In this paper, we show that the nondestructive detection of chemicals hidden in envelopes can be achieved using terahertz waves in a simple two-step procedure: First, scattering of the terahertz waves is an indicator of the presence of powders in the envelope; second, the identification of the chemicals is done by spectral fingerprinting. In addition, we examined the possibility of measurement by diffuse reflection.

抄録全体を表示

Blind deconvolution is a method of recovering transmitted signals from only received signals. The probability distribution method is one of blind deconvolution methods. This method has two problems that has slower convergence, and that its reliability is lower. In this paper, we propose the new algorithm for solving two problems above. The proposed algorithm is as follows. 1)it is based on the adaptive processing with each sample. 2) Kurtosis is adaptively estimated by recovered signals with each sample. 3) Cost function is decided by kurtosis. 4) Transmitted signals are recovered by received signals using decided cost function on the sample time. We confirm validity of the new algorithm by computer simulation.

抄録全体を表示

Recently, domestic accidents caused by sudden illness are increasing in Japan. We developed a detection and alarming system for the dangerous state in a bathroom. FG sensor which has been developed in our laboratory is used in this system to get respiratory information of bathing person. It becomes possible to watch the person in a bathroom by this sensor, without invading his privacy. We applied the developed system to an actual bathroom, and carried out several experiments supposing various situations. As a result of the experiment, we confirmed that this system can acquire the person's respiratory information and can emit alarms correctly for emergency.

抄録全体を表示

An assistance of safe driving is one of the most remarkable field in ITS. In this paper, a system for detecting dangerous vehicles from rear scene using image processing is proposed. The system is very useful for avoiding collisions which occur when changing lanes, or warning following drivers of the sudden collision possibility. The system is comprised of three image processing parts; the Lane Detection, the Vehicle Detection, the Vehicle Observation. Especially, in the Vehicle Detection, a new method for detecting only front faces of following vehicles without their sides is proposed. It is achieved because boundary lines between the vehicle front and the side are detected stably by Area Over-Division and Area Integrate. The detection of boundary lines is treated as the detection of dividing lines which divide into two areas. In the system, the problem of detecting vehicles is divided into many simple steps, and in comparison to conventional researches that try to search correct things, incorrect things are searched and only those are removed on each step. Therefore, only correct front faces remain after many steps. This system has already been experimented as a real time system in-vehicle.

抄録全体を表示

In recent years, the person recognition system using the face picture can consider application to security system. This paper describes a method to have face recognition with the important pixels. These pixels are found out with the genetic algorithm. In the face recognition process, only important pixels to perform the reliable recognizing are extracted. Before the recognizing process, subject's head is adjusted by free locations, inclinations and scales.

抄録全体を表示

We propose use of a constant envelope pulse for the purpose of improving noise-rejection property in chaotic synchronization based communication systems. In a conventional system where discrete-time chaotic signals are transmitted through the pulse amplitude modulation, the correlator output variance increases as the spreading factor decreases, while the synchronization error increases as the spreading factor increases. Therefore, it is difficult to control the bit error rate only by adjusting the spreading factor. In the proposed system, use of pulse width modulation keeps the envelope of transmitted signals constant, which leads to the correlator output with zero variance. The synchronization error is kept small because the spreading factor can be set to be one without increasing the correlator output variance. We have a result of computer simulation showing that the proposed system achieves a bit error performance better than the conventional system. In addition, a new blind adaptive algorithm is proposed which suppresses inter-symbol interference.

抄録全体を表示

In process industries such as the chemical plants, a good control performance cannot be obtained by simply using the linear controllers, since most processes are nonlinear multivariable systems with mutual interactions. And now, in various fields, the neural networks are well known as the representative schemes to describe the nonlinear elements included in the systems. Also, many types of neural-net based control systems have been proposed, since they have the ability of function approximation, the training ability and versatility. However, the neural networks tend to require great deal of training iteration or careful adjustments of user-specified parameters. In this paper, a design method of neural-net based decouplers is proposed for nonlinear multivariable systems. Here, the decoupler is generated by the sum of a static decoupler and a neural-net based decoupler. The former is used so that the influence of mutual interactions is roughly removed, and the latter plays a role of compensating the nonlinearities and decoupling the remaining mutual interactions. Thus, by designing the control system as the hybrid system, the burden in training the neural networks can be considerably reduced. Finally, the effectiveness of the proposed control scheme is evaluated on a simulation example.

抄録全体を表示

It is well known that sliding mode control (SMC) is capable of tackling systems with uncertainties. However, the discontinuous control signal causes a significant problem of chattering. In this paper, a new and simple approach to chattering free SMC methodology is proposed. The main purpose is to eliminate the chattering phenomenon. As a result, the chattering is eliminated and error performance of sliding mode control is improved. The reduction of the chattering of sliding mode control is achieved by using a distance function which measure the distance between the trajectory of state errors and the sliding surface as the corrective control term instead of discontinuous sign function. Experimental study carried out on a magnetic levitation system is presented. Experiments verified that the proposed control has the advantage of less chattering in SMC.

抄録全体を表示

Owing to the rapid progress of the computer technology in recent years, lots of works about intelligent control schemes have been developed in the field of control engineering. However, PID control algorithms still continue to be widely used for most industrial control systems, particularly in chemical process industries. However, it is difficult to find a suitable set of PID gains, because most processes are represented by nonlinear systems. On the other hand, in recent years, various neural networks (NNs) have been proposed. These technologies of the NNs enable us to deal with the nonlinear systems, and they play an important role in the field of control engineering. Furthermore, a Cerebellar Model Articulation Controller (CMAC) has been proposed as one of artificial neural networks. This paper presents a new design scheme of intelligent PID controllers whose PID gains are generated by using CMACs. Owing to the high-generalizing ability of the CMAC, the proposed scheme does not need large learning time with comparison to the conventional NN. Furthermore, the computational burden can be drastically reduced and the robust tracking can be realized for untrained reference signals. The newly proposed control scheme is numerically evaluated on some simulation examples.

抄録全体を表示

In this paper, we propose a new catchcopy creation support system by using interactive genetic programming and EDR electronic dictionary. Formerly we proposed a catchcopy creation support system called “CMSS" (catchcopy making support system) as a way-of-thinking support tool. However, it was not practical enough because it couldn't reflect user's evaluation for the outputs. In addition, many unnatural outputs could be generated. In order to solve these problems and to obtain better catchcopies, we propose a new catchcopy creation support system called “CMSS2". We employ interactive genetic programming on the proposed system. It can reflect user's evaluation for the outputs. Furthermore, in addition to more than 100 refined grammatical rules, we propose several new fitness functions considering following points to obtain practical catchcopies: structure of a catchcopy, frequency and distance among words on dictionary, and the rhyme. The validity of the proposed CMSS2 is verified by experiments.

抄録全体を表示

A practical optimal reliability design of a system required high system reliability could be formulated as an appropriate mathematical programming model, however, because in the real world, we should concern some kinds of decision criteria. Particularly, system reliability and construction cost are basically conflict each other, so that when taking both of them into consideration, the system reliability design model can be formulated as a bi-objective mathematical programming model. In this research, we consider a bi-criteria redundant system reliability design problem which is optimized by selecting and assigning system components among different valuable candidates for constructing a series-parallel redundant system. Such a problem is formulated as a bi-criteria nonlinear integer programming (bi-nIP) model. In the past decade, several researchers have developed many heuristic algorithms including genetic algorithms (GAs) for solving multi-criteria system reliability optimization problems and obtained acceptable and satisfactory results. Unfortunately, the Pareto solutions obtained by solving a multi-objective optimization problem using a GA cannot be guaranteed its quality, and the number of the Pareto solutions obtained is sometimes not so many. In order to overcome such problems, we propose a hybrid genetic algorithm combined with a Fuzzy Logic Controller (FLC) and a local search technique to obtain the Pareto solutions as many and good as possible. The efficiency of the proposed method is demonstrated through comparative numerical experiments.

抄録全体を表示

Various stock prices predicting and sell-buy strategy models have been so far proposed. They are classified as the fundamental analysis using the achievements of the companies and the trend of business, etc., and the technical analysis which carries out the numerical analysis of the movement of stock prices. On the other hand, as one of the methods for data mining which finds out the regularity from a vast quantity of stock price data, Genetic Algorithm (GA) has been so far applied widely. As a concrete example, the optimal values of parameters of stock indices like various moving averages and rates of deviation, etc. is computed by GA, and there have been developed various methods for predicting stock prices and determinig sell-buy strategy based on it. However, it is hard to determine which is the most effective index by the conventional GA. Moreover, the most effective one depends on the brands. So in this paper, a stock price prediction and sell-buy strategy model which searches for the optimal combination of various indices in the technical analysis has been proposed using Genetic Network programming and its effectiveness is confirmed by simulations.

抄録全体を表示

Genetic Network Programming(GNP) is extension of Genetic Algorithm(GA). The GNP is suitable for an agent programming. The GNP can evolve anoperation program of the agent. But, the GNP has the premature convergence problem as an evolution technique as well as GA. On the other hand, to avoid the initial convergence of the GA, Immune Alogorithm(IA) which is introduced the immune suppression to the GA had developed.
Then if the GNP and IA are combined, a better algorithm for agent programming can be developed. Authors proposed Immune evolved Genetic Network Programming (IGNP). To compare the GNP and IGNP, the simmulation was done. As a result, IGNP is more exellent without the premature convergence than the GNP.
In this paper, the authors explain the IGNP outline and a effectiveness of the IGNP is stated through the simulation result.

抄録全体を表示

In highly dynamic electronic commerce environment, the need for adaptability and rapid response time to information service systems has become increasingly important. In order to cope with the continuously changing conditions of service provision and utilization, Faded Information Field (FIF) has been proposed. FIF is a distributed information service system architecture, sustained by push/pull mobile agents to bring high-assurance of services through a recursive demand-oriented provision of the most popular information closer to the users to make a tradeoff between the cost of information service allocation and access. In this paper, based on the analysis of the relationship that exists among the users distribution, information provision and access time, we propose the technology for FIF design to resolve the competing requirements of users and providers to improve users' access time. In addition, to achieve dynamic load balancing with changing users preference, the autonomous information reallocation technology is proposed. We proved the effectiveness of the proposed technology through the simulation and comparison with the conventional system.

抄録全体を表示

The e-Learning has been focused on not only the employee training at enterprises but also the lesson at universities in Japan. Since there are many methods of instruction such as educational software, electronic bulletin boards, and slides using computers and networks, the class content cannot be developed easily in e-Learning. Therefore, the designing class content for e-learning is more important than conventional lessons.
This paper discusses class content of a production system design for e-Learning. First, the study target of this lesson is determined and the story of the whole lesson is decided. Second, the structure of the learning theme is determined. Third, the educational materials including the educational software are developed and related to the content and structure. Finally, the demonstration experiments of the lesson are carried out at universities, the results of the questionnaires and tests are analyzed, and the effectiveness of the class including the educational method and the educational materials is discussed.

抄録全体を表示

WWW has developed rapidly, and it is becoming easy to make personal web sites. In general, we create and edit web pages by using a HTML authoring software or writing HTML source codes in a text editor. Then, we need to upload them to a web server. When we make and build our own web pages by the existing tools, it takes a lot of time and effort to complete the necessary tasks. In this paper, we propose a home page authoring support system in which we directly edit and make web pages on a web browser. Current experimental results demonstrate that our system can effectively support novices to create their web pages. Also, we show two real-world applications that effectively utilize our system.

抄録全体を表示

In this paper, we describe a new model of immune network based on biological immune response network. We propose an immunity like multiple-valued network with apoptosis mechanism. The model is based on the interaction between B cells and T cells and the biological apoptosis mechanism in human body. With the mechanism, a naturally immune system can be reproduced. The model is also applied to pattern recognition. It gets possible with a conventional model to restricting categories increase of memory patterns.

抄録全体を表示

The approach of spatio-temporal correlaton (STC) in which the object's flow is supposed to be point-symmetry for the searching point(pixel) has been proposed to improve the flow estimation of moving object. However, in the STC the shape of object becomes complex as Tree, the flows causes many errors. In this paper, we propose a plane-based spatio-temporal correlaton (PSTC) to cope with this. We show that the PSTC effectively estimates the flows of blurred “Tree" with Gaussian function through the experiments.

抄録全体を表示

This paper proposes a human body posture estimation method using neural network. The input feature vector of the neural network is composed with the result of analyzing a human silhouette extracted from camera image and the output vector of the neural network indicates the 2D coordinates of the human body's significant points. The proposed method is implemented on a personal computer and runs in real-time. Experimental results show the feasibility of the proposed method for estimating human body postures.

抄録全体を表示

In this paper we practically compared the performance of three blocked based parallel multiplication algorithms with simple fixed size runtime task scheduling strategy on homogeneous cluster of workstations. Parallel Virtual Machine (PVM) was used for this study.

抄録全体を表示

Wrong:ulrasonic image
Right:ultrasonic image

抄録全体を表示

Wrong:Analysis of Atmospheric Electric Field Variation Signals using Aspectum of the Digital Natural Observation Method
Right:Analysis of Atmospheric Electric Field Variation Signals Using Aspectum of the Digital Natural Observation Method

抄録全体を表示