電気学会論文誌C(電子・情報・システム部門誌)
Online ISSN : 1348-8155
Print ISSN : 0385-4221
ISSN-L : 0385-4221
122 巻, 7 号
選択された号の論文の22件中1~22を表示しています
  • 塩島 謙次, 末光 哲也, 重川 直輝
    2002 年 122 巻 7 号 p. 1085-1088
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    We have demonstrated AlGaN/GaN HEMTs with lightly-doped buried p-layers under the channel for the first time. A 2-μ-gate device showed good pinch-off characteristics, gm of 55mS/mm, and breakdown voltage of 70_??_90V. It was confirmed that the p-n junction successfully suppressed leakage current under the channel, and the ohmic metallization did not penetrate to the buried p-layer. Carrier confinement by the p-n junction was confirmed by both one-dimensional calculations by solving the Poisson equations and capacitance-voltage measurements. As for RF performances, the maximum fr of 5.5GHz and a carrier velocity of 1×107cm/s were obtained. These results indicate the potential of p-layer insertion into GaN-based FETs for reducing a short-channel effect and possibility of impact ionization. We can conclude that p-layers can be applied to GaN FETs in the same manner that they are to III-V semiconductor electron devices.
  • Takako Yasui, Junichi Ohwaki, Masato Mino
    2002 年 122 巻 7 号 p. 1089-1093
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    A 2-W supply optical powering system that maintains electrical isolation is described. The output power of this system is stable and higher than that of conventional systems, and is not affected by environmental stresses, such as bends in the optical fiber or increased PV cell temperature, owing to its feedback control function. The system mainly consists of laser diodes, optical fibers, photovoltaic (PV) cells, and a feedback control circuit, and can deliver 2 W of electrical power. In the feedback control circuit, the PV cell output current is compared to the target current, and the laser output is controlled to equalize them. The system also has a safety function that shuts down the laser if an optical fiber is cut. The system was tested under a high-voltage condition, and the results indicate that it will be useful for driving sensors and measurement equipment used in high-voltage environments.
  • 田村 宏樹, 唐 政, 石井 雅博
    2002 年 122 巻 7 号 p. 1094-1103
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    The neuron has the characteristic of reacting to a specific stimulus. C. Koch, T. Poggio and V. Torre guessed the retinal ganglion cells of a cat has a nonlinear interaction between synapses and realizes the movement reaction selection function by receiving a fast input and a slow input. In this paper, we propose a neuron model by including the interactions between excitation and inhibition and the difference of time of inputs. The model is used to simulate the movement direction selection function. In the simulations, we initialize the interaction and dendrite randomly and use the back-propagation-like algorithm to train the neuron model to learn the movement direction selection function.
  • 寺澤 靖雄, 八木 透
    2002 年 122 巻 7 号 p. 1104-1109
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Visual prostheses are devices to recover visual sense by means of electrical stimulation to the visual nervous system. Although the ultimate goal is to restore excellent human visual function, presently it is aimed to realize the minimal visual function in order to carry out specific tasks in our daily livers. The purpose of this study is to determine specifications of the visual prosthesis which can realize the above function. In this study we have developed a simulator which enables us to experience prosthetic vision using real-time image processing. Here we selected “reading” as one of the most typical tasks in our daily lives, and quantitatively estimated reading ability with the simulated prosthetic vision. We used the MNREAD, a commonly-used reading ability test. As a result, it was observed that the range from 4 to 8 was appropriate as a gray level. In addition, it was suggested that the size of a pixel affected the reading performance, though it was less influential than the gray level.
  • 石山 仁, 笠原 宏, 天野 和彦, 上馬場 和夫, 許 鳳浩
    2002 年 122 巻 7 号 p. 1110-1119
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    An internal-use medication experiment of Ninjin-to and Mao-to was performed to six healthy adult males. The radial artery pressure pulse and stroke volume before and after taking each medicines were measured for 2 hours at every of 30 minutes. Then circuit constant of the five-element electric circuit arterial model which authors reported previously was calculated from the measured radial artery pressure pulse and stroke volume. Comparison examinations were carried out among each individual on the change of the calculated circuit constant with the passage of time. As the results, the change of the radial artery pressure pulse due to the internal-use of Ninjin-to and Mao-to has been able to be shown quantitatively as change of the central and peripheral constants.
  • 井上 悟, 秋月 影雄, 犬島 浩, 実森 彰郎, 宇佐美 照夫, 栃川 哲朗
    2002 年 122 巻 7 号 p. 1120-1126
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    In the field of oil well drilling, the underground communication system applying mud pulse method or electromagnetic wave method is put to practical use. However these communication systems are only one-way telemetry systems allowing communication from the underground to the earth surface for data transmission. In this paper the authors describe a construction of a two-way communication system and the performance experiments of it. The modulation and demodulation techniques are obtained by combining with Barker code and the differential shift keying method. By this technique, the bit rate of data transmission is decreased and the communication in the low signal to noise ratio becomes possible. According to the experimentally proved result, signal to noise ratio of the received signal could recover to-4 dB, and agreed with simulation results. Moreover, in the communication experiment with a transmission depth of 4000m, the rate of recognition of the transmitted bit information became 100%. The carrier frequency in this experiment is 1Hz. The possible limit case of transmission is estimated as the bit rate of (the receive signals in the underground) is 0.03bit/sec, and one in the surface is 0.04bit/sec.
  • 梅崎 太造, 佐藤 省三, 木村 聡仁, 松本 憲幸
    2002 年 122 巻 7 号 p. 1127-1136
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    In this paper, an effective fingerprint verification method for varying qualities of fingerprint images is described. When an optical sensor is used for imaging a fingerprint, the image quality is degraded especially due to a lack of moisture on the finger surface. Under such a condition, the captured fingerprint images are often faded. First, we investigated a method for improving the quality of faded fingerprint images, which is based on high frequency emphasis of the image using Fourier transformation. After the image is enhanced, the peaks of the group delay spectra (GDS) which are the features extracted for verification are more distinct than those extracted from the original image. When the reference patterns are selected from the fingerprint data obtained in April, the verification rate with the high frequency emphasis is 90.8%, while the rate without the emphasis is 75.3%. However, for the data obtained in February which include many faded fingerprints, the verification rate strongly decreases. Second, we examined a method to adapt differences in seasonally dependent image quality based on updating a reference pattern in terms of the accepted input pattern. Using this method, when the reference patterns are composed of the February data, the verification rate is improved from 57.1% to 85.2%. Furthermore, by resetting the verification threshold value whenever a reference pattern is updated, the rate increases to 96.5%, which is equivalent to that when the all seasonal data are registered.
  • 中澤 健亮, 古屋 穂高, 前川 仁孝, 伊與田 光宏
    2002 年 122 巻 7 号 p. 1137-1143
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Stereo matching takes a long computation time. It is expected that we can speed up stereo matching by parallel processing. One of the parallel processing methods is line partitioned which is dividing the image domain into one line. This method assigns each line to a processor when searching point. Then, the load of the matching processing is distributed almost equally. However, it has the problem of cache memory in the case of read miss for dividing large domains. Therefore, this paper proposes the speed up scheme to improve the hit ratio of cache memory by dividing the domain into small blocks. Lastly, the effectiveness of the proposed method is shown by revealing the experimental results on the shared memory multiprocessor named ‘Sun Enterprise 4500’.
  • 瀬川 修, 高橋 誠, 西俣 憲幸
    2002 年 122 巻 7 号 p. 1144-1149
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    For utilization in electric power control and information systems, we have developed a false operation prevention system using speech recognition technique. In operation of a remote line-switching task, any operator must do a pair of utterance and action, what is called, “pointing and calling check”. We have paid attention to this behavior and proposed a model of human error check system by intelligent interface agent. The function of the agent is to check operator's utterances which declare the next operation step by using speech recognition and furthermore to check the consistency of both operator's utterances and pointing actions. To realize the model architecture, we have developed a grammar-based continuous speech recognition module and implemented the whole function of the system. In experiment, we have confirmed a practicality of the proposed system by three operator's simulated operations.
  • 鈴木 良昭, 鈴木 隆, 板宮 敬悦
    2002 年 122 巻 7 号 p. 1150-1158
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    This paper presents a design method of the indirect model reference adaptive control (MRAC) system for the plant with deterministic disturbances, robust to the modelling errors due to unmodelled dynamics and observation noise. In the proposed method, the structure of controller is determined based on the internal model principle to reject the deterministic disturbance and its parameters are adjusted by using the estimates of plant parameters. Robustness to the modelling errors is ensured by the use of the adaptive law with dead zone and the fixed compensator of integral type. The adaptive law prevents the adjustable parameters from drifting and the fixed compensator improves the control performance deteriorated by the use of dead zone. Stability of the MRAC system is analyzed using the concept of _??_ norm and the Bellman-Gronwall lemma, based on the properties guaranteed by the adaptive law. Effectiveness of the proposed method is demonstrated by the simulation carried out for a plant with second order nominal part and a step disturbance.
  • 福馬 均, 田中 京子, 山浦 逸雄
    2002 年 122 巻 7 号 p. 1159-1165
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    This paper describes the electrode arrangement on the trunk for accurate measurement of earth resistance of a tree. The number of electrodes to be put into the trunk should be reduced to minimum because of damage to the tree. Potential distributions inside the trunk are calculated when a single or multi electrodes is/are used for the current injection into the trunk. According to the results of calculation, it is found out that the averaged value over the potential distribution around the base of the tree leads to the accurate measurement. Though the deviation of potential distribution around the base is not negligibly small using a single current injection electrode, the accurate value of earth resistance of the tree with the error of ± 3% can be obtained when a potential pick-up electrode is set on the position within the angle ranging from either +56 to +114 or -56 to -114 degree showing the averaged value at the circumference of the butt. Field studies on the larch, oak and locust tree showed the same result as found in the calculation. In the proposed method, two electrodes on the trunk satisfy the accurate measurement condition and minimize the damage to the tree.
  • 八幡 幸司, 名嘉村 盛和, 玉城 史朗, 翁長 健治
    2002 年 122 巻 7 号 p. 1166-1171
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    In this paper, we considered the minimum initial marking problems of Petri nets. Petri nets have been widely recognized as the best mathematical tool for modeling and analyzing concurrent systems. However, the performance analysis is restricted by the potential state space explosion. Many papers have proposed approximation algorithms for Petri nets analysis. However, a few paper reported application of genetic algorithm to Petri net analysis.
    The minimum initial marking problem is one of the important problems for Petri net analysis. We proposed a new solution based on a Genetic algorithm with heuristic rearrangement of strings to solve approximately the problem. Effectiveness of the proposed method was shown by experimental evaluation and we found that the proposed method can obtain better solution with higher probability than AMIM for many cases.
  • 高田 等, 縄田 俊則
    2002 年 122 巻 7 号 p. 1172-1178
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    In this paper a novel approach to nonlinear control, called an Augmented Automatic Choosing Control (AACC), is presented. Considering the nonlinearity, a separative function is introduced and its inverse domain associated with the region of the system is divided into some subdomains. On each subdomain, the system equation is linearized by Taylor expansion around a suitable point so that a constant term is included in it. This constant term is treated as a coefficient of a stable zero dynamics proposed here. Thus the given nonlinear system approximately makes up an augmented linear system, to which the optimal linear control theory is applied to get the linear quadratic (LQ) control. These LQ controls are smoothly united into a single nonlinear feedback controller by automatic choosing functions of sigmoid type. Parameters of these functions are suboptimally selected with the aid of a genetic algorithm (GA). This approach is applied to a field excitation control problem of power system to demonstrate the splendidness of the AACC. Simulation results show that the new controller can improve performance remarkably well.
  • 越智 敬之, 萩原 将文
    2002 年 122 巻 7 号 p. 1179-1185
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    In this paper, we propose a machine design support system using evolutionary techniques. The proposed system creates “machine” which consists of multiple parts. It is composed of two phases: the fundamental-parts-generation-phase, and the machine-generation-phase where the generated parts are combined. On each phase, we use evolutionary techniques, Genetic programming. The system generates parts themselves, uses functional parts in advance, and expresses shape's structure by tree structure. Therefore it is possible to generate various and functional designs. We have implemented the system on PC, and carried out experiments to confirm that various novel objects can be obtained.
  • スイッチングQ-学習の提案
    釜谷 博行, 阿部 健一
    2002 年 122 巻 7 号 p. 1186-1193
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    The most widely used reinforcement learning (RL) algorithms are limited to Markovian environments. To handle larger scale partially observable Markov decision processes, we propose a new on-line hierarchical RL algorithm, which is called Switching Q-learning (SQ-learning). The basic idea of SQ-learning is that non-Markovian tasks can be automatically decomposed into subtasks solvable by multiple policies, without any other information leading to good subgoals. To deal with such decomposition, SQ-learning employs ordered sequences of Q modules in which each module discovers a local control policy based on Sarsa (λ). Furthermore, a hierarchical structure learning automaton is used which finds appropriate subgoal sequences according to LR-I algorithm. The results of extensive simulations demonstrate the effectiveness of SQ-learning.
  • Ting Wang, Yasuo Sugai
    2002 年 122 巻 7 号 p. 1194-1201
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Wavelet neural networks are networks employing nonlinear wavelet basis functions as the activation functions of the neurons. This paper presents a new type of wavelet-based neural network: the local linear adaptive wavelet neural network. A hybrid evolutionary programming and gradient descent algorithm is introduced to the learning of the proposed network. The local linear models which are used in some neurofuzzy systems are introduced as powerful weights instead of straightforward weights employed in the previous wavelet neural networks. Training is performed by using the evolutionary programming algorithm at first to search a good region in the parameter space and then employing the gradient descent algorithm to find a near optimal solution in that region. The experiments on a number of nonlinear dynamic system identification problems indicates that the proposed network with the hybrid EP/Gradient algorithm can successfully identify and describe the input/output relationship for an unknown complex system with a small number of wavelet basis functions and compared favorably to the traditional neural networks with the sigmoid activation functions and the previous wavelet neural networks with straightforward weights.
  • 名嘉村 盛和, 天雨 美秋, 松村 隆, 島袋 勝彦, 長山 格, 山城 毅
    2002 年 122 巻 7 号 p. 1202-1208
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Petri nets are a mathematical tool well suited for modeling and analysis of Discrete Event Dynamic Systems (DEDS). In this paper, we propose and implement hardware Petri nets on programmable devices. The hardware Petri Nets are generated by firstly modeling the target system with a GUI tool, secondly converting the net description into VHDL codes and finally implementing onto FPGA devices. The hardware Petri nets are useful for discrete event simulation in which they can represent perfectly the natural parallelism of the target system in the circuit level. Moreover, our proposed system is regarded as logic circuit development systems in which we can design desired logic circuits by using Petri nets and implement it directly on FPGA as logic circuits.
  • 築島 隆尋, 佐藤 匡弘, 大成 尚
    2002 年 122 巻 7 号 p. 1209-1217
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    The application of parallel computing system to MRP (Material Requirements Planning) is essential to achieve a real-time demand forecasting for a whole e-Marketplace near future. The MRP using loosely connected multi-computer system is examined here. New methods of synchronization, load balancing and data access are required to keep high parallel efficiency increasing PE's(Processing Elements). In this paper 2 synchronization methods are proposed. Item synchronization method can keep 96% parallel efficiency for the MRP with 120, 000 items on the 6 PE's structure and can be robust against unbalanced load. The processing speed of the experimental system increases in liner fashion.
  • 福寄 雅洋, 原 章, 長尾 智晴
    2002 年 122 巻 7 号 p. 1218-1225
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Recently, many works about Partially Observable Markov Decision Processes have attracted increasing attentions in reinforcement learning because the world cannot be always modeled on Markov Decision Processes. In this paper, Time-dependent Classifier System; TCS is proposed to solve perceptual aliasing problems. In TCS, it keeps the constant length of time sequential information, and uses them only when it is necessary to solve the perceptual aliasing. So, the changeable lengths of rules are proposed in TCS using the time tags. And the internal action to detect the information of 1 time-step is used to link the present information to the past information. And the several maze problems with Partially Observable Markov Decision Processes are experimented to verify TCS's effectiveness.
  • 中村 政俊, 吉戒 泰輔, 後藤 聡, 松村 司郎
    2002 年 122 巻 7 号 p. 1226-1233
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Construction of reasonable controller and automatic determination of BIR for a boiler system in a thermal power plant are investigated based on nonlinear separation control method by using actual data of the thermal power plant. The nonlinear separation control consists of nonlinear statics compensation and liner dynamics compensation, independently. The effectiveness of the proposed method was assured by numerical simulation studies.
  • 川勝 真喜, 小林 宏一郎, 内川 義則, 斎藤 正男, 小谷 誠
    2002 年 122 巻 7 号 p. 1234-1235
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    We have developed a helium collection and re-liquefaction system for a MEG measurement system, and used the system applying to a 195 channel SQUID system. The helium purchase quantity became half the amount in our former report. However, this SQUID system has the large amount of consumption of helium compared with other MEG systems. So, we applied the re-liquefaction system to 122-channel SQUID system (Neuromag 122TM) which used the general amount of the helium (about 100 liters/week). The helium purchase quantity became less than 1/4, and the whole expenses of cooling also became an about 43% decrease.
  • 森 武宏, 小亀 英己
    2002 年 122 巻 7 号 p. 1236-1237
    発行日: 2002/07/01
    公開日: 2008/12/19
    ジャーナル フリー
    Shifted Hurwitz stability has been one of traditional means to specify control performances. Still, its conditions have thus far scarcely come to the fore in studies of control systems analysis, in large part because they would be regarded as direct corollaries of the “shiftless” Hurwitz stability criteria. This note examines them afresh and presents new conditions and a characterization for the shifted Hurwitz property. The results obtained possess some advantages over those derived simply by shifting the Hurwitz stability conditions.
feedback
Top