巻号一覧
29 巻, 10 号
選択された号の論文の21件中1～21を表示しています
- |<
- <
- 1
- >
- >|
-
山田 光1993 年 29 巻 10 号 p. 1125-1132
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, a differential non-linearity (DNL) error analysis of successive approximation ADC is considered. The largest cause for the DNL of successive approximation ADC is due to the deviation in the output analog difference of included DAC between two adjacent code from the ideal value. The DNL error is the most difficult error to deal with since it cannot be eliminated by adjustment. The very high differential linearity is needed in the field of radiation pulse height analysis (PHA). Because of its high differential linearity, the Wilkinson type ADC is still extensively used for PHA. But it has very long conversion time. The successive approximation type ADC has quite short conversion time but has large DNL. The “Sliding Scale Method” by Gatti was introduced about 30 years ago to compensate the channel width inaccuracies of successive approximation ADC.
First, the DNL of successive approximation ADC is analyzed using admittance matrix of ladder-type resistance network. And the characteristics of channel width are investigated. After that, the following properties become evident.
(1) The resistance errors cause DNL at the specially fixed every 2n channels.
(2) The small resistance error carse considerably large DNL.
(3) Each of specially selected channel groups has uniform channel width, in spite of 2n weighted resistance error.
Finally, a new procedure to measure a smoothed curve of probability density function avoiding the DNL error is proposed by means of these results of investigations. The usefulness of this procedure is shown by computer simulation.抄録全体を表示PDF形式でダウンロード (1658K) -
杉本 隆夫, 川口 忠雄, 藤懸 洋一1993 年 29 巻 10 号 p. 1133-1140
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThe mechanical characteristics of a machine or parts depend on the material properties. Especially, hardness is an important index for the design of a machine. The Vickers hardness number is defined by indentation of the surface of a test piece by an indentater of a diamond-pyramid shape. Usually diagonal lines that indicate the indentation size are several micrometers to several hundred micrometers depending on the hardness of the material. The conventional method of measuring the indentation is applying to only the specular-polished specimen in an automatic measurement indentation. The new Vickers hardness testing system has been developed by using similar to the visual inspection of hardness on the image processor with auto-tuning binary level depending on a image data. Furthermore, the new system can not only measure a roughly-polished surface specimen while keeping the required measurement accuracy defined in Japanese Industrial Standard (JIS), but can also treat one indentation within 20 seconds including making the indentation, focusing time and image processing time. It is three times faster than the conventional automatic Vickers hardness testing system, and also six times faster than the conventional visual inspection.抄録全体を表示PDF形式でダウンロード (2638K) -
小野 敏郎, 孫 建新1993 年 29 巻 10 号 p. 1141-1146
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThis paper extends the authors' previous method on linear system identification which is independent of the dynamics of the measuring devices used for observing the input-output signals of the object system. The essence of the method lies in the double measuring concept to generate the suitable input-output data for identification by means of the convolution integral of a different pair of input-output signals obtained through double measurements.
The following extension and discussion are given in this paper:
1) the method is extended to be applicable to the linear system whose input-output signals are different kinds of physical quantities; and
2) another method is examined of using the observed input-output signals obtained through double measurements, which conducts the conventional identification procedure twice, to be feasible.抄録全体を表示PDF形式でダウンロード (903K) -
田中 幹也, 鈴木 隆1993 年 29 巻 10 号 p. 1147-1155
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThe design of continuous-time model reference adaptive control system (MRACS) is usually based on the reduced-order model of the plant which accompanies various modelling errors. Robustness of such a MRACS scheme which is designed by neglecting modelling error becomes very important problem. Authors declared robustness of the conventional MRACS scheme to modelling error arising from inexact pole-zero cancellation. However, in that case, the relative degree of the transfer function of the reduced-order model of the plant is limited in one.
In this paper, we propose the continuous-time MRACS scheme robust to modelling error arising from inexact pole-zero cancellation. This scheme relaxes the restriction on the relative degree of the transfer function of the reduced-order model of the plant. This scheme is characterized by using the auxilialy signal, which is synthesized from the output error between the plant and the reference model, to ensure robustness of the MRACS scheme to the modelling error.抄録全体を表示PDF形式でダウンロード (1149K) -
小林 伸明, 櫻井 光広, 加藤 明, 中溝 高好1993 年 29 巻 10 号 p. 1156-1162
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, we discuss the property of the limiting form derived from a standard discrete-time linear optimal regulator problem by letting the weighting matrix tend to zero. The role of the unobservable subspace which is obtained by using feedback gain in the limit is in detail studied for nonminimum phase systems. Then it is shown that there exists an explicit relationship between limiting form of discrete-time linear optimal regulators and the decoupling control problem.抄録全体を表示PDF形式でダウンロード (1317K) -
中嶋 淳, 石原 正, 杉本 欽一, 猪岡 光1993 年 29 巻 10 号 p. 1163-1170
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThe method of inequalities proposed by Zakian is an effective computer aided design method to find a controller satisfying various design specifications. Sugimoto et al. have recently developed a concise CAD systems based on the method of inequalities. However, this CAD system can not provide a design accounting the effect of pole-zero cancellations. We consider improvement of this drawback. First, for unity feedback systems with specified steady state characteristics, we give a simple pole-zero placement algorithm which can avoid specified pole-zero cancellations. Replacing the algorithm used in the CAD system developed by Sugimoto et al. with this algorithm, we construct an improved CAD system based on the method of inequalities. A design example is presented to illustrate the effectiveness of the proposed CAD system.抄録全体を表示PDF形式でダウンロード (1567K) -
申 鉄龍, 田村 捷利1993 年 29 巻 10 号 p. 1171-1175
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーQuadratically stabilization approach has been often used to solve a robust stabilization problem of a plant with structural uncertainty. A norm of a state vector of quadratically stable (q-stable) system is bounded by an exponential function for all bounded time varying perturbation. Thus, the approach seems to be effective to solve the problem.
However, q-stability of a system with unstructural uncertainty, which is described in frequency domain, has not been fully discussed. In this paper, we will show a sufficient condition of the q-stability of a system with both uncertainties, time varying parameter perturbation and frequency domain uncertainty. Using the condition a design of q-stabilizing controller is presented. An numerical example is given.抄録全体を表示PDF形式でダウンロード (819K) -
駒田 諭, 石田 宗秋, 大西 公平, 堀 孝正1993 年 29 巻 10 号 p. 1176-1183
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThis paper proposes a new robust hybrid position/force control of robot manipulators. The proposed method controls the second derivatives of control variables such as position and force in task coordinate system in order to realize robust and high response control. In order to realize this, the disturbances are estimated by a position-based disturbance observer and a force-based disturbance observer in task coordinate system, and are compensated by feeding back the estimated disturbances. The proposed method requires less computation effort and is robust against the disturbance and parameter variations.
The position-based disturbance observer has been proposed to linearize robot manipulators and realized the robust position control. However, when force control is performed, the force response is influenced by not only the nonlinearity of robot manipulators but also the characteristics of environment on which the force is imposed. Therefore, the force-based disturbance observer is developed to realize robust force control. A controller robust against the disturbance and parameter variations is realized by using the position-based disturbance observer and the force-based disturbance observer on performing the position control and the force control respectively. The effectiveness of the proposed method is shown by experiments using a direct drive robot.抄録全体を表示PDF形式でダウンロード (1693K) -
山口 功, 木田 隆, 上野 誠也, 田中 正樹1993 年 29 巻 10 号 p. 1184-1190
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThis paper discusses the optimal hand trajectories generation for a space robot manipulator. On the topic, the analytical optimal solution exists only for a simplified mathematical model. For the practical case where mass and moment of inertia of links are taken into account, large-dimensional non-linear two point boundary value problem must be solved to obtain the optimal hand trajectory. In this numerical approach, iterative calculation is required to improve the accuracy of solution that takes too much time and cost. In this paper new optimization algorithm by the method of Dynamic Programming (DP) is proposed to solve optimal hand trajectories for the general space robot model with a two-joint arm in the two dimensional coordinate system. This method has the advantages reducing the computational time substantially. Comparison with the numerical results by the method of optimal control theory are made to demonstrate the accuracy of the proposed method.抄録全体を表示PDF形式でダウンロード (1418K) -
古賀 雅伸, 古田 勝久1993 年 29 巻 10 号 p. 1191-1198
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThis paper describes a high-performance computer aided control system design language MATX which integrates numerical analysis, matrix computation, and symbolic manipulation in an easy way to use. One-variable polynomials, rational-functions, and symbolic matrices comprised of those elements are essential to the study of system theory. We take advantage of this fact and combine numerical computation and one-variable symbolic manipulation facility. Since MATX allows the definition and manipulation of these objects in a natural symbolic fashion, problem solutions are expressed almost exactly as they are written mathematically. Therefor, the user can solve a lot of control system design problems in much shorter time than describing programs in other programming languages. MATX will play an important role in developing and examining new design algorithms in system theory.抄録全体を表示PDF形式でダウンロード (1037K) -
小金澤 鋼一1993 年 29 巻 10 号 p. 1199-1202
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThis paper deals with the development of a fast algorithm of QR decomposition based on the Householder transform. This paper also shows the correspondence between the recursive QR decomposition based on the proposed algorithm and one based on the Givens rotational matrix. This fact implies a ratinale of the proposed method on the theory of orthogonal linear transform.抄録全体を表示PDF形式でダウンロード (715K) -
根岸 直樹, 相吉 英太郎, 羽深 嘉宣1993 年 29 巻 10 号 p. 1203-1212
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThere exist many learning algorithms for finding the synaptic weights of a layered neural network so as to minimize the output errors in response to the plural training data. The most well-known learning algorithms are the backpropagation procedures, in which there are two manners of minimizing the plural error functions in practice. One is minimizing the sum of all error functions totally, and the other is making a reduction of only one error function chosen temporarily and cyclically at each iteration. The former does not always keep the balance and the efficiency in reduction of the plural output errors, and the latter does not guarantee the convergence from the standpoint of numerical analysis.
In this paper, we then adopt the learning manner of minimizing the maximum valued function among the all error functions in order to reduce the output errors based on all training data uniformly, and propose a learning algorithm using a non-differentiable optimization method with the generalized gradient in order to guarantee the convergence. An application to the simple recognition problem with plural training patterns and its simulation results show efficiency of our learning manner and algorithm.抄録全体を表示PDF形式でダウンロード (1527K) -
辻 敏夫, 市延 弘行, 伊藤 宏司, 長町 三生1993 年 29 巻 10 号 p. 1213-1220
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーThe present paper proposes a method to estimate the motion intended by a subject from his EMG signals using error back propagation typed neural networks. Estimation of the motion from the EMG signals is useful for means of human interface in such fields as control of multi-functional powered prosthesis, teleoperation of robot manipulators, virtual reality.
The neural network used in the method can learn a mapping from the EMG patterns measured from four pairs of electrodes to six motions of forearm and hand intended by the subject. The experimental results for several subjects including an amputee show the following: 1) the method can discriminate six motions with the accuracy about 90 percent for several electrode locations, 2) ill-discrimination can be decreased by suspending discrimination using entropy of network output, 3) the neural network can adapt to some dynamic variations of the EMG patterns using on-line learning, and 4) utilizing frequency characteristics as well as amplitude characteristics of the EMG signals reduces the number of iterations required for learning convergence.抄録全体を表示PDF形式でダウンロード (2335K) -
伊能 教夫, 藤原 宏, 梅谷 陽二1993 年 29 巻 10 号 p. 1221-1226
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーLiving bones have the adaptive function to change the shape and the structure to outer forces exerting on the bones. The function interests even engineers because the order of the whole system is reformed and maintained properly by the way that each element of bone senses the mechanical states produced by outer forces at each position and adjust its material strength by itself. It is a kind of Autonomous Distributed Systems.
This paper describes a computational method to generate mechanical structures based on the adaptive function of bone. The original idea was proposed as the Growing-Reforming method. But the application of a three-dimensional problem has not yet been examined.
We examined the possibility of application to the three-dimensional problem by changing Young's modulus of the elements referring to the Growing-Reforming method. A three-dimensional cantilever on which terminal outer forces are exerting was set for this study. First, we used a methematical model that the target stress value was constant for Young's modulus. The computer simulation showed that the model did not converge to the desirable mechanical condition.
Next, we discussed the reason why the model diverged using basic equations of the finite element method. And we pointed out that the target stress value should be changed for the present Young's modulus of each element because the target stress value reflects the material strength of each element. We settled that the target stress value has a linear relationship to Young's modulus. The computer simulation showed that the new model generated an appropriate distribution of Young's modulus which can support the force efficiently.
We also examined a case of non-linear relationship between the target stress value and Young' modulus. The simulation result showed that the model generates a clearer three-dimensional structure.抄録全体を表示PDF形式でダウンロード (2333K) -
梅本 敏孝, 青島 伸治1993 年 29 巻 10 号 p. 1227-1231
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーIn the transcription system, peak frequency extraction is most important. Iguchi et al. have developed interpolation method in complex spectra. It is necessary for this method to calculate long data FFT to make frequency correction in low frequency range. So this method is not suitable for fast rhythm song. We have developed the adaptive method, by which frequency correction is not necessary. In this paper, we also use short time spectrum analysis method by complex first order system. We constructed transcription system and compared three peak extraction methods in calculation time and accuracy.抄録全体を表示PDF形式でダウンロード (1206K) -
稲葉 宏幸, 板倉 安正, 笠原 正雄1993 年 29 巻 10 号 p. 1232-1234
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーA new method of spatial filtering velocimetry is proposed for measuring the velocity vector of a random flow. The method uses three spatial filters which are located in a triangle, while the conventional methods uses two filters, that are perpendicular to each other. An estimated velocity vector can be obtained by a simple transformation, a weighted summation of the outputs of the central frequencies of the three spatial filters under the criteria of the least mean square error. The results of the computer simulation shows that this method realizes a highly accurate measurement of the velocity vector of a random flow.抄録全体を表示PDF形式でダウンロード (326K) -
今野 和彦, 奥山 大太郎, 中鉢 憲賢1993 年 29 巻 10 号 p. 1235-1237
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーA sound velocity measurement method with a small quantity of liquids which is measuring the transmission coefficients of sound pressure is newly developed. The principle and system are described and the experiments with water and ethyl alcohol of 0.1∼0.3cc has been demonstrated. From the results, measured velocities are coincide with 0.2% with the usual method have been attained.抄録全体を表示PDF形式でダウンロード (290K) -
岩井 善太, 足立 克己, 水本 郁朗1993 年 29 巻 10 号 p. 1238-1240
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーIn this paper, we consider a problem of a model output following control based on output feedback. It is shown that, if the plant satisfies the so-called ASPR condition, the model output control system can be constructed by using constant output feedback. It is also derived that the addition of relay element improves tracking performance.抄録全体を表示PDF形式でダウンロード (386K) -
香川 利春, 藤田 壽憲, 山中 孝司1993 年 29 巻 10 号 p. 1241-1243
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーWe propose the nonlinear model of the rubber actuator which includes the variation of compliance with the volume change. It is clear that the model is adequate to express dynamics of an artificial muscle.抄録全体を表示PDF形式でダウンロード (304K) -
村田 忠1993 年 29 巻 10 号 p. 1244-1246
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーA new method with the fuzzy inference to an interval estimation of a part reliability (MTBF) is proposed. The width value of the interval estimation to a part MTBF could become more narrow at same failure data and same confidence coefficient.抄録全体を表示PDF形式でダウンロード (849K) -
井田 正明, 片井 修1993 年 29 巻 10 号 p. 1247-1249
発行日: 1993/10/31
公開日: 2009/03/27
ジャーナル フリーMultiobjective linear programming problems are considered, where coefficients of objective functions are given by intervals. For this problem, two kinds of efficient solutions are known, namely, necessarily efficient and possibly efficient one. This paper proposes discrimination methods for those efficiencies.抄録全体を表示PDF形式でダウンロード (406K)
- |<
- <
- 1
- >
- >|