All issues
Volume 26, Issue 8
Displaying 1-19 of 19 articles from this issue
- |<
- <
- 1
- >
- >|
-
Tetsuo SAWARAGI, Osamu KATAI, Sosuke IWAI1990 Volume 26 Issue 8 Pages 854-861
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSCurrently, the fuzzy control techniques have been successfully attempted, but many of those have been restricted only to the automatization of local control tactics and have not taken into account more global control strategies. In this paper, at first a ship-maneuvering simulator is constructed that can be operated by the human operator as well as by the computerized fuzzy controller. Then, in order to improve the conventional fuzzy controller toward the more intelligent one, we propose an implementation algorithm for analyzing the skillful human operator's control action sequences with reference to his internal model, which is a symbolical and qualitative model of the ship's dynamics affected by his control operations. By relating each operation to his global planning structure determined by this model, the algorithm can identify a number of distinctive planning phases from the operation history, and the parameters for the fuzzy controller that are implicitly interwoven into the operator's action sequence can be specified. Finally, an intelligent fuzzy controller is constructed that utilizes the identified control parameters properly and adaptively according to the progress of planning phases.View full abstractDownload PDF (1168K) -
Toshiyuki TAKAGI, Shohachiro NAKANISHI, Keizo UNEHARA, Yasuyuki GOTOH1990 Volume 26 Issue 8 Pages 862-869
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSThe authors propose a system constructing fuzzy controllers which are automatically formed by means of two neural networks. One of these neural networks acts as a classifier of the dynamic responses of a controlled object system, and the other one as a knowledge for the dynamic characteristics of the object system. The fuzzy control rules in the controller are automatically formed from the information based on the neural networks. Moreover, the system provides a self-tuning mechanism using fuzzy inferences to adequately control the object system. Finally, the system proposed in this paper is applied to a frequency control of power systems.View full abstractDownload PDF (1018K) -
Tetsuya OKAMURA, Tohru MURAKAMI1990 Volume 26 Issue 8 Pages 870-877
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSAn adjustment support system using computer for start-up operation of cyclotron is being developed in order to support operators who, using their experiences and intuition, tune up dozens of components to maximite beam efficiency.
This paper describes a method of beam pattern diagnosis and adjustment guidance based on fuzzy relation as one function of the adjustment support system. The followings are the basic processes of this method.
(1) The computer extracts features of beam pattern which correspond with operators' criteria.
(2) The computer calculates causative possibility of each adjustment parameters by inverse operation of fuzzy relational equation, which formulates the relationship between criteria of beam pattern and adjustment parameters.
(3) The computer displays those causative probabilities in agraphical manner as a operation guidance. Not only the probabilities but also the criteria of beam pattern and their relationship are displayed in one figure.
(4) Operators select one adjustment parameter referring to the display and other informations such as history of adjustment, and they tune the parameter.View full abstractDownload PDF (1312K) -
Shigetaka HOSAKA, Isao TAKAMI1990 Volume 26 Issue 8 Pages 878-885
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSThe fault diagnosis method using the fault trees for diagnosis is developed, which describe the relationships among the faults, causes and observations. In this method, the causes and their probabilities are estimated by the fault trees for diagnosis and the data of observations.
The probabilities of causes are calculated with the failure rate of them, and with Fussell-Vesely importance which is use in the Fault Tree Analysis. So, it is possible to diagnose objectively.
This method applied to the draining pump plant fault diagnosis system. In this system, the failed equipment is searched using the fault trees, which is constructed from designer's and maintenance man's knowledge, and the data of observations. To make the human operator understand the results easily, the system presents the estimated causes, the probabilities of the causes, and the process of diagnosis in the picture of display devices. Also, the fault trees, the system block diagrams, and the pictures of the component are shown on the CRT display devices. In the block diagram, the failed parts flicker to show the causes easily for the operator. Moreover, a maintenance guide is shown. In this paper, the diagnosis method and the feature of human interface of the system are described.View full abstractDownload PDF (1939K) -
Nobuo SANNOMIYA, Chisaku OONISHI, Keiichi AKIMOTO1990 Volume 26 Issue 8 Pages 886-893
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSIn energy-consuming processes, various kinds of primary energy are consumed and, at the same time, are converted into byproduct energy. A part of byproduct energy is utilized efficiently for supplying necessary energy to another processes. Then we have a complicated interrelationship between generation and consumption of energy.
This paper deals with an optimal planning for supplying several kinds of byproduct gas produced in steel works as fuel of the boilers in a joint electric power plant. The generation of the byproduct gas changes with time according to the production schedule in steel works. On the other hand, a constant gas supply is desirable for the boiler operation in the electric power plant. An optimization technique is applied to coordinate such requirements and to find the optimal sequence of gas supply of which the steel works must notify the joint electric power plant in advance.
In our earlier paper, a mixed-integer programming model has been used in order to formulate the complicated operational constraints, and a decomposition algorithm has been applied to solve the problem. However, it is found out from further investigation that the algorithm requires much computation time for several numerical results based on real operation data. In this paper, a heuristic approach, called iterative improvement method, is proposed to save computation time. By examining numerical results, it is observed that the proposed algorithm is efficient, requiring less computation time than the branch and bound method. Further, the algorithm is evaluated for the possibility of making the gas supply plan on line.View full abstractDownload PDF (1091K) -
Tsuyoshi GOROMARU, Hideo HANAFUSA, Yoh YONEZAWA, Ryozo ITOH, Hayashi M ...1990 Volume 26 Issue 8 Pages 894-901
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSProper knowledge of static and dynamic behavior for refuse combustion process is necessary not only for operation of refuse incineration plants but also for designing the automatic combustion control system of the plants. However, the incineration process is very complex and the experimental measurement of dynamic characteristics is not easy because of fluctuation of physical and chemical characteristics in the refuse as a fuel.
This paper deals with the studies concerning the analyses of dynamic characteristics to the refuse incinerator, and then the automatic combustion control system aimed at stabilizing steam flow rate in the waste heat boiler of the incineration plants. In order to reduce fluctuation taking place in the steam flow rate, we propose effective utilization of refuse agitation, which is practically available for a quick recovery of the heat release rate in the furnace. As a result of simulation, we show that good responses to reference inputs are obtained for the PI control system with an effective agitating apparatus, in comparisonwith the conventional PID control system without agitating one. Furthermore, the refuse incineration process is shown by that of the system with time delay in state variables.View full abstractDownload PDF (983K) -
Tomio WATANABE1990 Volume 26 Issue 8 Pages 902-907
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSA machine which adapts to the speech of any human speaker and responds at an optimal utterance speed would be useful for the smooth exchange of information in human-machine communication, and the present study is concerned with the development of such a system. The author has already proposed a method for adapting machine conversational speed to speaker utterance speed. Based on sensory evaluation, this paper demonstrates that the speech activity adaptation itself is more essential for the smooth exchange of information than the adaptation to speech speed. Also, by analyzing the characteristics of speech activity estimation error as a function of observation time (talkspurt-silence period), the minimal number of talkspurt-silence periods required to estimate speech activity is determined. Finally, a conversational messaging system has been developed, which plays back stored speech segments controlled by the speech activity adaptation algorithm.View full abstractDownload PDF (876K) -
Kenji TANAKA1990 Volume 26 Issue 8 Pages 908-915
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSIn large-scale systems, systems failures mean not only physical failures but include failures in decision making proress. The systems failures can't be perfectly prevented in design level, so that it is desirable that the system improves itself by using information obtained from the occurred failures in operation level. To realize such a self-improvement process, we propose a ‘soft’ systems methodology for manmachine systems which consist of a engineering system and its management system.
The systems methodology is based on adaptive maintenance which is a kind of metasystem approach. The adaptive maintenance is an adaptive process to improve the inadequate cognitive model which is inconsistent for the real states. We discuss some adaptation properties of this maintenance and show that the internal adaptive process irrelevant to the changes of environment is indispensable to obtaining an adequate model, in addition to the external one for the changes. According to the proposed systems methodology, man-machine systems may develop through the adaptive maintenance that form a positive feedback.
The methodology is introduced by goal seeking model in consideration of systemic properties and satisfactory principle of evaluation so that it is also applicable to general control systems.View full abstractDownload PDF (1243K) -
Osamu KATAI, Hiroshi KAWAKAMI, Tetsuo SAWARAGI, Sosuke IWAI, Tadataka ...1990 Volume 26 Issue 8 Pages 916-923
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSAn EBL-based method for acquiring conceptual design knowledge in physical systems was proposed and implemented to a system based on the idea that such knowledge can be acquired by analyzing the structural features of existing designed objects.
Since any object can be interpreted via various design rationalities such as teleological, causal and economical ones, it can be modeled as a hierarchy which consists of designs goals, subgoals, structures and substructures toward attaining those goals. Through organizing domain-specific knowledge of the EBL system according to the above hierarchical model, this system yields a systematic explanation of how structures attain design goals. This results in a generalized version of Functional Diagram used in Value Engineering. From the Functional Diagram, general design knowledge of various levels can be acquired by a single positive instance of designed objects. The operationality of the acquired knowledge is then analyzed with reference to its modes of acquisition.
The acquisition method is implemented to a system using PROLOG.View full abstractDownload PDF (1171K) -
Shoji TOMITA, Shigeru FUJIMURA, Akira SUZUKI1990 Volume 26 Issue 8 Pages 924-926
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSWe propose a method to construct rules for a process monitoring system.
It is derived from analyzing knowledge-base development works of an intelligent training system for Distillation Tower Process we have built.
This method is useful to the tasks that there are many focusing points like a process observation.
In this paper, we show the actual construction of the monitoring rules, and examine a generic method of development.View full abstractDownload PDF (433K) -
Takeshi TSUJIMURA, Takayuki YASUNO, Tetsuro YABUTA1990 Volume 26 Issue 8 Pages 927-929
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSThis paper describes the learning method to create nonlinear transformation of inverse kinematics using a three-layer neurall network, showing effects of network parameters such as output function, number of hidden units, learning rate, momentum constant through simulations.View full abstractDownload PDF (329K) -
Riichi ADACHI1990 Volume 26 Issue 8 Pages 930-932
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSSome orthogonal weights are selected from pseudorandom numbers and accumulated successively to cover whole learning space in the two-layer feedforward topology for pattern classification.View full abstractDownload PDF (324K) -
Shigenobu KOBAYASHI1990 Volume 26 Issue 8 Pages 933
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSDownload PDF (155K) -
Masashi MIZUNO, Toshio ENDO1990 Volume 26 Issue 8 Pages 934-941
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSThe fluorescent magnetic particle test is quite common for detecting surface defects of square steel billets. While this method has high sensitivity, it has low consistency and inefficiency because of detection by the naked eyes. This paper reports the development of an automatic system for this test by using the high sensitive plumbicon camera and the image processing technology. An experimental equipment was made that consists of a hardware with plumbicon camera, frame memory, one CPU and CRT, and software for basic image data processing and special processing to extract the only signals, as defects, which have sharp convex shape in transverse direction to billet length. Consequently, detection rate of seam defects deeper than 0.3mm was 83%, and over-detection rate was 78%. This result is not satisfactory for practical use. Accordingly, to raise up detection rate above 90%, and to lower over-detection rate below 20%, sensitivity of pick-up system was improved and defect length judgement processing was added to the above-mentioned software. And defect detection efficiency was also raised up by parallel operating ten CPUs for image processing. Improved equipment was installed in practical fluorescent magnetic particle test plant and examined to check the performance. Consequently, defect detection rate has gone up to 98%, and over-detection rate down to 17%. This result is sufficient enough for practical use.View full abstractDownload PDF (1938K) -
JIACHUN Fan, Kenji KUROSU1990 Volume 26 Issue 8 Pages 942-947
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSIn this paper, we treat the repetitive control for linear mechanical systems in the discrete domain. First, we investigate the stability of the repetitive control system using the basic feed-back control theorem and note that the stability condition can be related to the positive real condition. We next propose an available control algorithm based on the concept of positive real property for the plant whose transfer function is not positive real (e.g. a strictly proper transfer function). The value of parameters in the algorithm can be found by using the Nyqist plots. We also discuss the characteristics of the designed system and present some results of hybrid computer simulations for a mechanical system. It is shown that the designed repetitive control system is insensitive to changes in plant parameters and the error converges in spite of the high frequency of the reference signal.View full abstractDownload PDF (717K) -
Hongye DU, Yoshikazu HAYAKAWA, Seizou FUJII1990 Volume 26 Issue 8 Pages 948-955
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSFor centralized control systems, an elegant solution to the simultaneous stabilization problem has been obtained by showing that the problem of stabilizing two plants with a common controller is equivalent to that of stabilizing one plant with a stable controller (called the strong stabilization problem). As far as the authors know, however, there have been only few studies on the simultaneous/strong stabilization problem for decentrally control system.
In this paper, by using the parameterization of all the decentralized stabilizers that the authors have obtained already, a relationship between the decentrally simultaneous stabilization and the decentrally strong stabilization is derived. Then it is shown that in general, the decentrally simultaneous stabilization is not equivalent to the decentrally strong stabilization, but is reduced to the strong stabilization with structurally constrained controllers. Furthermore, a simple necessary and sufficient condition for decentralized systems with 2 inputs-2 outputs-2 stations to be strongly stabilized is given.View full abstractDownload PDF (886K) -
Masakazu SUZUKI, Kiyotaka SHIMIZU1990 Volume 26 Issue 8 Pages 956-963
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSArray algebra and array optimization technique are concisely reviewed. The array is a natural generalization of the vector-matrix and it provides the more convenient tool than the vector-matrix does. It is shown that a wide class of distributed parameter systems (DPS) is approximately transformed into array dynamical systems in a unified manner. In the array system formulation, the well-systematized lumped parameter theory is available, at the same time, the spatial structure of the DPS is preserved and it is easy to utilize the SpatialState-Distribution-Pattern (SSDP) to control. The optimal regulator problem for the linear array systems is formulated and controllers for the problem are optimally designed by use of array optimization technique. The resultant controller realizes the most basic SSDP feedback in the sense that the structured states are feedbacked. The array formulation, in other words, the discretization of continuous distributions into forms suitable for computer management with their spatial structure preserved, provides a methodology for SSDP analysis and design, and it can be applied to various fields other than DPS approximation.View full abstractDownload PDF (1009K) -
Jun-ichi TANJI, Mitsuo KINOSHITA1990 Volume 26 Issue 8 Pages 964-971
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSA multi-input/multi-output self-organizing fuzzy controller for multivariable cooperative control systems is presented. One developmental problem to be tackled is the interactive responses of the controlled variables when the controller is applied to a real multi-input/multi-output system. In the fuzzy controller described, linguistic decision rules which authers name “cooperative rules”, are utilized for regulating interactive responses of the controlled variables instead of a cross controller.
The analytical basis for the multi-input/multi-output self-organizing fuzzy controller is described using a modified scheme of model reference adaptive control systems. A detailed formulation is given regarding specific response evaluation and rule modification in the cooperative rule learning process of the controller. The schematic model presented here includes the former model for a single-input/single-output controller, which is a general extension of the self-organizing fuzzy controller from a scalar system to multivariable systems.
Control simulations of a BWR plant and an HTGR (experimental reactor) plant are presented showing the performance of this fuzzy controller. Results show the cooperative rules are effective for improving the control performance of multivariable systems.View full abstractDownload PDF (1025K) -
Kaname YOSHIDA1990 Volume 26 Issue 8 Pages 972-979
Published: August 31, 1990
Released on J-STAGE: March 27, 2009
JOURNAL FREE ACCESSWe often encounter the optimization problems with variant coefficients or the problem whose structure are same as each other but different only coefficients, and are usually required to solve them quickly. In the present paper we deal with the geometric programming problems with variant coefficients. In geometric programming, if the degree of difficulty of the given problem is 0, the dual problem to the given problem has only one feasible solution i.e. optimal solution excluding coefficients which can be easily obtained by solving a simultaneous linear equation.
So we try to apply the approximate decomposition method for the given problem and decompose it into a set of 0 degree of difficulty subproblems. A equation, whose solution is the approximate optimum of the given problem, can be derived from the optimums of the subproblems. The equation is constructed with the variant coefficients, therefore we can use the equation to obtain an approximate optimum by substituting any values into the coefficients.
Moreover we can easily and quickly solve the equation. At last we check our method to apply some numerical examples and show the usefulness of this method.View full abstractDownload PDF (997K)
- |<
- <
- 1
- >
- >|