A practical design procedure of coupled-resonator band elimination filter (BEF) is studied. A small-sized BEF is fabricated with Low Temperature Co-fired Ceramics (LTCC) having high dielectric constant. Deterioration of band elimination characteristic due to approach of the constituent resonators is recovered with the proposed design procedure. But, in a practical fabrication process, additional physical restrictions are placed on this structure, e.g., the length and the impedance of inter-stage transmission line make another problem for its realization. They are overcome by adjustment based on computer simulation, and a fabricated LTCC BEF shows a good performance in an experiment carried out at the 2GHz-band.
As the deregulation of electric business proceeds, each company needs to construct a risk hedging system. So far many companies have not been taking much care of this suffciently. In this paper, we address the nodal price hedge issue. Most companies have risks for the nodal prices which tend to be highly volatile. There's almost no doubt that such a company actually needs hedge products to make profits stable. We suggest the usage of FTR for this purpose. First, we briefly note the mechanisms of nodal price in PJM market and FTR, and suggest the mathematical formulations. Then we show some numerical examples and discuss our findings.
This paper presents an efficient G-XML data management method for mobile devices. G-XML is XML based encoding for the transport of geographic information. Mobile devices, such as PDA and mobile-phone, performance trail desktop machines, so some techniques are needed for processing G-XML data on mobile devices. In this method, XML-format spatial index file is used to improve an initial display time of G-XML data. This index file contains XML pointer of each feature in G-XML data and classifies these features by multi-dimensional data structures. From the experimental result, we can prove this method speed up about 3-7 times an initial display time of G-XML data on mobile devices.
Chaos control is stabilizing the state of a chaotic system into the peculiar unstable periodic orbit (UPO). In chaos control proposed until today, the target system is known in many cases. However, when the generating mechanism of chaos is unknown, the control only based on time series data observed from the system is also required. Delayed feedback control (DFC) applies control input based on the difference between the τ-time delayed state and the current state. Where τ denotes the periods of UPO. Although this method needs correctly no chaos orbit if the periods of UPO is known, there exists the limitation of the odd number property. As the method to compensate this fault, there is prediction-based feedback control (PFC) using the prediction value of τ-time future state. However, PFC needs to calculate this prediction value analytically by using the known mathematical model of the target system. Then, in this paper, chaos control for unknown chaotic systems is proposed. This technique has the hybrid type control input to improve faults of DFC and PFC. The prediction value to be used in the control input is determined by using neural network or fuzzy neural network. The control inputs are impressed only near periodic points of the target UPO using the concept of the unstable periodic region.
In general, a material discrimination task is achieved by following two steps: One is an analytical step for estimating constructive elements of the object, and the other is a computational step for identifying its material. The first step is simple because only measuring plural parameters is required to reach the correct answer. But the second step is complex and various kinds of techniques have been proposed. Then, one of the neural network models called a self-organizing map (SOM), a good tool for topology-preserving projection, is adopted in this paper. Firstly, each material is examined with a micro multi-functional tactile sensor. Secondly, in the training stage, the average data of several measurement trials are applied to the SOM. Finally, in the test stage, the actual measurement raw data, which are completely new to the trained SOM, are applied and the measurement object is estimated from the location of the emerging “winner” neuron. As a result of computer simulations, it is found xperimentally that measured information space is divided into several regions representing each material successfully. Furthermore, an acquired feature map for material discrimination is useful against several untraining data.
In multivariable control systems, controlled variables interact each other in general. In this case one compensation for a control loop exerts harmful influence upon other control loop through the interaction. If we could take away this interaction by some ways for a complete one to one corresponds between the manipulated variable and the controlled variable, then the design of control systems results in the individual compensation of the single loop control. A cross controller was used for removing the interaction thus far mainly, but it was complicated and the controller was high-order. On the other hand, the precompensator by the Inverse Nyquist Array method was also used for noninteraction of the controlled variables. However it is difficult to remove the interaction of the controlled system, which has very strong interaction, when the precompensator is used only one. This paper proposes the series and/or parallel connection of the precompensators for a multi-stage noninteracting PID control, and the multi-stage precompensators are applied to the control of the fluid temperature and liquid level interacting plant to confirm the effectiveness.
In this paper, a new notion of reachable set for linear time-delay systems is introduced and a method to evaluate the reachable set is proposed. The evaluation method is given in linear matrix inequality form. An important application of reachable set analisys is constrained controller synthesis. We also propose synthesis method of constrained H∞ controllers for time-delay systems.
Genetic Network Programming (GNP) is a kind of volutionary methods, which evolves arbitrary directed graph programs. Previously, the program size of GNP was fixed. In the paper, a new method is proposed, where the program size is adaptively changed depending on the frequency of the use of nodes. To control and to decide a program size are important and difficult problems in Evolutionary Computation, especially, a well-known crossover operator tends to cause bloat. We introduce two additional operators, add operator and delete operator, that can change the number of each kind of nodes based on whether a node function is important in the environment or not. Simulation results shows that the proposed method brings about extremely better results compared with ordinary fixed size GNP.
Recently, systems are becoming more complex and larger than ever, so numerous attempts have been made to introduce biological features into artificial systems, because many biological systems in the nature exist as one of the most complex systems. Multi agent system with symbiotic learning and evolution have been recently proposed. It is named Masbiole. In this paper, Masbiole is reviewed and the method for evolving multi agent systems is proposed. From simulations on a multi objective knapsack problem, it has been clarified that Masbiole has better performance than that of conventional multi objective genetic algorithms.
Up to this time, the vessel-mounted pinger system is used in the passive ranging system. But when we need to estimate the running ability of the underwater vehicle easily and economically, we cannot use the vessel-mounted pinger system to be expensive. So that we must track down the running locus of the underwater vehicle from the radiated noise of the vehicle. In this paper, I proposed the new passive ranging system which located the moving target. Then I computed of the time-delay difference on the output of a cross-correlation function between two radiated noise signals which were received by the acoustic sensors. And I solved three-dimensional equations for the acoustic velocity and these sensors' co-ordinates positions by the nonlinear least squares method. Still more, I analysed the radiated noise signal of the target fishing boat which were collected in MITOHAMA test sea zone. In view of these facts, the experimental result established the facts that this study's new passive ranging system can track down the running locus of the target vessel. Because I compared this study's system and the radio signal serveying system about the tracking result of the running locus of the target vessel, as summarize the results, I can explain the measured values by two different method almost agree for the running locus of the target vessel.
A learning method of the Hopfield neural network is presented for efficiently solving combinatorial optimization problems. The learning method adjusts the balance between the constraint term and the cost term of the energy function so as to keep the Hopfield network updating in a gradient descent direction of energy.This paper describes and analyzes the learning method and shows its application to the traveling salesman problem (TSP). The performance is evaluated through simulating 100 randomly generated instances of the 10-city traveling salesman problem and some TSPLIB benchmark problems. The simulation results show that the performance of the proposed learning method on these test problems is very satisfactory in terms of both solution quality and running time. The proposed learning method finds the optimal solutions on the test TSPLIB benchmark problems in very short computation time.
Our final goal is to establish the model for saccadic eye movement that connects the saccade and the electroencephalogram(EEG). As the first step toward this goal, we recorded and analyzed the saccade-related EEG. In the study recorded in this paper, we tried detecting a certain EEG that is peculiar to the eye movement. In these experiments, each subject was instructed to point their eyes toward visual targets (LEDs) or the direction of the sound sources (buzzers). In the control cases, the EEG was recorded in the case of no eye movemens. As results, in the visual experiments, we found that the potential of EEG changed sharply on the occipital lobe just before eye movement. Furthermore, in the case of the auditory experiments, similar results were observed. In the case of the visual experiments and auditory experiments without eye movement, we could not observed the EEG changed sharply. Moreover, when the subject moved his/her eyes toward a right-side target, a change in EEG potential was found on the right occipital lobe. On the contrary, when the subject moved his/her eyes toward a left-side target, a sharp change in EEG potential was found on the left occipital lobe.
In distributed appplication system, Totally Ordered Group Communication Service(TO-GCS) is very important, and it is a powerful infrastructure for building distributed fault-tolerant application such as distributed shared memory, Computer Supported Cooperative Work and distributed monitoring. Much work has been dedicated to Totally ordered multicast protocol. One of them, Adaptive Totally Ordered Multicast Communication(ATOP), is able to dynamically alter the message delivery order in response to changes in the transmission rates of group members. Based on this, a protocol ATOP/MBS is proposed and discussed. In this paper, we improve the ATOP/MBS and propose a protocol ATOP/improved MBS, which use no dummy message and can adapt well in case of transmission frequency changing in comparatively short period. According to the simulation results, TO-GCS with ATOP/improved MBS can be carried out at low delivery latency and fluctuations of delivery latency.
We propose a new image enlargement method employing a codebook and a fuzzy similarity measure in order to realize both of smoothing and sharpening in various types of image enlargement. In the proposed method, codes (vectors) in codebooks, which are generated by Self-Organizing Maps, are regarded as fuzzy if-then rules, and unknown pixel values in the image are estimated by a fuzzy inference. In this inference, fuzzy similarities between the input and the antecedent parts of the rules are calculated, then the estimation is achieved by taking a center of gravity of the consequents. In this paper, we apply this technique, with a codebook of Lena, to various standard digital images except Lena. The experimental results show that the proposed method is superior to other image enlargement techniques.
The guaranteed cost control problem of the decentralized robust control for largr-scale systems with the norm-bounded time-varying parameter uncertainties and a given quadratic cost function is considered. Sufficient conditions for the existence of guaranteed cost controllers are given in terms of linear matrix inequality (LMI). It is shown that the decentralized local state feedback controllers can be obtained by solving the LMI. The problem of guaranteed cost control for large-scale systems under the gain perturbations is also considered.
A new genetic algorithm is proposed for solving job-shop scheduling problems where the total number of search points is limited. The objective of the problem is to minimize the makespan. The solution is represented by an operation sequence, i.e., a permutation of operations. The proposed algorithm is based on the framework of the parameter-free genetic algorithm. It encodes a permutation using random keys into a chromosome. A schedule is derived from a permutation using a hybrid scheduling (HS), and the parameter of HS is also encoded in a chromosome. Experiments using benchmark problems show that the proposed algorithm outperforms the previously proposed algorithms, genetic algorithm by Shi et al. and the improved local search by Nakano et al., for large-scale problems under the constraint of limited number of search points.
In this paper, we propose a genetic algorithm including genetic operator based on empirical knowledge. We demonstrate the advantages of the proposed method by applying it to a problem of navigation of robot. In the proposed method, when a robot can’t complete a task, the weight, which is corresponding to gene, of the activation network which decides the behavior of the robot is revised directly by using the empirical knowledge. Heuristic rules used in the genetic operator based on empirical knowledge are determined to constructs a robot which has better ability than before modified by designer. Computational experiments show that the proposed method raises fitness in earlier generation than the conventional GA.
A human perceives a set of feature points (FPs) as a cluster when he/she finds a mass of collected FPs in a sample space. We have been trying to develop a clustering technique working like clustering in human perception. For such a clustering technique, we introduce the concept of perceptual position in this paper. The perceptual position is based on the assumption that human perception of relative positions among FPs changes depending on the arrangement of FPs around those FPs. To implement a perceptual position, each FP is encoded by a fuzzy set. We then describe a clustering technique using perceptual positions. Computational experiments were carried out to determine the effectiveness of the clustering technique using perceptual position, and the results showed that clustering by the technique using perceptual position is more compatible with clustering by human subjects than is clustering using a conventional fuzzy c-means (FCM) algorithms.
This paper describes automatic adjustment of a michelson interferometer using genetic algorithms. The michelson interferometer consists of optical components (such as mirrors, lens, and prisms) that must be physically positioned with micron-meter precision to obtain optimal performance. Therefore, it is very difficult to use an interferometer outside for environmental measurement such as air pollution, because outside use causes mis-alignment of optical components. In order to overcome this problem, we propose automatic adjustment method using genetic algorithms that realize the optimal and quick alignment of optical components of interferometer. We also develop new compact mirror holder that allows portable and on-site use of interferometer. We confirmed the advantage of this system by the comparison with conventional adjustment algorithms. The proposed interferometer including the new compact mirror holders has been successfully adjusted by genetic algorithm in three minutes. The quick adjustment time indicates the possibility that the system can be used for on-site measurement.
In this paper, three-dimensional image of the face is acquired by the rangefinder, and the range image is generated from three-dimensional image. The estimation method of a face direction along horizontal and vertical axis from nose position is proposed. If the direction of the face can estimated automatically, it is possible to use as a human interface. Simulation results for three-dimensional image using two acquisition methods show that the validity of the proposed method is verified.
The potentiostat system consisting of a constant voltage driving negative feedback circuit is very useful in understanding of electrochemical interfacial phenomena. This paper proposes a designing method for a simplified potentiostat circuit for biomedical instrumentations and bio-sensors, by using conventional op-amplifiers and instrumentation amplifiers. The ranges are 10μV∼10mV, 0.1nA∼10mA in the frequency of 10Hz∼10kHz. Results of the loop frequency characteristics, the intrinsic noises and the Lissajous characteristic of a designed potentiostat consisting of low noise and low bias-current op-amps. and instrumentation amps. are shown.
This paper investigates which feature of the face is focused on by human and/or computer in the classifi-cation of attributes, such as gender and human race, from facial image. The classification experiments withhuman subjects and/or artificial neural networks are carried out using the center of the face as the stimulus and/or input image. The experimental results show that human and the artificial neural network can classify the attributes from low resolution facial images and the feature parts are good agreement between human and the artificial neural network.
In the study by Yasuda et al., a basic scheme of subjective contour extraction was proposed. It succeeded in extracting subjective contours from relatively simple figures. However, the scheme when applied to the figure of a subjective zebra against a complex striped background resulted in generation of extraneous contours and in addition failed to generate certain subjective contours. Therefore the basic scheme has been augmented by three new rules (two rules to delete extraneous subjective contours and a rule of multiple stages extraction). With the augmented scheme, the subjective contours of the zebra were extracted with reasonable results.
We observed broadband spectra in VHF band at Tateyama, Japan, to investigate anomaly in long distance propagation of TV broadcasting waves. We found that TV broadcasting waves from Malaysia at 48.25MHz (distance is 5200km) and China at 49.75MHz (distance is 1800km) can sometimes propagate. Seasonal and daily variations of received broadcasting waves were investigated through one-year observations. Malaysia broadcasting waves were frequently received in spring and autumn, but China broadcasting waves were received mainly in summer. We may conclude that radio waves of Malaysia TV are reflected by F2 layer and those of China TV are reflected by sporadic E layer in summer in the ionosphere.