Al-doped zinc oxide (AZO) or Ga-doped zinc oxide (GZO) films with c-axis orientation have been deposited on glass substrates by pulsed laser deposition (PLD) using an ArF excimer laser (λ= 193 nm) or Nd:YAG laser (λ= 532, 355, 266 nm). The film deposition took place at substrate temperatures of room temperature (25 °C) ∼ 200 °C. For the GZO film deposited by irradiating a pulsed laser beam of energy density of 1 J/cm2 at repetition frequency of 10 Hz, we obtained the 416 nm-thick GZO film with the lowest resistivity of 2. 89×10-4 Ω·cm for the case at a substrate temperature of room temperature. An average transmittance of more than 80 % in the visible region was obtained for the GZO films fabricated, providing useful functionality as TCO films in the visible region. Moreover, we studied whether the film properties were subject to influences of the position of the plume (in the central axis or at the periphery) generated between the substrate and the target.
Creeping discharge and light emission characteristics in Ne gas for the configuration with plane electrodes on the dielectric plate with a stripe backside electrode have been investigated. In the case of the configuration without backside electrode, discharge inception voltage increased monotonically with increasing gas pressure. For the configuration with a backside electrode discharge inception voltage decreased when the backside electrode became wide. Moreover, it tended to saturate over 30kPa for the configuration with a backside electrode. By observations of discharge using image-intensified CCD camera, discharge developed along the dielectric plane surface at 30kPa on 60kPa, on the other hand, discharge grew in the gaseous space away from dielectric surface. These facts suggest that the effect of backside electrode strongly depends on the their width and gas pressure.
In order to create the implant material which has high biocompatibility, a new hydroxyapatite (HAp) coating method, the laser-assisted laser ablation (LALA) method, has been developed. In this method, control of assist laser irradiation timing (delay time after an ablation laser irradiation) is very important for the quality of a HAp coating layer. We studied the influence of the timing on adhesion strength , crystallinity and Ca/P value of the HAp coating. Quality of the HAp coating layer obtained at a delay time of a few microseconds was extremely high. On the other hand, the layer obtained over 10 microseconds had more amorphous component and contained other calcium phosphate compounds from HAp. Using the present method, we succeeded in fabricating thin film (≦1 μm) of HAp with high crystallinity.
Three kinds of clock recovery circuit for 40-Gbit/s optical RZ signals are fabricated and their characteristics are compared. The three schemes comprise a phase lock loop (PLL) including a voltage-controlled oscillator, an injection locking, and a logic operation by T-flip-flop. The injection-locking scheme is then applied for 80-Gbit/s signals, where the input optical signal is converted into a 40-GHz signal by an optical modulator driven by the output of the locking circuit. Characteristics as hold-in rage, frequency response, and phase noise are measured and compared among the schemes. The properties of the PLL and the injection locking circuit were relatively similar and those of the TFF circuit were uncharacteristic compared with others. Performance of the circuits was confirmed by error free transmission and by stable phase noise over 24 hours. Phase noise transfer properties of the three schemes are briefly discussed.
A major drawback of orthogonal frequency division multiplexing (OFDM) is the high peak-to-average power ratio (PAPR) of the transmitted signal. PAPR reduction techniques by using neural networks have been proposed to reduce the PAPR problem in OFDM transmitter. These techniques require side information to be transmitted from the transmitter to the receiver in order to recover the original data symbol from the received signal. In this paper, we propose a novel technique to reduce PAPR of OFDM signal. Proposed technique is based on Tone Injection(TI) and dose not use any side information to be transmitted from the transmitter to the receiver. Moreover, the proposed model is designed with VHDL for a FPGA device, and evaluated the performance.
Fast responses without any steady-state errors are required for control of a sinusoidal PWM inverter. There are two key points. One is that its reference voltage waveform is not constant but sinusoidal. It is necessary to design a compensator by considering these sinusoidal variations of the reference waveform to get a good steady-state and fast transient response. A sinusoidal tracking compensator which is based on the optimal control and the internal model principle has been proposed by the authors to realize a response with no deviation for a periodic sinusoidal reference input. The other is that compensator and feedback (FB) gains have a trade-off between responses and sensitivities to noises and they can not be too large. Feedforward(FF) compensation can improve responses without alternating system stabilities. This paper proposes a PWM control system which has a sinusoidal FB and an inverse system FF compensators. According to the inverse system FF compensation, appropriate nominal control input values are determined to give the reference steady-state response. Therefore the input values are computed by deconvolving a set of system response equations inversely. The proposed system is investigated and validated through theoretical and experimental results for a rated, an unbalanced, and a rectifier loads by using a simulator Matlab and a DSP system at the switching frequency of 20kHz. The proposed control showed better performances than the original FB and the preview controls.
Stability analysis of a linear uncertain polynomial system in real parameter space is one of central themes of recent control theory. It is generally very difficult to analyze the stability of such a system because the number of its characteristic polynomials to be analyzed becomes enormous. A parallelotope of polynomials is a typical type of uncertain polynomials. Though a number of analysis methods for such a parallelotope are at our hand, they mainly focus on stability check of given polynomials and hardly give information on how for the certain interested parameter can be extended for stability. To solve this problem, we propose a novel tool for stability analysis called “Stability Feeler". This concept leads to an algorithm which determines D-stability intervals along a given line in the polynomial coefficient space. With the tool, stability problems of parallelotopes can be successfully resolved. Two numerical examples are shown to support this concept.
In controller design of dynamical systems, it is desirable that the controllers can guarantee the stability for not only nominal plants but also perturbed ones, because the plant models may have some uncertainties in their parameters. By this motivation, several researches for the robust controller design have been conducted. In this paper, a feedback controller design method is considered, which achieves enlarging the degree of robust stability for the linear SISO closed-loop systems. Choosing stability radius for the plant parameters as the degree of robust stability, and regarding this as an object function, we formulated this maximization as an optimization problem with constraints coming from prescribed pole placement. The difficulty of this formulation is that while pole placements give linear constraints, the object function is not a smooth one, which calls for an optimization technique for non-differentiable functions. To cope with this, we successfully used the bundle method. We show some numerical examples to illustrate our proposed design method.
In this paper, we propose a new decision support system (DSS) for dealing stocks which improves the traditional technical analysis by using neural networks. In the proposed system, neural networks are utilized in order to predict the “Golden Cross” (“Dead Cross”) several weeks before it occurs. Computer simulation results concerning the dealings of the TOPIX and Nikkei-225 confirm the effectiveness of the proposed system.
Mobility enhancement technologies by incorporating strain in MOSFETs have been recognized as key technologies for scaled CMOS devices. The most promising channel materials for n- and p-channel MOSFETs are tensily-strained Si and compressively-strained Ge (SiGe), respectively, from the viewpoint of their high mobility values. In this paper, dual channel CMOS structures with strained Si-on-insulator (strained-SOI)-nMOSFETs and strained SiGe-on-insulator (strained-SGOI)-pMOSFETs are demonstrated as well as their high channel mobility and current drive enhancements. Strained Si channels on a relaxed SGOI substrate and Ge-rich strained SGOI channels are located on the nMOS and pMOS regions of the same wafer, respectively. The dual channel structure was fabricated by a CMOS process combined with the Ge condensation process, in which the epitaxially grown SiGe layer on the SOI substrate was locally oxidized at high temperatures. As a result, significant electron- and hole-mobility enhancements for the strained SOI and SGOI channels were observed as well as the drain current enhancements. Based on the measured mobility for the nMOS and pMOS channels in the CMOS devices, CMOS performance enhancement of 30% was estimated.
Fast-recovery and low on-state-forward-voltage drop (Vf) in pin-diodes can be simultaneously obtained with a relaxed SiGe crystal anode layer. We have successfully fabricated 280V-class SiGe pin diodes with a recovery time of 160ns and Vf of 0.86V at a forward current density of 100A/cm2. A device simulation predicts a possibility of Vf=0.7V@100A /cm2 and a recovery time of 20ns by controlling carrier lifetimes at p and i layers independently with SiGe.
There are many kinds of life infrastructure such as gas and water pipes, communication and power cables, are buried in the floor or wall of buildings. The layout information of these structures is very important in maintenance and repairing constructions. It is expected that by launching an acoustic wave directly into a pipe, the position of the discontinuities of the pipe can be estimated from the reflective waves precisely. In this paper, detection of the discontinuities of gas pipe is studied experimentally with an acoustic chirp signal. In order to distinguish L-junction and T-branch, the propagations of an acoustic wave are computed and studied by FDTD analysis. A differential method is proposed to reduce the directive wave. The experimental results show that the method improved distinctness of the reflective signals. It is also confirmed experimentally that the sidelobes of the compressed pulse are decreased by multiplying a Hamming window function to the output chirp signal. The proposed system is able to detect the discontinuities in 36m long gas pipe, and distinguish the L-junction and T-branch according to their reflection characteristics.
The observed images are often corrupted by Gaussian noise. If the image is embedded in small-amplitude Gaussian noise, the noise can be removed by applying Wiener filter. Recently, the BayesShrink wavelet method has attracted considerable attention as a denoising technique. In this paper, we propose a method for removal of Gaussian noise of large amplitude as well as of small one, which can not be removed only by exploiting the BayesShrink wavelet method. Our approach is a combination of the BayesShrink wavelet method with the directional adaptive center weighted median filter. Applying the proposed method to an image corrupted by large-amplitude Gaussian noise, a clean image can be obtained.
This paper presents an adaptive scheme to estimate the time-derivative of a signal without any knowledge of its dynamics. The adaptive estimator is formulated as an adaptive observer with the σ-modification update law. The algorithm is applied to an adaptive differential filter for the detection of moving objects within a scene acquired by a stationary camera. We adopt four nonlinear intensity functions to reduce the influence of measurement noises. The MATLAB simulations are performed to demonstrate its performance.
This paper addresses the fuzzy control problem for a class of discrete-time nonlinear systems with disturbance. In case not all the system states are available, state observer is proposed. Consequently, the whole system behavior can be attributed to a kind of the standard singularly perturbed form. The system design is based on the estimated values from the state observer. It is shown that the proposed fuzzy controller guarantees the output of the considered system follows the desired value while maintaining all signals involved in the system stable.
This paper proposes a new tuning method for PI( proportional plus integral ) controllers based on partial model matching to refer to the standard 2nd order lag model. The method consists of four steps : (1) to measure step responses of a 1st order lag plus dead time process controlled by PI controller, (2) to extract features of the step responses, (3) to estimate damping coefficient ζ and natural angular frequency ωn of the control system, and (4) to change control parameters Kp, Ti to adjust the damping coefficient and the natural angular frequency to the desired values. The result of simulation show that this method is so practical that prediction errors to step responses after changing parameters are small and it is easy to evaluate characteristics of the control system after changing parameters.
In this paper put forward are neuron-type models, i.e., neural network model, wavelet neuron model and three layered wavelet neuron model(WV3), for estimating traveling time between signalized intersections in order to facilitate adaptive setting of traffic signal parameters such as green time and offset. Model validation tests using simulated data reveal that compared to other models, WV3 model works very fast in learning process and can produce more accurate estimates of travel time. Also, it is exhibited that up-link information obtainable from optical beacons, i.e., travel time observed during the former cycle time in this case, makes a crucial input variable to the models in that there isn't any substantial difference between the change of estimated and simulated travel time with the change of green time or offset when up-link information is employed as input while there appears big discrepancy between them when not employed.
This paper presents a method of position sensorless control of ultrasonic motor (USM). The rotor position of USM is estimated based on the input voltage information. The characteristic of input voltage versus rotor position is expressed by a quadratic function. Since the parameters of quadratic function are adjusted on-line, estimated rotor position agrees well with actual rotor position against load torque and motor temperature changes. Position sensorless control is achieved by using estimated rotor position instead of measured rotor position. The validity of the proposed method is confirmed by experimental results.
The purpose of this paper is to investigate the performance of the associative memory model using Aihara's chaotic neural network with different activation functions. Sigmoid function, a monotonous function, was used in Aihara's original model. However, in the static associative memory, it is reported that the storage capacity of the network is improved when a non-monotonous function is used as the activation function. To improve the associative ability of chaotic neural network, kinds of non-monotonous functions have been proposed to serve as activation function. This paper investigates their difference as to retrieval ability, and proposes an advanced non-monotonous function. By computer simulation, we discuss about what kind of shape is good to improve the associative ability of chaotic neural network.
To evaluate a train schedule from the viewpoint of passengers' convenience, it is important to know each passenger's choice of trains and transfer stations to arrive at his/her destination. Because of difficulties of measuring such passengers' behavior, estimation methods of railway passengers' flow are proposed to execute such an evaluation. However, a train schedule planning system equipped with those methods is not practical due to necessity of much time to complete the estimation. In this article, the authors propose a fast passengers' flow estimation method that employs features of passengers' flow graph using preparative search based on each train's arrival time at each station. And the authors show the results of passengers' flow estimation applied on a railway in an urban area.