-
Toshiharu KAGAWA, Masashi SHIMIZU
1987 Volume 23 Issue 10 Pages
1003-1008
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper deals with how to correct the indication of the float-area type flow meter when line pressure changes. Float-area type flow meters are often used for small range flow rate measurements. In this range of measurement, flow rate characteristics of the meter are affected by the pressure drop in the small annular slit between float and tube. According to the analysis, new correction method for float-area type flow mether is introduced. The experimental data are in satisfactory agreement with the calculated results of the flow meter.
View full abstract
-
Takanori NINOMIYA, Takashi HIROI, Yasuo NAKAGAWA
1987 Volume 23 Issue 10 Pages
1009-1016
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Recently many types of practical robot vision systems have been developed. However, most of them are based on the 2-D image processing technique and can recognize objects only on a known plane. On the other hand, much fundamental research effort on 3-D recognition has been made, but achieved results are still time-consuming and impractical.
In this paper, we describe the compact 3-D sensor for robots. 3-D information about objects can be detected by the light section method. Projecting a slit light onto the object and taking its image from the oblique direction, the distances from a point on the slit light plane to the light stripes, which are intersections of the object and the slit light, can be measured based on the triangulation. The proposed sensor projects two slit lights one by one, which are arranged like X-shape. The light stripes neighboring the intersection of two slit lights in the taken images are guaranteed to be on a continuous surface. Therefore, once the sensor is directed to the desired surface just roughly, the orientation and the distance of the surface in the 3-D space can be detected through the calculation of the distances to the points neighboring the intersection of the slit lights, without moving the sensor.
The developed sensor consists of two slit light sources and a solid state TV camera. Infrared LED chips are used as the light source and Camera Control Unit is separated from the sensor itself. They enable us to realize a light-weight and compact sensor which is suitable for being mounted on a robot arm. The size of the sensor is 200mm
W×57mm
H×75mm
L and the weight is 400g. According to the experimental results, the repeatability of the detected position of the cylindrical object is σ=0.04mm for the
x-y coordinate, σ=0.09mm for the distance and σ=0.2° for the orientation and the acquisition time is about 0.1 second.
View full abstract
-
Toshihiro TSUMURA, Masafumi HASHIMOTO, Akira ISHIDA
1987 Volume 23 Issue 10 Pages
1017-1023
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
We now develop an active type of positioning system for roboted vehicle moving on three dimensional space by use of a laser beam and a corner cube (retroreflector). In this system a vehicle, first, searches a corner cube target fixed on a ground reference point and emits a laser beam toward the target. When the laser beam captures the target, the vehicle geometrically determines the own position and attitude based on the directions of the target measured by the laser beam.
In order to continue the accurate position finding even during movement of the vehicle, the above positioning system requires a laser tracking control which always directs the laser beam toward the center of corner cube target.
This paper presents a laser tracking control method. First we show that the directional deviation of the laser beam relative to the center of target, which is needed for accurate tracking control, is easily determined by the active use of unique reflectional characteristics of the corner cube. Then we design the laser tracking controller on which the feedback gains are obtained based on LQ regulator theory. Further we show the capability of our proposed tracking method through some experiments using a proto-type system.
View full abstract
-
Ichiya SATO, Takao YONEYAMA
1987 Volume 23 Issue 10 Pages
1024-1029
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper describes the development of a machine condition diagnosis system using acoustic emission (AE) techniques. The system is applicable to rotating machinery such as steam turbines, generators and rolling mills.
The basic algorithm of the system discriminates between various abnormal conditions of machines using the relationship between AE waveform characteristics and frequency characteristics of an envelope detection signal of the AE. In order to analyze the waveform characteristics, the following waveform parameters are calculated: mean value, AE event, duration time, rise time, and AE energy. The frequency characteristics of the envelope detection signal of the AE are obtained by fast Fourier transform (FFT) calculations. Using the results of the waveform analysis, the AE signal is identified as either a continuous type or a burst type. Then using the results of the frequency analysis, the AE signal is identified as either a wide band type or a narrow band type, and the latter is further classified as either a rotation tuned type or an untuned type. These categorize the AE signal by six types of abnormal conditions.
Suitable application software for a diagnosed object are provided from rubbing diagnosis, rotor crack diagnosis and journal bearing diagnosis software which were developed using the algorithm described above. Thus, abnormal conditions and source locations can be monitored in detail. The system uses an interlocutive form with the operators, who can use graphic, data, or trend displays for the diagnosed results.
View full abstract
-
Smell Identification Model by GMDH
Masayoshi KANEYASU, Tomoji OISHI, Akira IKEGAMI, Hideo ARIMA, Shoichi ...
1987 Volume 23 Issue 10 Pages
1030-1037
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper describes an analytical method and experimental results of identifying and quantifying smells using an electronic system composed of an integrated sensor and a microcomputer. The integrated sensor with six different elements on an alumina substrate was fabricated by using thick film techniques. The elements are kept at around 400°C by a Pt heater mounted on the sensor back. Since each element was made from different semiconductor oxides, they possess different sensitivities to material odors and the integrated sensor can develop specific patterns corresponding to each odor as a histogram of conductance ratios for each element. The microcomputer identifies the scent on the basis of similarities calculated by comparing standard patterns stored in the memory and a sample pattern developed by the integrated sensor. The scent is then quantified by using the sensor element with the highest sensitivity to the smell identified. A model for the smell quantification in state of mixture of different two scents has been developed. The model is essentially based on the GMDH algorithm, and is in need of some data sampled from the integrated sensor in different temperature controlled by a voltage of the Pt heater. The experimental results show that smells can be successfully identified and quantified with the electronic system.
View full abstract
-
Shigenori OKUBO
1987 Volume 23 Issue 10 Pages
1038-1045
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
In this paper we discuss a design of model following control systems (MFCS) for discrete time systems and the robust stability conditions. In case that a controlled object has unstable zeros, it has been difficult to design a stable MFCS by using conventional methods. The reason is that a characteristic polynomial of the closed loop system involves a characteristic polynomial of unstable zeros. In this paper we use the method to involve a zero characteristic polynomial in control input signals. This method is one of a signal synthesis. Under the condition that all of the future values of reference inputs are known, we can synthesize bounded signals which are necessary for control inputs. Futhermore we discuss robust stability conditions when the transfer function of a controlled object has errors. We use a formulation whose errors exist in a numerator matrix and a denominator matrix of a right coprime decomposition as additive forms. Such a formulation does not lose the generality, but makes quantitative analysis of robustness easy. From Rouche's theorem and norm calculations of matrices, we obtain robust stability condition. In robust analysis of discrete time system by this paper's method, there is no limitation of polynomial degree of error matrices. This property does not exist in continuous time systems. So we can expand this paper's robust stability conditions to the case that error matrices are general analytic functions.
View full abstract
-
Tadashi ISHIHARA, Kazuhiro ITO, Hiroshi TAKEDA
1987 Volume 23 Issue 10 Pages
1046-1052
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Frequency-domain properties of Mita's digital optimal servosystem extended to general input delay case are investigated. It has been pointed out that the closed loop transfer function matrix of this optimal servosystem contains explicitly that of the optimal regulator used in the derivation. In addition to this interesting property, this paper shows that the guaranteed stability margins and the disturbance rejection property of the optimal servosystem have explicit relations to those of the optimal regulator. The influence of the choice of the cost function and that of the input delays on these properties are clarified by use of the well-known asymptotic properties of the optimal regulator. These results provide useful guide in the design of the digital optimal servosystem.
View full abstract
-
Kazuo AIDA, Toshiyuki KITAMORI
1987 Volume 23 Issue 10 Pages
1053-1059
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
In this paper, a design method of preview control system which does not use LQ approaches is presented.
One of the authers has proposed the design methods of IPD and PID type control system based upon partial knowledge about plant dynamics. In these design methods, σ
i which denotes the first order moment of the impulse response of every control loop in multi-input-output control system is used as a parameter representing the rise time.
A preview controller is designed such that it overtakes the delay of response to the reference signal by σ
i. This preview action has the following effects.
(1) For IPD type control system, the tracking and steady-state errors are improved.
(2) For I
2PD type control system whose design formula is given here, in addition to the effects of (1) the overshoot of response to the step reference signal is suppressed.
Two numerical examples are presented to demonstrate the above effects.
View full abstract
-
Tatsuya MASUDA, Katsuhiko FUJII
1987 Volume 23 Issue 10 Pages
1060-1067
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Recently the analytic hierarchy process (AHP) developed by T.L. Saaty attracts attention as one of the particularly useful vehicles for allocating resources, analyzing policy impacts, and making a decision.
This paper presents some useful theorems for the sensitivity analyses of priorities (i.e., relative importances) used in the AHP. These theorems are derived from the principle of hierarchical composition which is expressed in the form of reachability matrix. Using the theorems, we can easily calculate the degree of effects caused by local or global changes in the priorities of some criteria, and also we can examine about the possibility of rank reversal among alternatives.
This paper, also, presents an application of the theorems to the problem of house selection, and demonstrates their effectiveness.
View full abstract
-
Qingxing LIN, Atsunobu ICHIKAWA
1987 Volume 23 Issue 10 Pages
1068-1075
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
To introduce colors in Petri nets is believed to be an effective mean to reduce the size of Petri nets when the same type of devices are involved in a system. The notion of colored Petri net (CPN) can be defined in various ways depending upon how the color matchings are specified in the firing conditions. The concepts of the structural color matching and the firing color matching are introduced in this paper in order to define a new type of CPN suitable for representing the system consists of devices which can handle multiple processings. The sequencial control system (SCS) is represented by the CPN from its logical element level to the system function level in a single type of model. Equivalence of description power of the CPN and PN is proved, the way of representing SCS by the CPN is described and the properties of the CPN when it represents SCS are discussed.
View full abstract
-
Kenji ONAGA, Qi-Wei GE, Norihiko ONO
1987 Volume 23 Issue 10 Pages
1076-1083
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper proposes a technique of designing a periodic firing sequence σ=δδ… and its enabling initial marking
MI for a bounded live Petri Net PN=(
P, T, E, MI) with an aim of minimizing the number of tokens in
MI, |
MI|_??_Σ
MI(p).
Let
A=A+-A- be the place-transition incidence matrix of the PN. It is well known that if the initial token distribution
MI is sufficiently abundant (namely
MI=A-X0) then it is possible to construct a periodic firing sequence σ=δδ… with δ=
X0 satisfying
AX0=0. This paper presents a method of constructing multi-round firing of the period, δ=δ
1δ
2…δ
k, and its enabling initial marking
MI with an aim of minimizing the token number |
MI| through use of a partial ordering of the transition, induced by a given feed-back edge set
F, and a decomposition
X0=Y1+Y2+…+Yk such that in the
i-th round δ
i transition
t fires
Yi(t) times in succession in accordance with the ordering
J.
View full abstract
-
Kiyoshi WADA, Miyoichi EGUCHI, Setsuo SAGARA
1987 Volume 23 Issue 10 Pages
1084-1090
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
It is well known that generalized least squares (GLS) requires a large amount of memory and time for computation, although the estimates are accurate. One way to avoid this disadvantage is a usage of modified GLS in terms of correlation function (GLSC). In the case of small data size, however, the estimated results are unreliable even for the stationary sequences postulated in GLSC.
In the present paper, we propose a new GLS based on a product moment matrix to remove these defects. The proposed algorithm gives the same result as GLS but reduces computational time. It is especially effective in the case where the input is nonstationary and many iterations are required to get the good results.
View full abstract
-
Shigeru UCHIKADO, Kimio KANAI
1987 Volume 23 Issue 10 Pages
1091-1093
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
-
Kazuo AIDA, Toshiyuki KITAMORI
1987 Volume 23 Issue 10 Pages
1094-1096
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
-
Joe IMAE, Kyojiro HAKOMORI
1987 Volume 23 Issue 10 Pages
1097-1099
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
-
Joe IMAE, Kyojiro HAKOMORI
1987 Volume 23 Issue 10 Pages
1100-1102
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper proposes a new second-order algorithm for solving continuous time optimal control problems, and proves that accumulation points generated by the present algorithm, if exist, satisfy Pontryagin maximum principle, a necessary condition for optimality.
View full abstract
-
Masanori ITO, Kenichi ABE
1987 Volume 23 Issue 10 Pages
1103-1105
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
-
Jongkol NGAMWIWIT, Noriyuki KOMINE, Michihiko IIDA, Yasuo NOZAKA
1987 Volume 23 Issue 10 Pages
1106-1108
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
-
Fumio MATSUDA, Kazuhiko SUZUKI, Masafumi ITOH, Shuzou HATTORI, Minoru ...
1987 Volume 23 Issue 10 Pages
1109-1111
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper discribed the diviation of the parameters on a binary image and its characteristic. The value of the deviation of parameters gives the order; the minimum and maximum of the distance from the gravity to the perimeter, area and perimeter.
View full abstract
-
[in Japanese]
1987 Volume 23 Issue 10 Pages
1112-1113
Published: October 30, 1987
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS