Because of the characteristic features of optical fibers, their transmission loss varies for an instant when a bend or a torsion is caused to them by engineering works, etc., and, in the worst case, this may affect the communication system at work.
In the present optical fiber communication systems, various services are transmitted through a single optical fiber as a multiplex transmission line. Therefore, the deterioration in the quality of a circuit, which may be caused by engineering works, has an extremely great influence.
With a view to finding the method to avoid such a serious situation, we carried out an optical fiber engineering work as an experiment which was so designed as to reproduce the condition of an actual engineering work as closely as possible. As a result, the characteristic of the loss variation of the existing circuits and its effect on the optical communication systems were clarified.
In addition, based on the result of this experiment, we prepared a guideline for the method of designing the circuits to eliminate the effect of optical fiber engineering works on the existing circuits as far as possible.

View full abstract

In the field of power system calculation, much study has been made in the area of applying parallel computers in order to increase computation speed. However, unlike examples in other application areas, such as circuit simulation or computational fluid dynamics, the performance gain is small. Also, the performance gain begins to saturate just after small number of processors are utilized. This is due to the essential feature of power system calculation: mainly (1) The size of the problem is usually much smaller than the other application fields. (2) The sparsity of the problem is high. These two features results in the fact that effect of interprocessor communication overhead is severely reflected.
In this study, the authors tried to determine the most suitable parallel architecture for Power System Calculation, restricting the number of processors to 16 or so. The architectural feature we wanted to detemine are Memory architecture and Inter processor connection method. The candidates we chose were (a) Distributed memory architecture connected by an inter-processor communication network, (b) Distributed memory architecture connected by a bus, (c) Shared memory architecture connected by a bus, and (d) Composite distributed/shared memory architecture connected by a bus. We evaluated each architecture using the data for transient stability analysis from a single PE excution, and evaluating each memory operation, necessary interprocessor communication and computation. The results showed that while it was very hard to attain much speed up using a distributed shard memory architecture connected by an interconnection network, it was possible to attain a speedup of 6.5 using 16 PEs on a Composite distributed/shared memory architecture connected by a bus.

View full abstract

A simple method was proposed to calculate the work of human ventricle that has been transmitted to arterial system so as to make clinical laboratory examination cheap and quick. Measured arterial pressure was approximated by a serial expansion of time as an input for the equivalent electrical circuit of arterial system. The transmitted ventricular work to arterial system is obtained by calculating the power at the equivalent electrical circuit of arterial system which parameters are different among the patients. This method provides the cheapest cost in the routine laboratory examination for ventricular function.

View full abstract

In this paper, we propose a new network element named a “distance-based cosine element” for neural networks. We also derive a learning algorithm based on the back-propagation algorithms for multilayer networks. The distance-based cosine element inputs a squared distance between an input pattern vector and its weight vector, and uses an affine transformation of cosine function as its output function. The proposed distance-based cosine network is able to improve its learning speed as well as convergence rate because its output function does not have any saturated regions which cause slow learning speed of the back-propagation learning using sigmoid elements.
We demonstrate the advantages of the proposed network by solving N-bits parity problems and Fisher's Iris classification problem. Experimental results indicate that our distance-based cosine network consistently obtains better results than the conventional sigmoid network in terms of both the learning speed and the convergence rate.

View full abstract

PDP (Plasma Display Panel) is a self light emitting matrix display device using luminescent phenomenon with gas discharge. It has such features as thin depth, wide view angle, free from geographical distortion, and high brightness. Gray scale is controlled by sub-field driving method using binary weighted pulse width modulation (LSB to MSB sequence). Sub-field method essentially has a flicker problem for displaying moving images. When two close gray values are displayed in particular interval (field frequency), flicker phenomenon appears.
In this paper, time-amplitude co-modulation sub-field method using RBC (Redundant Binary Code) is proposed. This driving method increases conventional two level gray scale {0, 1} to three level gray scale {0, 1/2, 1}, combining light emitting time and amplitude, and selecting driving waveform not to produce flicker. Even if two close gray values are displayed alternately in field time, light emitting frequency is no more divided into frame frequency. Proposed driving method improves flicker problem compared to conventional sub-field method.

View full abstract

This paper reports on a design method of the intermediate frequency active band-pass filter (IF BPF) using two pairs of a new current conveyor (CFCC II) and some passive CR elements. The CFCC II consists of a conventional second-generation current conveyor (CC II) and an additional current follower. The quality factor (Q) of the filter is calculated from an approximate equation, considering the frequency limit and the stability of the filter obtained from the observed characteristics of CFCC II. The second-order IF BPF is simply designed by using the approximate equation and the stable operation of the filter is experimentally confirmed under the condition of Q≤100 at 455kHz. This design method also validates in the fourth-order IF BPF which is realized by cascading two sets of the second-order TF BPFs.

View full abstract

In earlier papers, it has been reported that a fast and precise servo system, which has a low-sensitivity for parameter variation and disturbance, can be realized with simple structure by using a high order disturbance observer. However, a clear and simple design method satisfying specifications for the robust stability, the influence of the measurement noise, and the relative stability (damping characteristics) for large parameter variation has hardly been proposed. In this paper, we clarify the class of the robust servo system realized by adjusting the order of the disturbance observer and the control parameters. We apply this strategy to the design of a second delay system, such as position servo system, and realize high performance robust servo system using high order disturbance observer.

View full abstract

In this paper, we propose a system bus arbitration method suited for composing processor for packet change. This arbitration method has the following features: (1) low arbitration overhead, (2) high performance bus throughput, and (3) only a few number of arbitration signal lines requiring hardware on a small scale. Arbitration architecture provides one signal line for connecting the processors of each peripheral channel in a ring and circulates a 1-bit token among them. Also, in situations where the number of peripheral channels increases, we consider dividing the channels into multiple groups and propose an independent bus composition method designed to reduce the packet waiting times. The characteristics of the proposed method are examined by constructing and analyzing a Markovian model of the system. Finally, we verified the accuracy of analysis by comparing the results against simulation and use these results in discussing the best divisions of channels in terms of packet delay.

View full abstract

This paper describes a self-learning ULR fuzzy controller using temporal back propagation. The ULR fuzzy controller is a multi-layer feed-forward network in which each node performs an unidirectional linear response (ULR) function (node function) on incoming weighted signals. In order to achieve a desired input-output mapping, the weight parameters are updated with a temporal back propagation such that the state variables can follow a given desired trajectory as closely as possible. The temporal back propagation algorithm is used to train the ULR fuzzy controller to a variety of problems. We demonstrate the effectiveness of the self-learning ULR fuzzy controller by applying it to a benchmark problem in intelligent control-the inverted pendulum system. Experiments show a very good control performance and self-learning capability of the ULR fuzzy controllers.

View full abstract

In this paper, we propose an algorithm based on Thymus Learning System (TLS) for elucidating an emergence of complexity generation which an elongation system generates with the self-nonself recognition mechanism. TLS has the immunological tolerance as an incorporation of non-self. TLS shows various interesting features from the viewpoint of the information engineernings such as immunological memory and tolerance, pattern recognition. We present that our immune system provides a suitable methology for a complexity emergent problem of the self-defense process in evolution.

View full abstract

This paper presents an direct design scheme of model reference adaptive control (MRAC) for pneumatic servo system with additive external forces. In this design scheme, the controller contains the mechanism which can eliminate the constant disturbances caused by additive external forces. As the result, the proposed design scheme works well in the presence of the constant disturbances. The effectiveness of the proposed design scheme of MRAC has been confirmed by experiments carried out using the existent pneumatic servo system.

View full abstract

The aim of this paper is to propose adaptive control systems and a methodology for the design of them. We use multiple controllers to control a nonlinear plant whose parameters widely change without making use of any plant models. The basic ideas of adaptation by the proposed multiple controllers are both to settle the plant output to the desired reference signal while gradually narrowing down uncertainties of plant parameter values by switching and to keep the plant output stable for a series of switching. Multiple controllers consist of several basic controllers and a switching rule. We introduce two robust performance criteria into each basic controller, by which we try to guarantee stability and responsiveness of the system affected by parameter fluctuations. Switching rule utilizes only tracking error signal between the plant output and the desired reference signal, and simply switches a basic controller in a predefined sequence at a constant switching time intervals. The procedure for constructing multiple controllers is divided into two processes, the generating process and the selecting process. In the former process, a number of candidates for the basic controllers are produced. In the latter process, several basic controllers are selected from the candidates, so that basic controllers will satisfy robust performance requirements for multiple controllers. To illustrate the effectiveness of our approach, we apply the proposed system to the cart-pole balancing problem.

View full abstract

The cell operating temperature of a photovoltaic (PV) panel is an important information required to calculate its electrical power output. Two computing schemes proposed in this study to estimate the PV cell temperatures employ neural networks to learn and perform nonlinear functional mapping. Output of each computing scheme is the PV cell temperature estimated at 5 minutes intervals corresponding to a given profile of weather inputs, namely the solar radiation intensity and the ambient temperature. A feed forward neural network trained using the back propagation algorithm realizes the nonlinear relationship between weather inputs and the corresponding PV cell temperature. Results of the study show that the cell operating temperature of a PV panel can be very accurately estimated using this approach. The method can be applied in the PV systems simulators and controllers. The same technique may be used for modeling in the other areas where it is hard to obtain deterministic models but easy to obtain input-output data.

View full abstract

I proposed an fuzzy inductive learning algorithm IDF, where IDF generates rules in form of fuzzy decision tree. I verified the effect of IDF by numerical experiments. However, it is necessary for IDF to collect enough training samples. On the other hand, it is difficult to collect enough samples in real world problems. Because we don't know indispensable samples for a inductive learning algorithm and we can't collect samples that include all combination of attribute values. Thus, it is necessary to generate rules from the collected samples. The training samples are usually insufficient to learn all rules. After we generate the rules once, it is necessary to collect new samples and refine the rules. Thus, the paper proposes new approach to collect necessary samples and refine rules in IDF. The approach identifies evaluation samples with bad evaluation accuracy and collect the samples as training samples. The approach refines the rules with new samples. The paper also verifies the effect. of new approach by munerical experiments with a real world problem.

View full abstract

Although many researches have been done in the field of distributed autonomous systems, most of them discuss only the theory aspect with simulation but no experimental results. Among the systems with real experimental results, the works which take into consideration the system dynamics are still rare. In this paper, a distributed autonomous mechatronics system is proposed as a test bed for studying autonomous decentralized control. The proposed system consists multiple actuators loop-connected into a wheel shape. The system has significant characteristics of strong internal constraints and thus requires strong co-operation to control. In order to design an initial controller as the basis for future studies, Genetic Programming and Genetic Algorithms were employed to directly search for suboptimal torque generating functions. A real experimental system has been constructed and experimental results have confirmed the performance of the controllers obtained by the genetic methods.

View full abstract

This paper proposes a new method, which uses dynamics of dynamical system, for improving learning efficiency in multilayered neural networks. The shape of the error function is analyzed using Monte-Carlo simulation, and the learning efficiency is improved on the basis of the shape analysis and the adaptive learning rate.
The proposed method uses the adaptive learning rate, which is determined from the gradient of the error function. The proposed method is applied to XOR and 5-parity check problems and the effectiveness and feasibility of the proposed method are verified.

View full abstract

To eliminate the droplet formation during the laser ablation thin film growth process, we proposed a new laser ablation technique named “Eclipse Method”, which facilitated removing this droplet problem completely. Especially, YBa₂Cu₃O_y films obtained by this method had T_c_??_93K and droplet-free surfaces. To achieve the growth rate enhancement in low pressure condition and to introduce the activated species on the substrate in eclipse method, we further proposed a new composite shadow mask method, “Eclipse Angel Method”. Furthermore, we found the Ar addition effect to improve the film qualities of Bi-layered crystals.

View full abstract

The problem of robust stabilization of uncertain single-input and single-output (SISO) nonlinear systems is considered. Based on Lyapunov methods in conjunction with input/output linearization approach, a new approach is proposed to designing a class of continuous and bounded state feedback controllers using the possible bound of the uncertainty for uncertain nonlinear systems. It is shown that the state feedback controller developed in this paper can guarantee exponential stability instead of uniform ultimate boundedness for the nonlinear systems with uncertainties. Because of its continuity, our continuous controller can guarantee the existence of the solution to the dynamical system in usual sense. Moreover, we can easily select the control parameters, so that the method can be expected to have a better transient response. Finally, a numerical example is presented to demonstrate the effectiveness of the approach developed in this paper.

View full abstract

The digital terrain maps become widespread, and Geographical Information Systems (GIS) have come into the limelight. One of the key technologies needed in GIS is data fusion reasoning. The function of data fusion is to consider various geographical data, such as “road” and “slope”, and to make total evaluation. In this paper, we propose a data fusion reasoning technology using uncertain knowledge. Data fusion knowledge contains some uncertainty. For example, the knowledge for evaluating mobility costs is uncertain because it is qualitative, such as “we want to refuse the steep place.”. We introduced two uncertainty reasoning mechanisms to represent such data fusion process. One is fuzzy reasoning, and the other is Dempster-Shafer theory. We also offer knowledge editing facilities for describing data fusion knowledge, such as a data-flow diagram editor for designing data fusion process and a membership function editor for describing data abstraction methods. These knowledge editors facilitate the development and modification of data fusion knowledge base for GIS.

View full abstract

This paper presents the visual servoing using the H_∞ filter with verification of effectiveness of the robust filtering method. We show that the H_∞ filter is useful for a real system as the robust estimator against disturbances. We first describe the model of a visual servoing system by the relation between the moving target and the camera. Next, we apply the H_∞ filter to the system to estimate the velocity of the target and the relative position of the camera and the target. Finally, we make the comparative experiments between the H_∞ filter and the Kalman filter. The real-time experimental results show the effectiveness of the proposed method.

View full abstract

This paper describes a technique for recognizing human clothes by computer. It is almost our daily routine to look for a person, and we focus our attention on the fact that one's clothes have often much information on identifying a person.
In the proposed technique, clothes in an acquired image are described by a set of line segments using Hough transform, and the discrete relaxation is employed for finding match between the line segments and model clothes also represented in a set of line segments. The degree of the match is evaluated to judge whether the clothes in the input image agree with the model clothes. Experimental results show satisfactory performance of the matching by the discrete relaxation. Applications of the proposed technique may include computer identification of a particular person one wants to find based on his/her clothes recognition.

View full abstract