計測自動制御学会論文集
Online ISSN : 1883-8189
Print ISSN : 0453-4654
ISSN-L : 0453-4654
21 巻, 8 号
選択された号の論文の21件中1~21を表示しています
  • 小林 彬, 坂本 泰彦
    1985 年 21 巻 8 号 p. 771-777
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    This paper proposes a general method capable to process paired comparison data with order effect. The method is able to estimate psychological interval scale and to estimate the degree of order effect.
    First, conditions that the model should satisfy are investigated. And it is indicated that band-function model satisfies such conditions. Next, some properties about order effect are defined on band function model. Third, the algorithm to obtain interval scale is reviewed. Then, in order to estimate characteristics of judgement from paired comparison data and psychological interval scale, an algorithm for curve fitting is proposed. Last the whole method proposed in this paper is applied to the paired comparison data about discomfort of burst sound. Results assures that order effect and psychological interval scale is estimated by this method. Furthermore, the psychological interval scale is compared with LA, slow-max value as physical scale on burst sound.
  • 志水 清孝, 石塚 陽, 太田 裕久
    1985 年 21 巻 8 号 p. 778-785
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    We study a mathematical programming problem whose objective and constraint functions include the extremal-value functions. This problem can be regarded as an extension of well-known “min-max problem” and/or infinitely constrained problem”, and it becomes a nondifferentiable optimization problem. In this paper, we present some optimality conditions for such problem by use of a new theorem of the alternative and Gauvin's results concerned with the extremal-value functions. Moreover, as a computational method for our problem, we propose applying the Mifflin's nondifferentiable optimization algorithm by means of the generalized gradients. A numerical example is also presented.
  • 勝部 泰弘, 堀口 和己, 浜田 望
    1985 年 21 巻 8 号 p. 786-791
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    In order to design control systems, it is necessary to have a mathematical model which will adequately describe the motion of a given system. Modelling problem is a significant field in control theory, and has been much studied in relation t signal processing theory.
    In this paper, we concern with a modelling method using 2-point Padé approximation on discrete-time systems. The rational function of z (difference operator) obtained by usual Padé approximation preserves the first some terms of the given power series expansion at a point. When we choose z=1 (z=∞) as that point, the first parts of time moments (Markov parameters) of the model coincide with the given ones in version of discrete-time systems. On the other hand, 2-point Padé approximation provides the transfer function which fits the desired characteristics at two points respectively, hence selecting z=1 and z=∞ as these two points, we can obtain the model with both steady and transient response taken into account.
    The outline of this paper is as the following: after definition of 2-point Padé approximation problem in Sec. 2.1, a few subjects on it are discussed in the next four sections. They are (1) an explicit form of 2-point Padé approximation (Sec. 2.2), (2) manimal order's reakization problem (Sec. 2.3), (3) Nuttall's compact form (Sec. 2.4), (4) relation to continued fraction expansions and numerical efficient algorithm based on it (Sec. 2.5). In Sec. 3, a numerical example of modelling via 2-point Padé approximation is shown for demonstration of its effectiveness.
  • 大久保 重範
    1985 年 21 巻 8 号 p. 792-799
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    A model following control system (MFCS) can let output signals follow desired ones. Since MFCS permits general signals as desired ones, its application's range is wide. This paper describes the design of nonlinear MFCS of which the inner states are bounded under disturbances, and the output signals converge to desired signals asymptotically. In a nonlinear MFCS the boundedness of inner states is important. In this paper we set the nonlinear part (f(v(t))) of the controlled object as ||f(v(t))||≤α+β||v(t)||γ, and show the boundedness of inner states by separating the nonlinear part into two cases, i.e. 0≤γ<1 and γ≥1. There are not strong constraints in such a setting of the nonlinear part. In the case of 0≤γ<1, the fact that characteristic roots of system matrix are stable guarantees the boundedness of inner states. In the case of γ≥1, the conditions that the transfer matrix H(p) from f(v(t)) to v(t) is positive real, and the coefficient tensor of the highest degree of f(v(t)) is negative definite guarantee the boundedness of inner states. We show the availability of this design in a numerical example of which the nonlinear degree is five.
  • 藤井 信夫, 市川 雅司, 香西 治彦
    1985 年 21 巻 8 号 p. 800-805
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    In this paper are discussed the domain (shape) optimization problems in which a boundary value problem is a main constraint The approach developed by one of the authors is applied to slightly extended problems This approach is also applied to a problem of fluid mechanical engineering, namely, the minimumdrag problem in low Reynolds number limit. The outline of the approach is as follows; the existence of the variation of the solution, corresponding to a domain variation, to the boundary value problem is to be shown; the variational equation which relates the variation of the solution to the domain variation is to be derived; a neccessary condition is, then, to be obtained for each problem with a subsidiary constraint by the well-known Lagrange multiplier rule and by introducing adjoint variables. The domain optimization method demonstrated in this paper can be applied to other problems similar to those treated here.
  • 小畑 秀文
    1985 年 21 巻 8 号 p. 806-812
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    This paper describes a study on texture segmentation based on a two-dimensional linear prediction method. The principle that the division of an image which gives the minimum of the total sum of final prediction errors over the whole image corresponds to the true segmentation is shown if each texture region consisting the image can be assumed to be a homogeneous gaussian random field. An edge detection method which is based on the principle is also given. Although this method needs a great amount of computation, a practical procedure of computation using Dynamic Programming is developed. The effectiveness of this edge detection method is demonstrated experimentally.
  • 西原 元久, 嶋田 智, 田辺 正則, 山田 一二, 松岡 祥隆
    1985 年 21 巻 8 号 p. 813-820
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    This paper discusses the temperature dependence of the nonlinearity in silicon diaphragm piezoresistive pressure sensors. Minimization of the temperature dependence is very important in designing high-precision pressure sensors which can be used over the wide temperature range. The amount of temperature dependence can be evaluated quantitatively by using the non-linear characteristics formula of the piezoresistive effect in p-type silicon diffused layers. From the characteristic analysis of silicon diaphragms in {110} and {100} planes by this method, the following results are derived. (1) The amount of the temperature dependence of the non-linear error increases in proportion to the increase in strain added on the silicon diaphragm. (2) Optimum gauge pattern which minimizes the non-linear error and its temperature dependence is found to exist on the {110} plane.
    Using these results, high-precision semiconductor pressure sensors are developed. The specifications are as follows: (1) The nonlinear error is less than ±0.1%. (2) The output voltage is larger than 65mV/V. (3) The operating temperature range is between-40°C and 120°C.
  • 大竹 由晃, 西田 直矩
    1985 年 21 巻 8 号 p. 821-826
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    Facility location problems involve the selection of potential facilities that will be used as sources to satisfy a known demand at various customers that are to be serviced by the facilities. Associated with each potential facility is a capacity limitation on throuput and a cost function which is broken down into a fixed cost plus a transportation cost. Although a number of literature on the facility location problem have been published, problems considered in these literature assume that the demand at customers is constant.
    The parametric facility location problem (PFLP) treated in this paper is to solve a facility location problem, in which the demand at various customers is varied continuously over a specified range. The variation of the demand at customers is formulated with a parmeter in the problem. In this paper, a branch-and-bound algorithm for solving PFLPs is developed. Also, a new concept of the parametric pegging test is proposed to improve the computational efficiency of the algorithm. Finally an illustrative example is given to show how the present algorithm works.
  • 前田 幹夫, 村上 周太, 尾上 敏一
    1985 年 21 巻 8 号 p. 827-834
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    This paper deals with a medical decision problem on the choice of clinical tests or treatments for anovulatory (sterile) patients with functional infertility under the fuzzy information evaluated by a physician. He has to decide an appropriate action considering many aspects, such as effectiveness, cost, time, mental and physical pains, and etc., caused by clinical tests and or treatments. A fuzzy decision analysis technique developed here is one that contains such multi-objectives and also fuzzy information about the consequence of each attribute, the grade of disease group of anovulation and his medical knowledge and experience among diseases, clinical tests and treatments. Those information are measured by λ-fuzzy measure and a conditional fuzzy measure proposed by Sugeno. To give the best decision, the fuzzy expected utility for each alternative, which implies the grade of goodness, is calculated using the fuzzy integral.
    As an application of fuzzy decision analysis on the choice of tests or treatments for anovulatory patients, forty three samples were examined on the basis of their clinical records under cooperation with a physician. As results of analysis, thirty seven samples gave the same agreements between the best action obtained through this analysis and the actual behavoir of the physician.
    Finally, we got comments from the physician that this system worked as he did and that this could be a useful tool to support physicians' medical judgement and decision making.
  • 鬼沢 武久, 菅野 道夫
    1985 年 21 巻 8 号 p. 835-841
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    This paper is concerned with proposing a fuzzy set of unreliability based on 'likelihood of failure' which is expressed by a fuzzy set instead of failure rate in the conventional fault tree analysis, and presenting a method of fault tree analysis. In this analysis we consider the fault caused by a component, human error, environment and common mode failure. A causal relation of common mode failure is expressed by a fuzzy relation. Standby redundant systems including series systems and parallel redundant systems are analyzed. Especially we define an operator for unreliability of a standby redundant system.
    Five criteria are defined to evaluate the results of the present analysis,
    i) the numerical unreliability which gives the greatest grade of the membership of the fuzzy set
    ii) 0.9-cut of the fuzzy set
    iii) the fuzziness of the first criterion
    iv) the potentiality of failure
    v) the importance of a basic event.
    As an illustrative example, a circuit diagram of a power supply in a nuclear power plant is shown and its fault tree is analyzed, where the top event is a loss of electric power supply to engineering safety features. It is shown that the present method gives us more useful informations than a probabilistic method does.
  • 熊本 博光, 鈴木 俊雄, 井上 紘一, 池西 憲治
    1985 年 21 巻 8 号 p. 842-848
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    Most expert systems (ES) have been programmed in artificial intelligence languages such as LISP and PROLOG. This paper proposes a new approach which programs an ES for plant failure diagnosis in a data manipulation language (DML) of a relational database. The personal computer DML of dBASE II or K-MAN is used here. The approach is first justified. Then a knowledge acguisition and representation scheme of incremental refinement type is introduced for the failure diagnosis, and the inference engine and others are implemented accordingly by the DML. The resulting ES is overviewed and it is demonstrated by application to two subsystems of a ship engine propulsion system.
  • 築山 俊史, 諏訪 基, 白井 良明
    1985 年 21 巻 8 号 p. 849-855
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    This paper describes a system called DIP which was designed to perform experiments conveniently for development of new approaches in image processing. Image processing experiments require many but inhomogeneous functional devices such as video digitizers for data collection, conventional computers, special hardware for data proceesing, display devices and image memories for temporary store. Systems for the experiments have to select functional devices and to perform them efficiently according to user's programs. Especially, systems must realize high speed transfer of image data, because each frame of TV images contains huge amount of data and transfers of image data are frequently repeated among functional davices in experiments. For realizing an efficient system, switching mechanisms are considered to play important roles. In DIP a network system is implemented for programmable interconnection of functional devices. Distributed control mechanism is adopted for control of the system. Each functional device is activated and performed by control data broadcasted from a computer via the network. Computers never supervise each control sequence of selected functional devices. With the network system any two devices are connected directly and an image data is transferred via the network at a high speed. In additions users can extend and change configuration of the system with ease, according to their demands, because DIP can allow to contain more than one computer and any kind of functional devices.
  • LSIウェハ上異物の自動検出技術の研究
    小泉 光義, 秋山 伸幸
    1985 年 21 巻 8 号 p. 856-862
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    Contaminants have a direct effect on LSI yield. At present, visual inspection of Si wafers for contaminant is an essential part of the semiconductor QC process. However, the conventional QC process requires much time and significantly limits LSI yield. Therefore, a means of detecting these particles to facilitate early rejection of the contaminated wafers is needed.
    Several systems have been developed which automatically detect surface contaminants or particles on unpatterned wafers by measuring the intensity of scattered laser light. However, no system has yet been applied to inspecting patterned surface, because micro-patterns on the surfaces behave as diffraction gratings producing noise signals which are hard to distinguish from signals produced by particles.
    In the present technique, information about particles is obtained by measuring the polarization of the scattered light. The basic idea in the present technique is that, if an analyzer placed behind the objective could cancel only the light reflected from the pattern, then the light reflected from contaminants would be readily detectable in the absence of pattern noise. Therefore, the relation is determined between the analyzer angle and the electric vector of the incident light which would effectively cancel the pattern noise. It has been verified that particles as small as 1μm could be detected from the pattern noise, by experiments.
  • 清水 基夫, 長井 賢, 大木 永光
    1985 年 21 巻 8 号 p. 863-868
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    An optical fiber liquid sensor for the remote detection of very small amount of liquid leak was developed. The sensor consists of an optical prism and focusing lenses and has the following features.
    (a) long distance, maltiple point detection for more than 10km
    (b) insensitive to water
    (c) safety in explosive or high voltage environment
    In this sensor, the optical signal from the input fiber is collimated by a Selfoc lens, and the incident angle to the detection surface of the prism is adjusted to a specific angle which is determined by the refractive indices of the prism and the liquid to be detected. The sensitivity of the sensor was 5dB against a differnce in refactive indics of 0.02, according to the experiment using various kind of liquids. The resulting data of the expriment were well matched to the theoretical value, and indicated that the sensor can discreminate liquids of differnt refractive indices. The sensitivity against the liquid quantity is also very high as to detect a small spot of kerosine on cloth. An experimental system for long distance leak detection was evaluated using an optical fiber of 1.6km long. The system works by the same principle to optical fiber fault locator (OTDR), with a range gate circuit controlled by a digital delay circuit to detect sensor distance. A 10km leak detection system with 17 liquid sensors was estimated to be possible using 1.3μm wavelength and such system will be suitable for pipe-line or plant maintenance application in chemical industry.
  • 山形 仁, 阿部 健一, 星宮 望
    1985 年 21 巻 8 号 p. 869-874
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
    A new method of voltage clamping of biological membranes with time-variant parameters has been developed. In our previous system, the control ability was not sufficient in the active state where the membrane showed a considerable increase of conductance. In this paper, we discuss a new multimicroprocessor system to overcome the above problem and realize fast adaptive control of biological membranes. The system was developed through four approaches as follows.
    (1) Projection algorithm was applied to the parameter estimation to design the optimal time-pattern of the effective loop conductance in the negative feedback loop.
    (2) Adaptive control method by iteration was design as follows: in each stage of realtime sampling, the deviation current from the preceding controlled current was controlled by voltage feedback, and the modified current was injected to the membrane in order to charge up the membrane capacitance to the command voltage. This method was repeated until the difference between the command voltage and the membrane voltage reached some minimum level.
    (3) Multi-microprocessor system which had master-slave structure was developed to realize high speed sampled-data control, parallel processings of parameter estimation and display of experimental results.
    (4) System performance was tested on the electronic exitable membrane model which simulated conductance changes of smooth muscle membranes.
    From the experimental results of the performance test, it is concluded that the adaptive voltage clamp method is very effective in the active condition as well as steady state condition of several kinds of exitable membranes.
  • 原田 博之, 柏木 濶
    1985 年 21 巻 8 号 p. 875-877
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
  • 安藤 嘉則, 鈴木 正之
    1985 年 21 巻 8 号 p. 878-880
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
  • 小林 敏弘
    1985 年 21 巻 8 号 p. 881-883
    発行日: 1985/08/30
    公開日: 2009/03/27
    ジャーナル フリー
  • 1985 年 21 巻 8 号 p. 885a
    発行日: 1985年
    公開日: 2009/03/27
    ジャーナル フリー
  • 1985 年 21 巻 8 号 p. 885b
    発行日: 1985年
    公開日: 2009/03/27
    ジャーナル フリー
  • 1985 年 21 巻 8 号 p. 885c
    発行日: 1985年
    公開日: 2009/03/27
    ジャーナル フリー
feedback
Top