計測自動制御学会論文集
Online ISSN : 1883-8189
Print ISSN : 0453-4654
ISSN-L : 0453-4654
31 巻, 12 号
選択された号の論文の17件中1~17を表示しています
  • 金山 尚樹, 金子 真, 辻 敏夫
    1995 年 31 巻 12 号 p. 1915-1923
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    Artificial Active Antenna is a new family of sensing system that can detect the contact location between an insensitive flexible beam and an environment through the measurement of the rotational compliance of the beam in contact with the environment. This paper discusses the 3D Active Antenna composed of an insensitive flexible beam, two actuators to move the beam in 3D space, two position sensors to measure the actuator position, and a two-axis moment sensor to evaluate the contact force. The lateral slip, which is the inherent characteristic for 3D Active Antenna, strongly depends on the pushing direction, the friction at the point of contact, and the normal surface of environment where the antenna makes contact. Generally, such a lateral slip overestimates the rotational compliance, deteriorates the sensing accuracy directly. In this paper, we discuss how to detect the lateral slip by using the sensors implemented and how to determine the next pushing direction for avoiding the lateral slip. We show that the two-axis moment sensor is effectively utilized for detecting the lateral slip. We propose a sensing algorithm which eventually lead to the pushing direction providing the exact contact distance. We also show a practical approach enabling us to detect the contact distance with permissible sensing accuracy, by at most three active motions.
  • Harumi KAMADA, Nobuharu AOSHIMA
    1995 年 31 巻 12 号 p. 1924-1929
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    A windowed CFOS (complex first order system) filter is presented in this paper. This filter is an extension of a finite impulse response CFOS filter which has been already reported. The name, “windowed”, indicates that the absolute value of complex output of the filter can be made equal to the well-known window functions. If we select an appropriate window function, we can get a filter which has good localization not only in the time domain but in the frequency domain. Using this filter, we could realize a constant Q-value filter bank. Design techniques of the filter are described. Then we applied the filter bank to analyze the human voice with DSP. The proposed algorithm is beneficial in application to real-time processing.
  • 瀬部 昇, 北森 俊行
    1995 年 31 巻 12 号 p. 1930-1935
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper is concerned with the diagonal dominance. The diagonal dominance is the property that the diagonal elements are relatively larger than the off-diagonal elements. And it is an important property for multivariable feedback systems with diagonal controllers.
    The original diagonal dominance was defined by 1-norm or ∞-norm, where the matrix p-norms are induced norms. And generalizations were made by using either scaling or Perron-Frobenius root. But all the dominance loose the phase information, because these depend only on the absolute values of each elements of the matrix.
    This paper proposes a new diagonal dominance defined by 2-norm. This diagonal dominance is less conservative than the other dominance. This paper also clarifies the relation between the diagonal dominance and strictly positive realness.
  • 杉江 俊治, 川西 通裕
    1995 年 31 巻 12 号 p. 1936-1944
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper is concerned with the exact description of the structured uncertainty for robust control. So far, the structured uncertainty of the plant was often expressed as affine perturbations of the coefficient matrices in the state-space or the descriptor form representation. However, this type of expression is sometimes very conservative for the uncertainty due to physical parameter perturbations. Therefore, in this paper, a systematic method to describe the physical parameter perturbations in an exact way is given based on the LFT and the descriptor form representation. It is shown that the proposed uncertainty description combined with the mixed μ analysis gives a sharp robust performance bound compared to the conventional uncertainty expressions via numerical examples.
  • 坂本 登, 鈴木 正之
    1995 年 31 巻 12 号 p. 1945-1953
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    In this paper the concept of passivity is generalized from the view-point of the rate of energy dissipation, introducing a supply rate with an index γ. A stabilizability result for dissipative systems with the supply rate is obtained. Next, a design method of γ-passive systems is discussed by using a bilinear transformation of nonlinear systems, called Cayley-transform. Finally, in the case of linear systems, it is shown that the index γ can be used as a measure of phase of transfer functions. A shaping method of the phase curve of closed loop transfer functions is also proposed. A generalization of so-called passivity theorem is obtained, which is shown to be effective, by using an example, for analysis of robust stability of feedback systems.
  • 平田 健太郎, 山本 裕, 片山 徹, Allen TANNENBAUM
    1995 年 31 巻 12 号 p. 1954-1961
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper studies the interrelationships among varied solutions to the 1-block sensitivity minimization problems. Of particular interest are the direct relationships among the Nehari-type solution, Nevanlinna-Pick interpolation solution and the skew Toeplitz theory. In spite of the obvious improtance of such relationships, their direct equivalence is not available in the literature, except rather indirect proof via Sarason's theorem. It is shown that the Pick matrix is closely related to the orthogonalization of eigenvectors of the compressed shift, which in turn leads to the equivalence relationships among these solutions.
  • 谷川 民生, 佐伯 正美
    1995 年 31 巻 12 号 p. 1962-1969
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    An H design method of a robust controller with tuning parameter is applied to a hard disk positioning control. On the gain diagram, sensitivity and complementary sensitivity functions can be moved along frequency axis by tuning parameter. A good system with sensitivity and robust stability being balanced adequately can be obtained by tuning it. Analytical solutions of Riccati equations which contain a parameter are required in the design. Since the plant transfer function has a special form of a/sn in this case, they can be easily constructed from numerical solutions of Riccati equations and it is not necessary to solve it analytically. This controller is compared with a fixed controller designed by a mixed sensitivity design method by simulation.
  • 則次 俊郎, 高岩 昌弘
    1995 年 31 巻 12 号 p. 1970-1977
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper describes one method to attain a robust positioning control of pneumatic driving system comprising a pneumatic cylinder and PCM digital control valves.
    A pneumatic driving system has become one of interests in the industrial fields, however, the control characteristics are inevitably affected by the friction force and parameter changes of plant due to air compressibility.
    The control characteristics of a pneumatic actuator depend on a pressure response. So if a pressure can follow the reference value which compensates the influences of the friction force and the parameter changes of plant, the friction force and the parameter changes of plant do not affect the control characteristics.
    A meter out speed control method, widely used in a pneumatic system, is employed. This method has a superior energy efficiency but gives a nonlinear pressure response. From these points of view, we proposed a positioning control system including an active controlled pressure loop.
    Two disturbance observers are employed in this control system. One is applied in a pressure control loop to improve the robustness against the influence of the piston speed which acts as a disturbance and to compensate the nonlinearity owing to the meter out method, the other is in kinetic part, from the pressure to the piston speed, to generate the compensating signal to counterbalance the friction force and the parameter changes of plant.
    Through some experiments about the robustness against the payload, the positioning accuracy and so on, the validity of this control method has been confirmed.
  • 近藤 正
    1995 年 31 巻 12 号 p. 1978-1987
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    A neural network which has a self-selection ability of network structure is described. This neural network algorithm is a revised one of the neural network which can identify a nonlinear system whose structure is very large and complex. By using a criterion based on a prediction error, a neural network structure which is adapted for a complexity of a nonlinear system can be selected automatically. This neural network is applied to an identification problem of a rolling model and a physical and neural network combined model is identified. It is shown that the physical and neural network combined model gives better prediction results as compared with a conventional physical and statistical combined model.
  • 田中 一男, 池田 貴幸
    1995 年 31 巻 12 号 p. 1988-1995
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper presents stability analysis for a class of nonlinear systems and a design method of fuzzy controllers to stabilize the nonlinear systems. First, a stability condition is introduced in terms of Lyapunov stability theory. Next, stability conditions for a class of uncertain systems are given from the viewpoints of quadratic stabilization, H control theory and Linear Matrix Inequalities. Furthermore, some stability conditions are derived for a generalized class of uncertain systems. The derived stability conditions for a generalized class are used to analyze stability of nonlinear control systems designed via parallel distributed compensation. Some important issues for the stability analysis are remarked. Finally, we demonstrates controller design examples. It is pointed out in the examples that a conservative result of stability analysis can be avoided by introducing the singular value decomposition.
  • 吉井 伸一郎, 鈴木 恵二, 嘉数 侑昇
    1995 年 31 巻 12 号 p. 1996-2004
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    In the domains with the purpose of acquisition of adaptive strategies, Lamarckian mechanism which focuses on the inheritance of acquired characteristics is supposed to be an effective approach, because it is important to use the information which is acquired on the experiences in environment. However, the information useful for adaptation depends on the landscapes of problems to be applied. Therefore, in order to realize an adaptive Lamarckian evolution, it is needed some control mechanism considering which characteristics are desirable. In this paper, we examine that Lamarckian effects are dependent on problems such as Travelling Salesman Problem and Job Shop Scheduling Problem. Then, we construct an adaptive evaluation module for acquired characteristics, based on Genetic Supervision, and achieve an adaptive Lamarckian evolution by this mechanism. Genetic Supervision is a concept to determine the direction of adaptation innately. In this paper, we discuss the proposed method through the numerical experiments.
  • 門藤 博, 田所 嘉昭
    1995 年 31 巻 12 号 p. 2005-2010
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper proposes a pitch detection technique for transcription based on the extended-notch Fourier transform (E-NFT). The E-NFT can calculate Fourier coefficients with three data points per one frequency component. It can use to analyze a damping signal such as piano sound, and detect the start and end points of musical sounds to determine musical notes.
    First, we introduce the E-NFT and then describe the realization of the automatic transcription system using the E-NFT. To simplify the processing of the E-NFT, we decompose musical sounds to three octave frequency bands using bandpass filters. Last we demonstrate the automatic transcription of clarinet (monophony) and piano (triplephony) sounds of an elecronic keyboard. It can be shown that the proposed method has a small amount of calculations and good time response comparing with other methods.
  • 伊藤 誠, 稲垣 敏之
    1995 年 31 巻 12 号 p. 2011-2017
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    There exist several ways for conditioning belief functions (Bel) in the Dempster-Shafer theory of evidence. One of them is the conventional Dempster's rule of conditioning. In the conditioning, Bel is updated based on information claiming that a certain proposition, say B, is true. Evidence is not always complete enough to prove that the single proposition, B, is true. Dubois and Prade have proposed a belief updating mechanism in which several propositions (say, Bi's) can be possibly true. The belief updating is done by applying the Dempster's rule of conditioning to each proposition Bi.
    This paper clarifies problems which are caused by the use of the Dempster's rule of conditioning in the Dubois and Prade's mechanism. The problems are: (1) the Dempster's conditioning imposes strict (maybe too strict) requirement on a prior Bel, and (2) an updated Bel can be completely irrelevant to a prior Bel in some situation. We propose in this paper a new rule of conditioning to resolve the above problems. Compared with the conventional Dempster's rule of conditioning, our conditioning rule is less restrictive in the sense that a broader class of Bel's can be conditionalized. Our rule of conditioning thus enables the Dubois and Prade's belief updating mechanism useful in more general settings. Another distinctive feature of our conditioning rule is that a prior Bel can be conditionalized flexibly through an appropriate choice of a value for the continuous parameter in the conditioning rule.
  • Kazuyuki KONDO, Kenzo WATANABE
    1995 年 31 巻 12 号 p. 2018-2020
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    A novel switched-capacitor bridge for high speed and accuracy capacitance measurement is developed based on a two-step quantization. The first coarse quantization is performed by the charge-balancing M-bit A-D converter. The second fine quantization interpolates the first quantization into K bits by the single-slope A-D converter. An error analysis shows that the 12-bit resolution can be expected by MOS IC form. A prototype bridge using discrete components has confirmed the principles of operation.
  • 内冨 昭三, 隈崎 健二, 緒方 純俊
    1995 年 31 巻 12 号 p. 2021-2023
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper deals with the classification of Temperature patterns and the analysis of time varying images. The fractal dimension provides a usefull skill to understand the complexity of the topography.
    We prepared a set of data which includes the cooling process and heating process on the surface of a micom board, and computed the set of fractal dimension corresponding to the individual thermal image based on the power law.
  • 森 泰親, 秋元 博之, 中内 靖, 入道 真
    1995 年 31 巻 12 号 p. 2024-2026
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    In this paper, we propose a new design method of the generalized predictive control to locate all closed-loop poles in a desired region. As the tuning parameter, a weighting matrix of the cost function is used. The method is based on the bilinear transformation of the z-plane system and the application of the Routh-Hurwitz criterion on the transformed system. The effectiveness of the proposed design method is demonstrated by a numerical test.
  • 尾崎 信之, 舘 〓
    1995 年 31 巻 12 号 p. 2027-2029
    発行日: 1995/12/31
    公開日: 2009/03/27
    ジャーナル フリー
    This paper proposes a new algorithm to create virtual world by utilizing 2D scenery images in virtual reality. The algorithm based on Ullman's algorithm deals with perspective transformations, which is essential. We can construct only with several real images stored in advance without making 3D model by computer graphics.
feedback
Top