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
130 巻, 7 号
選択された号の論文の27件中1~27を表示しています
特集:平成21年電気学会電子・情報・システム部門大会
特集論文
<光工学>
  • 福地 哲生, 緒方 隆志
    2010 年 130 巻 7 号 p. 1095-1101
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Laser ultrasound is an effective method for noncontact flaw detection, material property measurement, and temperature measurement. In this research, laser ultrasound was applied to surface flaw detection of rotating specimens, for which conventional techniques such as ultrasonic inspection using transducers cannot be applied. A SUS316 specimen with artificial slits of depth 1 mm machined in the radial and tangential directions was used. By synchronizing the rotation frequency of the specimen with the laser repetition rate, the presence of the slits could be detected from the waveform and frequency spectrum of the surface acoustic waves. The results obtained when the specimen was under rotation was in agreement with the results obtained when the specimen was at rest. Furthermore, the detection position was scanned in the tangential direction by slight desynchronization of the specimen rotation and laser irradiation. The presence of slits could be detected with high accuracy by using the ratio of different frequency components of the obtained waveform.
<電気回路・電子回路>
  • 齊藤 健, 畑野 裕一, 斎藤 稔, 関根 好文
    2010 年 130 巻 7 号 p. 1102-1107
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Oscillatory patterns of electrical activity are a ubiquitous feature in nervous systems. Oscillatory patterns play an important role in the processing of sensory information pattern recognition. For example, earlier reports describe that the oscillatory patterns in the olfactory center of the land slug are changed by odor stimuli to the tentacles. Olfactory processing has also been studied in relation to rabbits and land slugs through the construction and use of mathematical neural network models. However, a large-scale model is necessary for the study of a model which has sensory information recognition by the oscillatory pattern. Therefore, the construction of a hardware model that can generate oscillatory patterns is desired because nonlinear operations can be processed at higher speeds than the mathematical model. We are studying about the neural network using hardware neuron models to construct the olfactory center model of the living organisms. In the present study, we discuss about the oscillatory pattern generation of the olfactory center using pulse-type hardware chaotic neuron models. Our model shows periodic, quasi-periodic and chaotic oscillations such as the olfactory center of living organisms by changing the synaptic connection weights.
  • 森村 俊
    2010 年 130 巻 7 号 p. 1108-1117
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Because of increase of communication bandwidth demand, upgrading the speed of optical fiber networks is a future issue. However, in the case of high speed transmission at wavelength of 1.5 μm band on the existing single mode fiber, transmission distance is limited due to chromatic dispersion. Thus, in addition to chromatic dispersion compensation devices, automatic optimization methods of the compensation value are important especially in the case of photonic networks, on which transmission distances often change dynamically. But problems remain such that signal quality measuring instruments are expensive under existing automatic chromatic dispersion compensation methods.
    In this report, focusing on spectrum profile change by long transmission of optical signals, a simple automatic control method of chromatic dispersion compensation is proposed. And some results of basic experiments are described. By doing feedback control for optimizing the chromatic dispersion compensation value, an error-free state of a 10 Gbps optical signal is demonstrated even if the dispersion value of the transmission line is rapidly changed.
  • 高遠 健司, 鈴木 大, 石井 威, 小林 正人, 山田 宏一, 天野 茂
    2010 年 130 巻 7 号 p. 1118-1124
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Many AM&FM broadcasting stations in Japan are connected by the leased circuit system of NTT. Small delay and high performance AD/DA converter was developed for the system. The system was designed based on ITU-T J.41 Recommendation (384kbps), the transmission signal is 11bit-32 kHz where the Gain-frequency characteristics between 40Hz to 15kHz have to be quite flat. The ΔΣAD/DA converter LSIs for audio application in the market today realize very high performance. However the performance is not enough for the leased circuit system. We found that it is not possible to meet the delay and Gain-frequency requirements only by using ΔΣAD/DA converter LSI in normal operation, because 15kHz the highest frequency and 16kHz Nyquist frequency are too close, therefore there are aliasing around Nyquist frequency.
    In this paper, we designed AD/DA architecture having small delay (1msec) and sharp cut off LPF (100dB attenuation at 16kHz, and 1500dB/Oct from 15kHz to 16kHz) by operating ΔΣAD/DA converter LSIs over-sampling rate such as 128kHz and by adding custom LPF designed Infinite Impulse Response (IIR) filter. The IIR filter is a 16th order elliptic type and it is consist of eight biquad filters in series. We described how to evaluate the stability of IIR filter theoretically by calculating frequency response, Pole and Zero Layout and impulse response of each biquad filter, and experimentally by adding overflow detection circuit on each filters and input overlord signal.
<情報通信工学>
  • 石川 了, 佐野 実, 望月 寛, 中村 英夫
    2010 年 130 巻 7 号 p. 1125-1131
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Railway signaling systems that transmit control information via rails have been deployed in many applications, for example, digital automatic train control (ATC) systems for controlling train speed. Since the performance of digital ATC systems depends on the signal transmission speed, recently there have been many studies aimed at realizing high-speed data transmission. However, it is difficult to increase the transmission speed because rails have strong attenuation in proportion to an increase of the frequency. In this paper, we aimed to increase the transmission speed by improving the modulation scheme to overcome these limitations. We proposed CDMA-QAM method that is combined code-division multiple access (CDMA) and quadrature amplitude modulation (QAM). And we evaluated in a field trial the CDMA-QAM rail transmission device developed using DSP. On the other hand, an analog ATC based on an amplitude modulation (AM) is still employed in some railway lines. It is difficult for their lines to introduce the digital signal due to track circuit configurations and interoperability conditions. So we studied a data transmission scheme that makes it possible to mix an analog signal and a digital signal, and evaluated the influence given to both signals using a developed device that generates the mixed signal of analog and digital signal.
<生体医工学・福祉工学>
  • 櫛田 大輔, 金澤 智宏, 北村 章
    2010 年 130 巻 7 号 p. 1132-1138
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    In rehabilitation, there are two roles of reinforcement of the expansion of the motion range and muscular power. The diseased part is requested to be operated by the external force in the former, and the load corresponding to patient's muscular power is requested to be given in the latter. Recently, there are a lot of researches that try such rehabilitation by the machine. The principal object is put on the motion control of the machine in those researches. The most important thing is a mechanism that patient's state is quantitatively evaluated. This paper proposes the mechanism that presumes the patient recovery by relating bodily features to EMG of the diseased part in rehabilitation. In addition, a new rehabilitation system, that contained the self adjustment of the load using those mechanisms and the consideration of fatigue, is proposed. Effectiveness of the proposed rehabilitation system is verified by the simulation work.
  • 武内 彬正, 谷 博雅, 森 雅英, 森口 裕之, 小谷 潔, 李 鍾国, 野城 真理, 神保 泰彦
    2010 年 130 巻 7 号 p. 1139-1144
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    The sympathetic nervous system is one of the principal sources for regulating cardiovascular functions. Little is known, however, about the network-level interactions between sympathetic neurons and cardiomyocytes. In this study, a semi-separated co-culture system of superior cervical ganglion (SCG) neurons and ventricular myocytes (VMs) was developed by using a polydimethylsyloxane (PDMS) chamber placed on a microelectrode-array (MEA) substrate. Neurites of SCG neurons passed through a conduit of the chamber and reached VMs. Evoked activities of SCG neurons were observed from several electrodes immediately after applying constant-voltage stimulation (1 V, 1 ms, biphasic square pulses) to SCG neurons by using 32 electrodes. Furthermore, this stimulation was applied to SCG neurons at the frequency of 1, 5 and 10 Hz. After applying these three kinds of stimulations, mean minute contraction rate of VMs increased with an increase in the frequency of stimulation. These results suggest that changes in contraction rate of VMs after applying electrical stimulations to SCG neurons depend on frequencies of these stimulations and that the heart-regulating mechanisms as well as that in the body were formed in this co-culture system.
<システム・計測・制御>
  • 朝日 一平, 二宮 英樹, 杉本 幸代
    2010 年 130 巻 7 号 p. 1145-1150
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    An eye-safe Raman lidar for remote measurement of hydrogen gas concentration was constructed. The lidar used a low power diode-pumped solid state laser of wavelength 349nm and pulse energy 120μJ. Detection was possible for H2 gas concentration in the gas cell of more than 4% at a distance of 5m from the lidar. The concentration of hydrogen gas leaked into the open air at a distance of 10m from the lidar was successfully measured in outdoor conditions. From these experimental results, the detection sensitivity of the receiver was evaluated by calculated using a biaxial lidar model. As a result, it was confirmed that H2 gas concentration of 1% up to a distance of 10m was able to detect using a Fresnel lens of aperture 305mm. This result showed that the eye-safe Raman lidar could be used for hydrogen gas leakage detection.
<メディア情報,ユーザ・インタフェース>
<音声画像処理・認識>
  • 茂出木 敏雄
    2010 年 130 巻 7 号 p. 1159-1167
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Using our previously developed audio to MIDI code converter tool “Auto-F”, from given vocal acoustic signals we can create MIDI data, which enable to playback the voice-like signals with a standard MIDI synthesizer. Applying this tool, we are constructing a MIDI database, which consists of previously converted simple harmonic structured MIDI codes from a set of 71 Japanese male and female syllable recorded signals. And we are developing a novel voice synthesizing system based on harmonically synthesizing musical sounds, which can generate MIDI data and playback voice signals with a MIDI synthesizer by giving Japanese plain (kana) texts, referring to the syllable MIDI code database. In this paper, we propose an improved MIDI converter tool, which can produce temporally higher-resolution MIDI codes. Then we propose an algorithm separating a set of 20 consonant and vowel phoneme MIDI codes from 71 syllable MIDI converted codes in order to construct a voice synthesizing system. And, we present the evaluation results of voice synthesizing quality between these separated phoneme MIDI codes and their original syllable MIDI codes by our developed 4-syllable word listening tests.
  • 鄭 光, 大恵 俊一郎
    2010 年 130 巻 7 号 p. 1168-1176
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    In this paper, a new method to detect thin and slow smoke for early fire alarm by using frequency image has been proposed. The correlation coefficient of the frequency image between the current stage and the initial stage are calculated, so are the gray image correlation coefficient of the color image. When the thin smoke close to transparent enters into the camera view, the correlation coefficient of the frequency image becomes small, while the gray image correlation coefficient of the color image hardly change and keep large. When something which is not transparent, like human beings, etc., enters into the camera view, the correlation coefficient of the frequency image becomes small, as well as that of color image. Based on the difference of correlation coefficient between frequency image and color image in different situations, the thin smoke can be detected. Also, considering the movement of the thin smoke, miss detection caused by the illustration change or noise can be avoided. Several experiments in different situations are carried out, and the experimental results show the effect of the proposed method.
  • 坂部 史生, 村川 正宏, 小林 匠, 樋口 哲也, 大津 展之
    2010 年 130 巻 7 号 p. 1177-1185
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    In this paper, we propose a video compression method for video surveillance applications, which varies coding efficiency based on importance of a scene. This method attempts to reduce the video data size according to a coding efficiency which is inversely proportional to unusualness in the recorded data; the lower coding efficiency for the more unusual scenes. The unusualness is defined as deviation from usual motions, which can be statistically learnt by using cubic higher-order local auto-correlation (CHLAC) in combination with a linear subspace method. We conducted an experiment using actual video data to confirm effectiveness of the proposed method. In the experiment, we changed frame-rates for the actual video based on the coding efficiency inversely proportional to unusualness. The data size of the actual video has been reduced to 48% by the proposed method. From the results of the experiment, the proposed method shows effectiveness in reducing the data size by looking into the content of the scene and making use of semantic redundancy.
  • 奥長 潤也, 大松 繁, 吉岡 理文, 小坂 利壽
    2010 年 130 巻 7 号 p. 1186-1191
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    The purpose of this paper is to find a central position of the seal impression in a bill automatically using a template matching method. The seal impression in a bill is affixed after the bill has been printed. Hence, the central position of the seal impression may be different for each bill. Since the seal impression includes much valuable information to check whether it is true or false bill, we must find the precise central position of the bill. Therefore, we use a template matching to find the central position. Background images are different for 1,000, 2,000, 5,000, and 10,000 JPY, although the seal impressions are the same. It may cause erroneous detection of the position. Then it is preferable to remove the background images. We propose a filtering algorithm to remove them. After that we adopt several algorithms to speed up the template matching. Finally, we show that the proposal method is useful to find the precise position of the seal impression.
  • 柴田 進, 西郷 知泰, 林 真司, 田口 亮, 梅崎 太造
    2010 年 130 巻 7 号 p. 1192-1197
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    This paper has presented the Shape from focus method using segmental focal images. The segmental focal image is defined by various focal lengths in one field of view. By using these images, the focal length at particular position changes in accordance with shifting the field of view. The experiment was conducted in a smooth, uninterrupted manner and allowed us a high-speed concatenated three-dimensional reconstruction. Furthermore, combining this method with frequency transfer analysis, a high-precision measurement was possible even if the shooting interval between source images were enlarged. 3-dimensional high speed measurement can be achieved by combining these two methods effectively.
<ソフトコンピューティング・学習>
  • 鵜根 弘行, 銭 飛, 平田 廣則
    2010 年 130 巻 7 号 p. 1198-1206
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Recently, there are numerous services, such as audio and video streaming, which are real-time communication services which guarantee bandwidth and low latency, called “QoS (Quality of Service).” To ensure the bandwidth and shorten the communication delay, it is required to solve the problem of finding paths which satisfy the QoS constraints made of two or more metrices, such as link costs and communication delay. However, this problem is shown as an NP-complete problem. We introduce a routing algorithm with a reinforcement learning scheme to obtain the paths which satisfy two or more constraints.
  • 長谷川 嗣人, 石亀 篤司, 安田 恵一郎
    2010 年 130 巻 7 号 p. 1207-1212
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    This paper proposes a Multi-Objective Particle Swarm Optimization (MOPSO) with particle density. In the proposed method, density of particles around every Pareto solution is calculated and a Pareto solution with low particle density is selected as gbest which is a best position visited thus far by all of the particles. Then, it is validated through a simulation with some Multi-Objective problems comparing to the sigma method which is the conventional to select gbest.
  • 井口 翔太, 小圷 成一, 岡本 卓, 平田 廣則
    2010 年 130 巻 7 号 p. 1213-1218
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    In recent years, the study of hardware implementation of Neural Networks (NN) has been getting more important. In particular, Block-Based Neural Networks (BBNN) which are one of NN have been attracted attention. However, the conventional BBNN are analogue NN (ANN). The digital hardware implementation of ANN is very difficult, because the input and output signals are represented as analogue values. Pulsed Neural Networks (PNN) which adopt a pulsed neuron (PN) model instead of the AN model have been proposed in order to solve this problem. The input and output signals of PNN are represented as a series of pulses, and thus the digital hardware implementation of PNN becomes easy. In this paper, we propose Block-Based Pulsed Neural Networks (BBPNN) introducing the PN model into BBNN in order to faciliate the implementation of NN on digital hardware. We use particle swarm optimization (PSO) for optimization of weights of BBPNN, because PSO can produce a globally optimum solution of nonlinear continuous optimization problems in practicable calculation time by high accuracy. To evaluate the proposed BBPNN, we apply them to XOR problem and autonomous mobile robot control problems. Computational experiments indicate that the proposed BBPNN and the conventional BBNN can produce about the same results.
  • 折登 由希子, 山本 久志, 辻村 泰寛, 神林 靖
    2010 年 130 巻 7 号 p. 1219-1227
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    The portfolio optimizations are to determine the proportion-weighted combination in the portfolio in order to achieve investment targets. This optimization is one of the multi-dimensional combinatorial optimizations and it is difficult for the portfolio constructed in the past period to keep its performance in the future period. In order to keep the good performances of portfolios, we propose the extended information ratio as an objective function, using the information ratio, beta, prime beta, or correlation coefficient in this paper. We apply the simulated annealing (SA) to optimize the portfolio employing the proposed ratio. For the SA, we make the neighbor by the operation that changes the structure of the weights in the portfolio. In the numerical experiments, we show that our portfolios keep the good performances when the market trend of the future period becomes different from that of the past period.
  • 木野 千晶, 鈴木 喜雄, 武宮 博
    2010 年 130 巻 7 号 p. 1228-1237
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Scientific Concept Vocabulary (SCV) has been developed to actualize Cognitive methodology based Data Analysis System: CDAS which supports researchers to analyze large scale data efficiently and comprehensively. SCV is an information model for processing semantic information for physics and engineering. In the model of SCV, all semantic information is related to substantial data and algorisms. Consequently, SCV enables a data analysis system to recognize the meaning of execution results output from a numerical simulation. This method has allowed a data analysis system to extract important information from a scientific view point. Previous research has shown that SCV is able to describe simple scientific indices and scientific perceptions. However, it is difficult to describe complex scientific perceptions by currently-proposed SCV. In this paper, a new data structure for SCV has been proposed in order to describe scientific perceptions in more detail. Additionally, the prototype of the new model has been constructed and applied to actual data of numerical simulation. The result means that the new SCV is able to describe more complex scientific perceptions.
<情報処理・ソフトウェア>
  • 田川 聖治, 佐々木 幸紀, 中村 弘幸
    2010 年 130 巻 7 号 p. 1238-1246
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    The structural design of Surface Acoustic Wave (SAW) filters is formulated as a constrained multi-objective optimization problem. Then three Evolutionary Multi-criterion Optimization (EMO) algorithms based on Differential Evolution (DE), namely, Multi-Objective DE (MODE), Non-dominated Sorting DE (NSDE), and Generalized DE 3 (GDE3), are applied to the three- and two-objective optimization problems of a balanced SAW filter. In order to compare the performances of the above EMO algorithms, several criteria including hypervolume are evaluated. As a result, it is shown that the performance of the EMO algorithm depends on the number of objective functions. Besides, in order to clarify the tradeoff relationship among the objective functions of the three-objective optimization problem, Principal Component Analysis (PCA) is employed.
  • 佐賀井 重雄, 朱牟田 善治
    2010 年 130 巻 7 号 p. 1247-1252
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    We are developing a tool which can simulate the emergency restoration process of electric power distribution facilities in order to examine the methods to speed-up the process.
    This report is aiming following two objectives; 1)describing an overview of the tool “restoration simulator” that simulates the emergency restoration process of the distribution facilities after a disaster, 2)evaluating effectiveness of various restoration methods to speed-up the restoration process using the tool.
  • 朱牟田 善治, 登藤 貴毅, 高橋 健吾, 石川 智巳
    2010 年 130 巻 7 号 p. 1253-1261
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    This paper describes a seismic damage estimation method for electric power distribution equipment to support its emergency restoration activity. The predominant factors associated with the seismic damage degree of electric power distribution equipment are classified based on the analysis of its seismic damage record of the 2007 Niigata-Ken Chuetsu-Oki earthquake. Based on the analysis, a seismic damage estimation model is proposed. The proposed model is applied to an actual damaged area and compared to existing damage estimation models used by local governments. Finally, the effectiveness of the proposed model and future subjects are discussed.
  • 坂本 浩, 萬代 雅希, 渡辺 尚
    2010 年 130 巻 7 号 p. 1262-1270
    発行日: 2010/07/01
    公開日: 2010/07/01
    ジャーナル フリー
    Machine to machine (M2M) is a promising technology to achieve an ubiquitous environment by uniting machines and machines over the Internet. The network used for M2M consists of core network and access network. This paper discusses effective controls of the wireless access network for M2M. Among typical examples of the wireless access network for M2M is a wireless sensor network (WSN). WSN for M2M may require energy efficiency, high reliability and throughput. For these requirements, in this paper, we propose a scheme to build a hierarchical sensor network using smart antenna. The proposed scheme uses omni-directional antennas together with smart antennas. Since smart antennas can extend communications distance, the proposed scheme enables reduction of number of hops to reduce the traffic load on relay nodes. As a result, the energy consumption, data collection ratio and throughput can be improved. We implement the proposed scheme on a real testbed. The testbed uses UNAGI as smart antenna nodes and Mica Mote as sensor nodes. In addition to the fundamental evaluation on the testbed, we simulate large-scale sensor networks. The results show the effectiveness of the proposed hierarchical sensor network with smart antennas.
特集研究開発レター
<ソフトコンピューティング・学習>
部門記事
feedback
Top