The 26th ISCIE International Symposium on Stochastic Systems Theory and Its Applications (Oct. 1994, OSAKA)
-
T. Honjo, K. Tanaka, T. Nakamizo
1995 Volume 1995 Pages
1-6
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
Although LMS adaptive filters have been applied to many signal processing applications, there exists a critical limitation of the conventional adaptive filter when it is applied for a real time signal processing. It is simply because of its required lengthy convolution (filtering) calculation. To overcome the limitation, we propose a new Delayed LMS algorithm which is implemented in Signed-Digit Number system. In the paper, a new pipelined adaptive filter is developed and its detailed analysis is described with an actual VLSI implementation.
View full abstract
-
Hirofumi Yogo, Tadashi Kitamura, Naoki Inagaki
1995 Volume 1995 Pages
7-12
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
A fast and new parameter estimation method is proposed in this paper. The proposed method utilizes the inverse system based on an adaptive digital filter (ADF). As an adaptive algorithm, "Accelerated Stochastic Approximation (ASA) " method is used. The ASA method adopts the time-varying and the optimized convergence factor, and converges within 25 iterations. As the applications of the proposed method, two examples of the parameter estimation of the ARMA model and the natural stop consonants are presented.
View full abstract
-
AKIRA OHSUMI, YOSHIMASA ONO, ATSUHIKO SHINTANI
1995 Volume 1995 Pages
13-18
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In this paper an active control problem is investigated for a class of flexible cantilevered beams tipped with dynamic actuators subject to random disturbance. The mathematical models of the system are hybrid, described by an Euler-Bernoulli type partial and a second-order ordinary differential equations. Since the actuator is tipped the control problem constitutes a kind of boundary control problems. By introducing boundary homogenization approach and the modal representation, it is reduced to the finite dimensional control problem. Simulation studies are provided.
View full abstract
-
Masanori Sugisaka
1995 Volume 1995 Pages
19-25
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper introduces a micro-robot for inspection, which is able to run inside curved underground empty tubes, developed recently in my laboratory. The micro-robot is able to run inside a tube with 100 mm diameter and 3R curvature in forward and backward dirctions by changing its speed based on a program. The experimental studies were performed in order to see the fundamental control functions of the micro-robot by driving it in both up and down and left and right directions. It is concluded from the experiments that the micro-robot developed in my laboratory satisfies the required design specifications.
View full abstract
-
Shin Ichi Aihara
1995 Volume 1995 Pages
27-32
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
Motivating a simple industrial example of constructing a pipe line, we present a stochastic hyperbolic system with a moving boundary region. This hyperbolic system is formulated as an Ito stochastic integral equation in Hilbert spaces. The mean square stability problem is considered under the circumstance that the moving boundary is under control.
View full abstract
-
Yoshiki TAKEUCHI, Takahiro ISHIKAWA
1995 Volume 1995 Pages
33-38
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper is concerned with the problem of optimal transmission of multi-dimensional Gaussian signals through a set of parallel channels with feedback. We especially consider the case where the feedback is corrupted by additive Gaussian noise. Under the constraint on the total power of the signals, we consider the problem of finding the optimal gain matrix, i.e., a set of optimal gains for the channels which maximizes, at each time, the decrease-rate of the estimation errors of the signals. The solution is given by computing the hypothetical optimal gain matrices for the two cases: we use the feedback; and we do not. The optimal gain matrix is determined by the fact which of the two hypothetical gain matrices gives the larger decrease-rate of the estimation errors.
View full abstract
-
Kazutatsu HATAKEYAMA, Mitsuo OHTA, Yoshie KODERA, Minoru FUJITA, Takur ...
1995 Volume 1995 Pages
39-44
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In the field of medical X rays image, it is important how to seize a quantity of being affected by radioactivity to the human body. Then, when dropping affection by radioactivity dose, quantum noise increases on inevitability, and some counter-measure methods become central problem. In this paper, a new trial of wide-sense digital filter with a smoother for estimating and smoothing the non-Gaussian stochastic systems will be proposed in the hierarchical form of successively parameter differentiated types which takes Kalman filter as the first term of algorithm. Then, the effectiveness of proposed recursive algorithm will be experimentally confirmed through several applications to real medical X-ray images.
View full abstract
-
Tomohiro Hachino, Zi-Jiang Yang, Teruo Tsuji, Torao Yanaru
1995 Volume 1995 Pages
45-52
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper proposes a new on-line identification method of continuous time-delay systems from sampled input-output data. In order to track the time-varying system parameters and time-delay, the recursive linear least squares (RLS) method is combined in a bootstrap manner with the genetic algorithm (GA) which has a high potential for global optimization. The time-delay is coded into binary bit strings and searched by the GA, while the system parameters are updated by the RLS method. Furthermore, the proposed method (GALS method) is hybridized with the sequential nonlinear least squares method to improve the speed of convergence. Simulation results show that the hybrid method has superior convergence property to the GALS method and yields excellent estimates in the case that the system changes with time continuously.
View full abstract
-
Yang Wu, Tohru Katayama
1995 Volume 1995 Pages
53-58
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper is concerned with the problem of determining unknown time delay and coefficients of a transfer function. First, when the plant input is a white noise, we show that the time delay can be estimated accurately by correlation analysis. Then, we can determine coeffcients of transfer function by least-squares (LS) method. In order to apply LS method, pre-filtering of input and output observations is necessary to which we use λ transformation-based pre-filtering. Another method to estimate time delay is considered by decoupling the estimations of time delay and system parameters based on the variable projection theory. Results of simulation are included.
View full abstract
-
J. Kim, K. Wada
1995 Volume 1995 Pages
59-64
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In this paper we estimate the parameters of a transfer function using output error(OE) identification, employing the Steiglitz-McBride(SM) and Evans-Fischl(EF) methods. The parameter estimates, as well as the frequency response calculated from these estimates, are compared with the results of an alternative algebraic estimation method.
View full abstract
-
Kunihiko OURA, Masafumi YOSHIOKA, Munihito MATSUDA, Kageo AKIZUKI, Izu ...
1995 Volume 1995 Pages
65-70
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In identification of a process operating in closed-loop control system, it is known that meaningful results can be obtained under experiment conditions such that structure of the process are known, the closed system has at least one delay-time and external noises. To select suitable structure of model is required first to identify the process with unknown structure. In open-loop identification, there are some effective criteria for selection of the structure of model : for example, AIC. But they aren't always for uses in closed-loop identification. In this paper, we make suggestion how to determine model structure for closed-loop identification based on numerical experiments.
View full abstract
-
R.W. Merchant, K. Wada
1995 Volume 1995 Pages
71-76
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
The analysis of discrete-time identification methods for continuous-time system models usually involves the assumption of a piecewise-constant (ZOH) input excitation. However, as has recently been discussed, i e. [1], this assumption is often violated in practice. This paper investigates estimation when the input excitation corresponds to a filtered ZOH signal. An exact discrete-time model representation for the sampled input-output signals is derived, and the effects on estimation investigated through simulation studies. The studies show how the appropriate selection of the various estimation options can yield relatively accurate continuous-time model estimates.
View full abstract
-
Toshiyuki Aoki, Kohji Kamejima
1995 Volume 1995 Pages
77-82
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper examines simultaneous contour matching for an object with facets, which are closed sets in subspaces of 3-D space, by approximating an optical flow.
First, the object's shape is described by using facets within a framework of manifold. However here the relationship between only two facets is considered. Second, an algorithm for 3-D object shape contour matching is derived by using a diffusion field under the condition. Third, a mathematical model of the diffusion field is formulated by the stochastic evolution equality in Hilbert space. The existence and uniqueness of the solution to the diffusion field equations are studied. Fourth, a cost function is proposed and the convergence of the algorithm is demonstrated. Finally, the structure of the optical flow is shown by deriving the vector field matrix, when the relationship between two facets is considered.
View full abstract
-
Jeong-Yong SONG, Kageo AKIZUKI, Hee-Hyol LEE
1995 Volume 1995 Pages
83-87
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper describes how to recognize hand written Hangeul character using the stroke order of the elementary segment. The recognition system is constructed of 4 parts : character input part, segment disassembling part, character element extraction part and character recognition part. The character input part reads the character and performs thinning algorithm. In the segment disassembling part, the input character is disassembled into elementary segments using the direction codes and the feature parameters. In the character element extraction part, we extract the character element using the stroke order and the Knowledge rule. Finally, we are able to recognize the hand written Hangeul characters by assembling the character elements, in the character recognition part.
View full abstract
-
Takayuki Ueta, Toshiharu Hatanaka, Katsuji Uosaki
1995 Volume 1995 Pages
89-93
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
Two-state Markov sequences play important roles in analysis of physical and engineering phenomena. When the Markov sequence cannot be observed directly but through some noisy observation system, a “most likely” estimate of the underlying Markov sequence should be estimated from its noise-corrupted observation sequence. Although the estimate can be obtained by the integer programming approach, it is so tedious for long sequences. In this paper, a simple approach based on the genetic algorithm is proposed to obtain the estimate. A numerical example illustrates its applicability. And restoration of noisy binary images composed by
N x
M pixels is considered as an application of the proposed smoothing method.
View full abstract
-
Kohji Kamejima
1995 Volume 1995 Pages
95-100
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
A stochastic evaluation scheme is presented for morphological event. The morphological event is comprehensively defined via imaging process that evokes an "avalanche of exploration" in image field. This intrinsically non-deterministic imaging process is associated with observed patterns by a conditional probability that is represented on a non-linear diffusion system. The scheme is applied to dynamic detection of fractal patterns.
View full abstract
-
Masaki Suwa, Hiroshi Miyasada, Yusuke Mizuno, Sueo Sugimoto
1995 Volume 1995 Pages
101-106
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In this paper, we propose a method of image restoration based on the simulated annealing(SA) [1],[2], and the wavelet representation(WR)[3]. In the past two decades, there has been much research activity in developing image processing for restoration of two-dimensional noisy images based upon statistical methods. Geman-Geman proposed the Baysian restoration of images based on simulated annealing(SA)[2]. But their restoration method has very slow convergence for getting the MAP(Maximum A Posteiriori) estimate and requires a very large computational cost. To overcome such shortage of their method, we propose here a new method that restores a degraded image based on the Gibbs distribution and the WR. In this method, we compute a WR of a degraded image and obtain the MAP estimate from the WR.
View full abstract
-
Fei Qian, Hironori Hirata
1995 Volume 1995 Pages
107-112
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In an artificial neural network, the behavior modification is accomplished by parameter adjustment. Often, the adjustable parameters in the neural networks are called synaptic weights or connection strengths, but their analogy to real biology reinforcement learning to update has been on pattern classification and simple time-delayed feedback control tasks. There has been no significant work to use reinforcement learning techniques to carry out structural learning. To construct the model of reinforcement learning systems, in this article, we generalize the stochastic cellular automata to stochastic learning cellular automata, which is a combined model of traditional stochastic cellular automata and random environments, and give some definitions of stochastic learning cellular automaton.
View full abstract
-
Yoji MORITA, Shigeyoshi MIYAGAWA
1995 Volume 1995 Pages
113-118
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
A vector autoregressive model(called VAR model) is constructed to investigate behavior of GNP and price in Japan which are influenced by demand and supply shocks. The system model is expressed by
x(
t) +
A0x(
t) ＋
A1x(
t－1) ＋…＋
Akx(
t－
k) ＝
ε(
t)
where
x is a vector with 4 variables; oil price, unemployment rate, real GNP and price. Responses of shocks
ε to the each element of the
x-process are considered under the condition such that
A0 is a non-zero matrix. In order to identify the matrix
A0 , structural restrictions are imposed on ”long run multiplier” of the system and such a system is called ”a structural VAR model”. The unit root and cointegration are checked in association with stationarity of the system and impulse responses and variance decomposition are obtained from the identified model.
View full abstract
-
H. Ashida, A. Sano
1995 Volume 1995 Pages
119-124
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This paper proposes a method for estimating the mixed spectrum which is composed of line and continuous spectra, the latter of which is characterized by an AR or ARMA noise model. Line spectrum is represented by multiple sinusoids. In order to avoid simultaneous minimization of a prediction error criterion with respect to all unknown parameters, we give an efficient iterative algorithm for estimating the frequencies of the sinusoids and other parameters separately. By adopting the genetic algorithm in choice of initial values of the AR or ARMA parameters in the iterative estimation, we can attain a globally optimal estimates of unknown parameters. The frequency estimate is given by the Toeplitz approximation method or its extension using a shifted correlation matrix of observed signals. The effectiveness of the proposed algorithm is validated in numerical simulations.
View full abstract
-
Akira IKUTA, Mitsuo OHTA
1995 Volume 1995 Pages
125-130
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In the actual sound environment, the random signal often shows a complex fluctuation pattern apart from a standard Gaussian distribution. In this study, an evaluation method for the sound environmnetal system is proposed in the generalized form applicable to the actual stochastic phenomena, by introducing two types of information processing methods based on regression models using an expansion series type and Fuzzy probability. The effectiveness and validity of the proposed method are confirmed by applying it to the observed data in the actual sound environment.
View full abstract
-
Tokuo Fukuda
1995 Volume 1995 Pages
131-136
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In this paper, the author investigate the properties of the newly developed concept of
fuzzy random vectors(
FRVCs), which have intrinsically both properties of fuzziness and randomness and they are considered to be obtained as fuzzy perceptions of ordinary non-fuzzy random vectors.
First, the concept of FRVCs and some of their measurability properties are reviewed and investigated. Secondly, applying the multi-valued logic, the expectations and the second statistical moments of FRVCs are defined. Finally, some properties of statistical moments of FRVCs are investigated theoretically.
View full abstract
-
Akihiro Takatsuka, Hideki Nio, Kazuhiro Ichikawa, Sueo Sugimoto
1995 Volume 1995 Pages
137-142
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
The feedforward layered neural networks have been applied to the various classification problem and control. The generalization capability of neural networks is one of the most important properties to be acquired such that it depends on the numbers of units or weights. In this paper, we propose a fast and effective structural learning algorithm to construct a suitable structure for neural networks and apply to the pattern recognition problem for handwritten numerical characters. At every learning step evaluations for unknown data are computed , and decide the stopping time of learning (test set validation). Furthermore we consider the validity to select an optimal neural network model by information criteria.
View full abstract
-
Yong-guang Zhang, Masanori Sugisaka
1995 Volume 1995 Pages
143-148
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In this paper we considered a time series which comes from multiple population. There is not any probability distribution information of the populations. We aim to classify the each samples or fuzzy prediction. Artificial neural network is a very useful tool to solve this problem. Based on the classification a invented probabilistic structure was established to give some fuzzy prediction or control. We showed two applications: one is the classification of Japanese chestnut in a food -making plant; another is for the prediction of stock price fluctuation.
View full abstract
-
Masaaki ISHIKAWA
1995 Volume 1995 Pages
149-154
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
The purpose of this paper is to formulate the stochastic electron beam processing systems. First, the mathematical model of the stochastic electron beam processing system is given. It is shown that the electron beam system leads us to the so-called free boundary problems of the Stefan type. Secondly, by using the change of variables and introducing the appropriate function spaces, the stochastic electron beam processing systems is transformed to the stochastic variational inequality. Finally, the existence and uniqueness properties of the solution of the stochastic variational inequality are proved by using the method of Galerkin.
View full abstract
-
Qun Jin, Nobuko Furugori, Yoshio Sugasawa
1995 Volume 1995 Pages
155-160
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
An aggregate approach of extended stochastic Petri net (ESPN) model and Markov renewal process with some non-regeneration points is applied to conduct performance analysis of an Ethernet-based client-server system. Petri net is used because of its highly visual nature that can give insight into the nature of the modeled system. Markov renewal process is introduced to increase the analytical power of the ESPN model. Modeling and analytical approach used in this paper is different from prevalent methods because of its potential advantage to distinguish the system states between regeneration points and non-regeneration points as well as to incorporate arbitrary time distributions to the ESPN model. The proposed approach may improve efficiency of performance analysis for most practical parallel/distributed systems.
View full abstract
-
Kimio SASAKI, Kohji UCHINO
1995 Volume 1995 Pages
161-166
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
With the aims of exploring humankind's biaural localization ability as well as developing refined devices for environmental recognition in robotics, a practical method for 3D-biaural localization of a stationary random acoustical source in near-field is proposed by newly introducing a rotation of detectors, in addition to the previously proposed method using amplitude attenuation as well as the conventional relative time delay between detected signals. The principle of the proposed 3D-localization and theoretical evaluation of the position estimate deriverd from auto- and cross-spectral analyses of detected signals are presented, together with the results of numerical analysis, illustrating the effectiveness as well as fundamental characteristics of this method under practical circumstances.
View full abstract
-
Satoru GOTO, Masatoshi NAKAMURA, Nishantha NANAYAKKARA, Tatsuhide MATS ...
1995 Volume 1995 Pages
167-171
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
This study is aimed to develop an automatic decision making of the time of washing the polluted insulators in substations. The factors of the automatic decision making are determined by the right state of the information of an expert's decision making. The automatic decision is made based on the index of the linear discriminant function. The proposed method is applied to the actual data of a substation and assured the usefulness. The proposed method accomplishes the role of the human decision making, automatically.
View full abstract
-
Kousuke Kumamaru, Jinglu Hu, Katsuhiro Inoue, Hideaki Ono
1995 Volume 1995 Pages
173-178
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
A crucial point in a model-based fault diagnosis system is its robustness to the influence of unmodelled uncertainty. Extending the framework of FDI system presented in our previous works to the case of presence of unmodelled uncertainty is a practical way to build such robust system. The framework of the robust FDI system includes three principal elements: the
robust identification of model
parameters, the Kullback Discrimination Index(KDI) including the information of unmodelled uncertainty, and the
robust decision making. In this paper, we will focus on the analysis of the KDI including the information of unmodelled uncertainty. Some simulations have been carried out to study the effects of the unmodelled uncertainties.
View full abstract
-
T. Sen Lee
1995 Volume 1995 Pages
179-184
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
Signal models and their statistical properties of a Doppler weather radar which employs clutter filter to suppress ground clutter are derived and validated by real data. These models are used to set signal thresholds imbedded in a number of hazardous weather detection algorithms.
View full abstract
-
Shogo TANAKA
1995 Volume 1995 Pages
185-190
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
In excavating tunnels for subways and power cables, shield tunneling machines are used. The paper presents an automatic detection system which detects anomalous objects to prevent the cutters of the machines from being damaged and also to avoid the construction delay caused by the accident. The detection system utilizes one transmitting and several receiving transducers (both sonic) located on the cutter plane and makes use of the modeling of the electric-sonic- electric transmission system in order to realize a high-accurate detection and characterization of the anomalous objects.
View full abstract
-
Norimasa Asano, Masamichi Ohba, Kageo Akizuki, Hiroshi Inujima
1995 Volume 1995 Pages
191-196
Published: May 05, 1995
Released on J-STAGE: May 28, 2018
JOURNAL
FREE ACCESS
Wavelet transform requires more computation time than Fourier transform. That leads to an obstacle of its engineering application. So the authors propose an efficient calculation method of wavelet transform. Using this method, the authors extract an acoustic signal embedded in noise. The performance of this method is demonstrated via comparing with band-pass filter approach.
View full abstract
