flow velocity and density measurement. A mass flowmeter based on this method is simple and compact in construction, and has an obstructionless configuration and a minimal pressure drop.抄録全体を表示PDF形式でダウンロード (1740K) -
佐藤 光男, 竹田 宏1987 年 23 巻 7 号 p. 677-683
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーMany studies have been devoted to the Markovian decision problem in which the transition probabilities are unknown, and various learning schemes have been presented for it. Most of these works are concerned with seeking convergence to optimal control inputs under presupposition that the time length is infinite. Considering the practical phase, however, we can say that a finite-time problem is important as well as the above mentioned infinite-time problem. In a finite-time problem, the goal of learning is not assurance of the optimal convergence but minimization of the total loss in the whole process.
Previously, for the finite-time two-armed bandit problem, which is the simplest model of learning control problems, we presented a learning scheme that was designed from introducing an idea of deviling the whole process into the first part of estimating the unknown parameters and the subsequent part of controlling the system.
In this paper, we design a finite-time learning scheme with the same idea for a Markovian decision model, and present a stopping rule of the estimation. The Bayesian method is used in the estimation part. As a performance criterion for the stopping, the maximum posterior probability is considered, and then a loss function defined on the basis of the posterior probability distribution is applied. Using the result of examination of a statistic property of the maximum posterior probability, we attempt to find a time point that the loss function attains its minimum in the sense of an approximation. The stopping rule presented here is such that the estimation should be stopped at the time point. Furthermore, we show some results of computer simulation to inspect accuracy of the approximation.抄録全体を表示PDF形式でダウンロード (990K) -
Mehdi NOURI-SHIRAZI, Nobuo SANNOMIYA, Yoshikazu NISHIKAWA1987 年 23 巻 7 号 p. 684-691
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーIn this paper we consider the state estimation problem of a discretetime linear stochastic system for a fairly general case where both the system disturbance and the observation noise are nearly Gaussian. First, the techniques developed hitherto are introduced for special cases where either the system disturbance or the observation noise is nearly Gaussian while the other one is Gaussian. Secondly, two possible filtering approaches are proposed for the general case by extending the filters constructed for the special cases. The performances of various approximate filters are compared with that of the Gaussian sum filter through Monte-Carlo simulations. A discussion is made on reducing on-line computational difficulties in the nearly Gaussian filtering.抄録全体を表示PDF形式でダウンロード (916K) -
栄坂 俊雄, 田川 遼三郎1987 年 23 巻 7 号 p. 692-698
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーWhen a digital computer is used as a controller of the control system, the error of the control variable due to finite wordlength implementation of the controller can not be avoided. This error is named ‘Quantization Output error of the Controller (Q.O.C. for short)’ in the present paper. The Q.O.C. causes an error of the controlled variable named ‘Quantization Output error of the Over all system (Q.O.O. for short)’, and this makes it difficult to complete precise control.
In this paper we point out that the magnitude of Q.O.O. depends on the algorithm of the controller. Then we indicate that the design of the DDC algorithm with small Q.O.O. is important in order to complete precise control. Furthermore, we propose a procedure for the DDC algorithm which may raise small Q.O.O..
We confirm the effect of the proposed procedure by simulation with two design examples.抄録全体を表示PDF形式でダウンロード (928K) -
新 誠一, 北森 俊行1987 年 23 巻 7 号 p. 699-706
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThe time delay is one of major problems in the application of the modern control theory, since the time delay system is infinite dimensional in the continuous time formulation. It can be represented as a finite dimensional system in the discrete time formulation, so that the adaptive control for the time delay system is mainly studied in the discrete time formulation.
However, the discrete time control has some disadvantages, that is, hidden oscillation, deterioration of the control quality, and introduction of unstable zeros. Therefore, it is also necessary to study the continuous time control without being satisfied with the discrete time formulation.
This paper is concerned with the global stability of the control system, where the continuity of the solution is assumed and the plant is restricted to the single input single output linear system with an input time delay.抄録全体を表示PDF形式でダウンロード (793K) -
吉田 和信, 西村 行雄, 米澤 洋1987 年 23 巻 7 号 p. 707-711
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーA design method of a relay controller is presented, from the viewpoint of the properties of the sliding mode, for a linear stable plant whose representative initial conditions are known. Namely, it is shown that the transient property of the relay control system can be explained qualitatively in terms of the speed of response in the sliding mode and the size of the sliding mode region. Using this relationship, we propose an interactive design algorithm for finding the feedback gain which gives an appropriate property of the sliding mode. Since the properties of this mode depend on the poles of the plant and the zeros of the open-loop transfer function, how to assign the zeros is essentially important. This paper also presents that the LQ optimal gain having one parameter can be utilized for the zero assignment such that the stability of the relay control system is taken into account.
A numerical example for a third order system is shown to demonstrate the availability of the proposed algorithm.抄録全体を表示PDF形式でダウンロード (632K) -
松尾 孝美, 冨永 明, 相良 節夫1987 年 23 巻 7 号 p. 712-718
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThe purpose of this paper is to investigate the problem of designing a minimal order function observer with arbitrary poles which estimates (n-m-μ)-dimensional partial state for a continuous-time observable n-th order system with an m-dimensional output.
A generic condition to construct a function observer with arbitrary poles is reduced to a restrictive condition for left eigenvectors of a reduced order observer by use of the following result: a full order observer converges to a reduced order observer as some poles approach infinity in the left half complex plane. On the basis of the restrictive condition, in μ≤m minimal order of function observer with arbitrary poles is given, whereas in μ>m upper bounds of minimal order are given.抄録全体を表示PDF形式でダウンロード (941K) -
崔 霖, 武藤 康彦, 市川 邦彦1987 年 23 巻 7 号 p. 719-726
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThe column properness of a polynomial matrix is one of the most important properties for the interactor of the linear multivariable systems. However it is observed that the properties of the column improperness play a more important role rather than those of the column properness in the derivation of the interactor. In this paper, the improperness index and di-improperness index which indicate the measure of the column improperness are defined and some of their properties are discussed. For this purpose, the “Γ-matrix” is defined using the coefficient matrices of the polynomial matrix, which palys the basic role for the derivation of the interactor. The relation between the improperness index and the rank of Γ-matrix of the same polynomial matrix is highlighted.抄録全体を表示PDF形式でダウンロード (895K) -
渡部 慶二1987 年 23 巻 7 号 p. 727-733
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThis paper is concerned with disturbance rejection and stability of Smith predictor control system. It is shown that there are two type of disturbance compensators, which are proper and strictly proper. Smith predictor control system with a proper disturbance compensator has better disturbance rejection, but less stability than the ordinary Smith predictor. By using a strictly proper compensator, it is possible to improve the characteristics of disturbance rejection without damaging the stability of Smith predictor. Moreover, Smith's system which is not robustly stable can be stabilized robustly by the strictly proper compensator.抄録全体を表示PDF形式でダウンロード (752K) -
前田 章, 坪井 晃, 古村 文伸1987 年 23 巻 7 号 p. 734-741
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーA new method to produce high-quality space-borne SAR (Synthetic Aperture Radar) imagery is proposed. It is known that complex SAR imagery must be interpolated before being converted to real imagery in order to prevent the degradation of spatial resolution. However, if the complex imagery has phase discontinuities or rotation, this method causes another degradation of image quality. In this paper, origins of phase discontinuities and rotation are analyzed. Especially it is shown that there exist essential phase discontinuities due to the discretization of azimuth frequency, and phase rotation due to the space-variant point spread function. Next, a method is proposed to accomodate these phase discontinuities and rotation, and its effectiveness is verified by experiments using SEASAT-SAR data.抄録全体を表示PDF形式でダウンロード (2319K) -
藤原 敏勝1987 年 23 巻 7 号 p. 742-749
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーIn recent power generating system in our country, nuclear power plants are used in baseloaded operation and the balance of demand and supply of the electricity is maintained mainly by thermal power plants.
In future, the above mentioned tendency will be accelerated by the increase of generating capacity of nuclear power plants so that thermal power plants will be obliged to bear more rapid and bigger amount of load change.
However, such a rapid and bigger amount of load change requirement as mentioned above in thermal power plants will not be satisfied by a ordinary PID control system only.
In this situation, we have deveroped an advanced control system named the predictive control system, and have applied this system to the steam temperature control of super critical pressure boiler plants.
This method is able to realize approximately invariant of controlled variables for the known disturbance e.g. the required output using a mathematical plant model.
The model itself is composed of a dead time and rational function, the parameters of which are obtained by the identification of plant.
The result of application shows the following performance.
1) The controlled deviation of steam temperature is less than half of that obtained by the feedback control with PID controller only.
2) The control is sufficiently stable.
It is understood that these results permit more rapid load change rate in comparison with a conventional rate.
This paper describes the principle of the method and the field test.抄録全体を表示PDF形式でダウンロード (1014K) -
横塚 勉, 関根 秀臣, 馬場 英治, 池田 眞治1987 年 23 巻 7 号 p. 750-756
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーA torque equation, for a case of do current excitation, is derived by applying the IBL law. The electromagnetic torque developed, with the dc current, can be obtained by use of speed electromotive force.
Based on the torque equation, a method is given for calculating the accelerating characteristics for the case of a step of dc voltage application. Starting characteristics under ac voltage applied condition are also given. Relation between the dimension ratio and the optimum number of poles taking account of starting characteristics is investigated.抄録全体を表示PDF形式でダウンロード (1267K) -
岩尾 範人1987 年 23 巻 7 号 p. 757-763
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーIn order to simplify the blast furnace control system, study was made to understand the relations among the reaction raw materials (X), the reaction products (Y), and the reaction rates (r) in the blast furnace by the method described below.
1) Basic model of B.F. operation
The material flow in the blast furnace is diagrammed according to the conservation laws of mass and number of atoms.
This diagram was proposed by the author as the basic model of B.F. operation, to clarify the relations among various process variables.
2) Method of obtaining reaction products (Y)
In seeking the reaction products (Y) from the reaction raw materials (X), it will be found that the number of equations is greater than the number of unknown variables (Y).
Hence, the selction of equations is required for obtaining the solution.
3) Application of generalized inverse matrix
The generalized inverse matrix is the method which uses all equations obtained. And its application is described here.
4) Method of obtaining reaction rates (r)
In seeking the reaction rates (r) in the blast furnace from the reaction products (Y) and the reaction raw materials (X) described in the foregoing paragraph, the node equations are formulated by the use of the basic model.
In this case, the number of node equations is also greater than the number of unknown variables (Y).
It is shown here that the reaction rates (r) must be sought by selecting independent nodes in accordance with the theory of graph.
5) Result of simulation
It is found that the solution loss value, which is an important B.F. operation index, is different for the traditional method and the generalized inverse matrix method.
This result indicats the necessity to find a method most suited for estimating the state of actual process.抄録全体を表示PDF形式でダウンロード (918K) -
田中 庸彦, 熊本 博光, 井上 紘一1987 年 23 巻 7 号 p. 764-771
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThis paper presents a “General Coverage Monte Carlo” method for evaluating the topevent probability of a coherent fault tree. All the minimal cut sets are assumed to be known. An order k coverage Monte Carlo is developed where the first k sums of product of an inclusion-exclusion formula need to be computed by deterministic algorithms and the contribution of the (k+1) st and the higher sums is evaluated by Monte Carlo.
The order k coverage Monte Carlo estimate always falls in between the corresponding upper and lower bounds obtained from an inclusion-exclusion formula. It is proven that variances of order k coverage Monte Carlo estimators monotonically decrease under reasonable assumptions as order k increases. The estimator's variance and coefficient of variation can be evaluated before and after the Monte Carlo execution. In contrast to the direct Monte Carlo, the coefficient of variation is proven to approach to zero as a system becomes more reliable and hence rare-event characteristic becomes more dominant. A numerical example is given.抄録全体を表示PDF形式でダウンロード (892K) -
桑原 正典1987 年 23 巻 7 号 p. 772-774
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーPDF形式でダウンロード (349K) -
松田 文夫, 岩崎 知弘, 服部 秀三, 上田 実1987 年 23 巻 7 号 p. 775-777
発行日: 1987/07/30
公開日: 2009/03/27
ジャーナル フリーThis paper describes a real time detection of a perfect image which includes a whole object contour line whose coordinates are filled into memory. The detection is performed as follow. If the contour line of the object deviates from the observation image area, the side of deviation is indicated through a status register which can be accessed any time from the system. If the contour line does not deviate, the coordinates of the contour line are filled into a memory.抄録全体を表示PDF形式でダウンロード (428K)
- |<
- <
- 1
- >
- >|