A gas sensing technique on a fire site was developed. An output of is-TPG in a necessary low frequency for gas sensing controlled was obtained by optimizing the seeder so far. A reflection spectrum of an architectural material that used N2O gas that was the model gas of CO gas generated at a fire was successfully obtained, and a realizability of remote spectrum sensing in the THz range was showed experimentally. An attenuation rate of the THz-wave when the long distance was propagated under the atmospheric pressure was measured experimentally, and an error margin with the calculation was within 10 %. A calculation result was able to endure practical use in the THz range to which the accuracy of the calculation was doubted.
A demand for non-contact and nondestructive internal defect detection is increasing. Using a photoacoustic effect, an internal defect could be detected by a low power semiconductor laser and an ultrasonic sensor. But the sensitivity of the ultrasonic sensor is low in detecting a photoacoustic signal. Then, an internal defect detection using a photoacoustic and self-coupling effect of semiconductor laser has been studied. To generate photoacoustic signal, a semiconductor laser with output power of 20mW is focused by a lens and is illuminated at 45 degrees on the sample. To detect photoacoustic signal by the self-coupled effect, another laser with output power of 10mW is illuminated at right angle. Sample defect can be detected by illuminating these laser lights at the same point on the sample. The self-coupling sensor can detect an internal defect with more sensitivity than the ultrasonic sensor does. An edge effect and a frequency characteristic are also observed. As a result, it was able to detect a very small defect of 0.07mm.
A well-known control system that can reduce the adverse effects of disturbances is a disturbance observer. Whenever we apply disturbance observer, disturbance frequency should be known and constant. However, in many cases of industrial systems disturbance frequency is varied for some frequency range. Therefore, it may be difficult to reduce the adverse effect of such disturbance by use of the traditional disturbance observer. In this paper, a design method of disturbance attenuation system that can cope with the frequency variation (DOFV) is proposed. The main idea of this design method is to combine DOFV with frequency estimator that can estimate disturbance frequency in real time. Even though the proposed disturbance attenuation system is low degree and low gain controller, it has superior steady state characteristic. Numerical simulations and experiments show the usefulness of the proposed disturbance attenuation system.
To display the break of paper, we propose the new method for calculating transformation of paper and for representing a minute structure of paper. Paper is defined as a “spring-mass” cubic mesh model. And, to express the flow of the paper fiber, some cells of the mesh are combined to one group. By giving “spring-mass” parallelepiped covering the group, the bending stress is added to the model. For minuteness expression of paper, display data is created from the model by the following 2 steps. At first, thickness of the mesh model is compressed, then many points are added on separated edges of the mesh to display jagged shape. As the result of computer simulation in various conditions, realistic break lines of paper were represented.
An entanglement purification of a quasi-Bell state from (3, 1) quantum error correcting code is considered. It is shown that the fidelity between a quasi-Bell state decohered by a bit-flip channel and an ideal quasi-Bell state is improved by the purification for wide range of the bit-flip probability of the channel.
An example of code that has quantum gain with both error probability and information criteria is shown. It is shown that BER for the codes is much smaller than that without coding even when quantum gain with information criterion is positive.
We propose a system for visually-impaired people to see a common document without disturbing others when they see a piece of paper together. The system captures the image of a document by a camera and finds the document region in the image. Next, it determines transformation matrix between the object and camera coordinate systems based on the assumption of a rectangular document. Finally the system presents a rectified document image on the display of a portable computer. The result of subjective evaluation by human is that for more than 80% of the samples, the rectified images are easier to see than the input image.
The particle swarm optimizer (PSO) is a stochastic, population-based optimization technique that can be applied to a wide range of applications. This paper presents a random time variable PSO algorithm, called the PSO-RTVIWAC, introducing random time-varying inertia weight and acceleration coefficients to significantly improve the performance of the original algorithms. The PSO-RTVIWAC method originates from the random inertia weight (PSO-RANDIW) and time-varying acceleration coefficients (PSO-TVAC) methods. Through the efficient control of search and convergence to the global optimum solution, the PSO-RTVIWAC method is capable of tracking and optimizing the position evaluate in the highly nonlinear real-time location systems (RTLS). Experimental results are compared with three previous PSO approaches from the literatures, showing that the new optimizer significantly outperforms previous approaches. Simply employing a few particles and iterations, a reasonable good positioning accuracy is obtained with the PSO-RTVIWAC method. This property makes the PSO-RTVIWAC method become more attractive since the computation efficiency is improved considerably, i.e. the computation can be completed in an extremely short time, which is crucial for the RTLS. By implementing a hardware design of PSO-RTVIWAC, the computations can simultaneously be performed using hardware to reduce the processing time. Due to a small number of particles and iterations, the hardware resource is saved and the area cost is reduced in the FPGA implementation. An improvement of positioning accuracy is observed with PSO-RTVIWAC method, compared with Taylor Series Expansion (TSE) and Genetic Algorithm (GA). Our experiments on the PSO-RTVIWAC to track and optimize the position evaluate have demonstrated that it is especially effective in dealing with optimization functions in the nonlinear dynamic environments.
This paper proposes a power efficient access method by polling for wireless mesh networks, which is robust to channel variation due to movement in surrounding communication environment. In the proposed method, all intermittent nodes periodically repeat ID transmission and succeeding short period of receiving state. A node having a packet to transmit goes into continuous receiving state. Right after receiving an ID transmission from an arbitrary node, it transmits the packet to the node of preceding ID transmission. In a conventional access method for wireless mesh network, a transmitting node specifies a receiving node well ahead of transmission, leading to a possible change in wireless link quality at the time of transmitting packet from the time of determining receiving node. The proposed access method is robust to wireless channel variation, in the sense that a transmitting node chooses a receiving node just before transmitting packet. The simulation results show its robustness to wireless channel variations. With the proposed access method, a whole network consumes about one sixth power compared with those f a conventional method.
There are a lot of nonlinear systems in the process control systems. However, it is difficult to deal with the nonlinear property by linear controller. Then, various methods have been studied in this field. On the other hand, a scheme called boosting is proposed in the machine learning field. This scheme can obtain the highly accurate condition by combining some less-accurate conditions. The boosting can obtain better results compared with the neural network under the less number of learning data. In this paper, we propose a design method of nonlinear PID control systems using boosting algorithm. The original boosting is only possible to deal with two-valued variable, so we extend the algorithm for function approximation using it. Finally, the simulation examjples are demonstrated in order to investigate the effectiveness of proposed scheme.
Since local signals appear time-locally and their waveforms are steep, conventional signal processing methods are generally inadequate to detect them. In this paper, by focusing on amplitude distribution forms of an observed signal, a simple signal processing method is proposed to detect the local signals and extract their waveform shapes simultaneously. Concretely, a characteristic waveform template which consists of representative data series of a simplified target local signal is newly introduced, and a certain event regulated with it is adopted. The local signals can be detected by evaluating whether the amplitude distribution forms of observed signals join to the event, and substitution of a conditional event for a joint event improves a detection performance. In addition, the proposed method has been extended to extract waveform shapes by introducing an enhanced waveform template, and automatic detection and extraction of waveform shapes can be performed simultaneously. The proposed method is applied to simulation signal data, and its effectiveness has confirmed from detection performances and extracted waveform shapes for the local signals.
1-bit signal processing based on delta-sigma modulation has been studied for hardware implementation of signal processng systems. In the 1-bit signal processing, finite word-length problems such as overflow and coefficent quantization error occur. To solve the problems, new design method with state space is proposed in this paper. Digital filters are designed to show the feasibility of the method. Firstly, L1/L2-sensitivity is shown to evaluate coefficient quantization error and L2 scaling constraints to prevent overflow. Secondly, state space equation is shown and L1/L2-sensitivity and L2-scaling constraints are extended to take filter structure and oversampling effects into account. Finally, the proposed method is shown to attain higher SNR than conventional ones.
In general, the production quantities and delivered goods are changed randomly and then the total stock is also changed randomly. This paper deals with the production and inventory control using the Dynamic Bayesian Network. Bayesian Network is a probabilistic model which represents the qualitative dependence between two or more random variables by the graph structure, and indicates the quantitative relations between individual variables by the conditional probability. The probabilistic distribution of the total stock is calculated through the propagation of the probability on the network. Moreover, an adjusting rule of the production quantities to maintain the probability of a lower limit and a ceiling of the total stock to certain values is shown.
In this paper, we describe a person's brain activity when he/she sees an emoticon at the end of a sentence. An emoticon consists of some characters that resemble the human face and expresses a sender's emotion. With the help of a computer network, we use e-mail, messenger, avatars and so on, in order to convey what we wish to, to a receiver. Moreover, we send an emotional expression by using an emoticon at the end of a sentence. In this research, we investigate the effect of an emoticon as nonverbal information, using an fMRI study. The experimental results show that the right and left inferior frontal gyrus were activated and we detect a sentence with an emoticon as the verbal and nonverval information.
Fatigued bills have harmful influence on daily operation of Automated Teller Machine(ATM). To make the fatigued bills classification more efficient, development of an automatic fatigued bill classification method is desired. We propose a new method to estimate bending rigidity of bill from acoustic signal feature of banking machines. The estimated bending rigidities are used as continuous fatigue level for classification of fatigued bill. By using the supervised Self-Organizing Map(supervised SOM), we estimate the bending rigidity from only the acoustic energy pattern effectively. The experimental result with real bill samples shows the effectiveness of the proposed method.
A new evolutionary method named “Genetic Network Programming with control nodes, GNPcn” has been applied to determine the timing of buying or selling stocks. GNPcn represents its solutions as directed graph structures which has some useful features inherently. For example, GNPcn has an implicit memory function which memorizes the past action sequences of agents and GNPcn can re-use nodes repeatedly in the network flow, so very compact graph structures can be made. GNPcn can determine the strategy of buying and selling stocks of multi issues. The effectiveness of the proposed method is confirmed by simulations.
In order to efficiently obtain an approximate solution of the traveling salesman problem (TSP), extended changing crossover operators (ECXOs) which can substitute any crossover operator of genetic algorithms (GAs) and ant colony optimization (ACO) for another crossover operator at any time is proposed. In this investigation our ECXO uses both EX (or ACO) and EXX (Edge Exchange Crossover) in early generations to create local optimum sub-paths, and it uses EAX (Edge Assembly Crossover) to create a global optimum solution after generations. With EX or ACO any individual or any ant determines the next city he visits from lengths of edges or tours' lengths deposited on edges as pheromone, and he generates local optimum paths. With EXX the generated path converges to a provisional optimal path. With EAX a parent exchanges his edges with another parent's ones reciprocally to create sub-cyclic paths, before restructuring a cyclic path by combining the sub-cyclic paths with making distances between them minimum. In this paper validity of ECXO is verified by our C experiments using medium-sized problems in TSPLIB, and it is shown that ECXO can find the best solution earlier than EAX.
This paper proposes a method for automatically extracting term knowledge such as case relations and IS-A relations between words in the headline sentences of the operating manuals for information equipments. The proposed method acquires term knowledge by the following iterative processing: the case relation extraction using correspondence relations between the surface cases and the deep cases; the case and IS-A relation extraction using the compound word structures; the IS-A relation extraction using correspondence between the case structures in the hierarchical headline sentences. The distinctive feature of our method is to extract new case relations and IS-A relations by comparison and matching the case relations extracting from the super and sub headline sentences using the headline hierarchy. We have confirmed that the proposed method to achieve approximately 90% recall and precision for extracting case relations and IS-A relations from operating manuals of a car navigation system and a mobile phone.
Various techniques that improve performance on SMP Clusters have been studied. Most of them use special hardware and non standard protocol, tending to raise their total cost and to spoil their flexibility. We propose CPU_NIC method which improves parallel performance on small-way SMP PC cluster only by loading driver. A proposal system is realizable with non intelligent switches. A proposal method uses the same number of NICs as CPUs and relates CPU and NIC with one to one. The transmitting frames are outputted from NIC related with execution CPU. Thus, by fixing transmitting NIC for every CPU, the sack frames which are easy to generate in communication load sharing to two or more NIC is able to be reduced and parallel processing performance is able to improve. As a result of measuring using NPB benchmarks on SMP cluster which consists of four SMP PCs, FT, MG and CG become 1.12 times, 1.29 times and 3.06 times faster respectively by proposal method, as compared with ordinary system without this method.
When chaotic dynamics is given to the neurons that compose the associative memory model, it searches for stored patterns in a pattern space chaotically. However, it has the fault that the judgment for whether the stored pattern is recollected or not is difficult because its behavior is always chaotic. As all dynamics of the chaotic neurons are chaotic, chaotic transition is repeated. One side, transient-chaotic associative network (TCAN) Lee proposed changes from the state of chaos to the state of stability (non-chaos) transiently. Additionally, it has the fast recollection speed, and has the characteristic, high memory capacity. However, the states of TCAN do not change chaotically. Based on these results, this paper proposes a transient chaotic associative memory model with temporary stay function (TCAMMwithTSF) which has two abilities: one is the fast speed at the states of the model converge to a stored pattern, like TCAN, the other is the ability that it searches the stored pattern in a pattern space chaotically, like chaotic neural networks. Finally, it is verified that what character TCAMMwithTSF has and its usefulness through simulation study.
Many studies for improving the phase noise characteristic are based on evaluating the characteristics of designed circuit. Because the oscillator acts as a nonlinear positive feedback circuit, it is difficult to expect the oscillation frequency and the amplitude accurately. Then, we proposed the Fr oscillator circuit which satisfied the impedance matching requirement. The effective approach can be examined because it is thought that the behavior of the element in the oscillator circuit can be understood by the open circuit of this circuit. This paper describes that the characteristic of the open circuit of the Fr oscillator is corresponding to it of the closed circuit well.
In this research, it aims at the development of the tactile display that information presentation is expressed with phantom sensation and apparent movement by few tactile elements. In this report, we suggest technical method to improve performance for the information presentation by the calibration using neural network. In the results which we performed calibration using suggested method, the effect of the calibration of 40% on the average was achieved for 19 subjects.