A joint method is proposed for estimating the systems state and identifying the unknown (stepwise) exogenous input to the stochastic dynamical systems by introducing the idea of pseudomeasurement. Both state estimation and parameter identification are processed by the Kalman filter, incorporating the pseudomeasurements. An application of the method to the change detection is also discussed briefly. The proposed method is tested through simulation studies.

View full abstract

We consider the estimation problem of stochastic volatility from stock data. The estimation of the volatility process of the Hull-White model is not in the usual frame work of the filtering theory. Discretizing the continuous Hull-White model to the discrete-time one, we can derive the exact volatility filter and realize this filter with the aid of particle filter algorithm. In this paper, we derive the optimal importance function and construct the particle filter algorithm for the discrete-time Hull-White model with jump processes. The parameters contained in system model are also estimated by constructing the augmented state for the volatility and parameters.

View full abstract

It is well known that least-squares (LS) method gives biased parameter estimates when the input and output measurements are corrupted by white noise. One possible approach for solving this bias problem is the bias-compnesation based method such as the bias-compensated least-squares (BCLS) method. In this paper, a new bias-compensation based method is proposed for identification of noisy input-output system. The proposed method is based on compensation of asymptotic bias on the instrumental variables type (IV-type) estimates by making use of noise variances estimates. In order to obtain the noise variances estimates, an overdetermined system of equations is introduced, and the noise variances estimation algorithm is derived by solving this overdetermined system of equations. From the combination of the parameter estimation algorithm and the noise variances estimation algorithm, the proposed bias-compensated instrumental variables type (BCIV-type) method can be established. The resuluts of a simulated exmple indicate that the proposed algorithm provides good estimates.

View full abstract

It is well known that the least-squares (LS) method gives biased parameter estimates of transfer functions when the output measurement is corrupted by white noise. One possible approach for solving this bias problem is the total-least-squares (TLS) method. Davila proposed a Recursive TLS (RTLS) algorithm based on minimization of the generalized Rayleigh quotient, but it has large computational complexity.In this paper, a new RTLS algorithm saving computational complexity is proposed. The simulation results of numerical examples illustrate the proposed method provides a good performance.

View full abstract

A method to measure the diameter of deformed bars in reinforced concrete structures using an electromagnetic wave radar is proposed. This method focuses on the measurement of the rib's pitch of the bar and then obtains the diameter of the bar from the standard relationship between the rib's pitch and the diameter. For the measurement of the pitch, variation of the propagation times of the electromagnetic waves to and from the bar along its length is first obtained and then Kalman filter and maximum likelihood method is applied to the dynamic model of the propagation time variation. Lastly, the frequency analysis of the proposed method is proved to be more reliable than the well-known FFT by experiment.

View full abstract

We have proposed an interpretation of MOESP types of subspace algorithms by using the Schur complement (SC) of the data product moment and proposed a unified framework for the subspace-based identification. Here we consider the introduction of exponential forgetting factor which windows the data matrices to apply the algorithms to the slowly time-varying system. The data matrix is windowed to reduce the influence of old data, which the forgetting factor or the sliding window can be used. Here it will show that the window weighting can also be reformed as the weighting of the data product moment and the proposed unified framework still kept consequently. Furthermore, this paper shows the proposed unified approach for the subspace identification will be reviewed at the point of view from the results of the Subspace-based identification using instrumental variables (SIV) approach by Gustafsson and discuss on the equivalent with the SIV and the proposed framework.

View full abstract

This paper is concerned with the identification of a discrete-time Wiener model, composed of a linear time-invariant (LTI) system followed by a static monotonous nonlinearity. Introducing a set of basis functions for the inverse nonlinearity, we define a criterion for the Wiener model identification according to the generalized error structure. We then develop a method of alternately identifying system matrices of the LTI system by a subspace method and the inverse nonlinearity by the least-squares method. Some numerical results are included.

View full abstract

In this paper, we present a prefiltering method for linear multi-input multi-output (MIMO) system identification. It is assumed that the system representation is the combined system with deterministic system and stochastic system based on orthogonal decomposition of the output process. The stochastic component can be defined clearly by the orthogonal decomposition of the output process based on the conception of Wold's decomposition. We consider removing the stochastic component from the output process. If it can be removed completely, by employing deterministic subspace methods, the parametrization problem included in the state-space model identification is completely bypassed. Also, if the stochastic component can be estimated precisely, this namely means a prefiltering for the system identification. For implementing this purpose, we employ LQ decomposition, also consider that the last block row of L-matrix is extracted. Also, the effectivities are shown in numerical experiments.

View full abstract

We investigate bifurcation phenomena between slow and fast convergences of synchronization errors arising in the proposed synchronization system consisting of two identical nonlinear dynamical systems linked by a common noisy input only. The numerical continuation of the saddle-node bifurcation set of the primary resonance of moments provides an effective identifier of the slow convergence of the synchronization errors.

View full abstract

It is well known that a self-organization plays an important role in various fields of engineering. The self-organization can create many kinds of spatio-temporal patterns such as spiral and target patterns in the BZ(Belousov-Zhabotinsky) reaction and a phase separation pattern in a high-polymer substance. In this paper, we focus attention on bacterial colony patterns among many spatio-temporal patterns created by the self-organization and study a mathematical modeling of the bacterial colony formation with chemotaxis such as Salmonella typhimurium (S. typhimurium) and Escherichia coli (E. coli). Since we cannot analyze the bacterial colony formation under the existence of fluctuations in the concentration of chemical substances and the population density of bacteria by the conventional deterministic model, we propose a stochastic model of chemotactic bacterial colony formations and study the influence of fluctuations on the bacterial colony formations by numerical simulations.

View full abstract

Epidemics such as the bird flue is still a menace to mankind. Especially, rabies is a viral zoonotic disease that causes acute encephalitis in mammals: more than 50000 people a year have died by rabies. Analyses of geographic spread of such epidemics are essential to contrive a plan to prevent it from going around. In this paper, noting that the infection rate of rabies contains the random fluctuations caused by the change of the environmental situation and considering the threshold from the biological reason, the stochastic model of rabies spreading with the infection rate of a saturation type is proposed. Using the proposed stochastic model, behaviors of rabies spreading and its spreading speed are studied by numerical simulations. The influence of the random fluctuations in the infection rate and the threshold on the rabies spreading is also analyzed.

View full abstract

We are concerned with a problem of the optimal selection of the gain matrix, over a given time interval, of a linear observation for the Kalman filter. The innovations process included in the Kalman filter has the same structure as the model of a set of parallel transmission channels with the optimal output feedback. In the linear coding problem for this set of channels, it is well-known that the optimal output feedback which minimizes the power of the encoded signal is given by the least-squares estimate of the linear term and that the channel output then becomes the innovations process. By applying a solution of the optimal transmission problem for this model, we obtain, at any time points, a set of the gains which maximize the mutual information between the observation and the signal under a constraint on the power of the innovations process. Finally, the optimal selection of the gain to minimize the estimation error is done by an optimization which is local in time.

View full abstract

Nonlinear systems described by the simple mathematical model often exhibit extremely complicated behavior called chaos. In this paper, chaotic behavior, exhibited by the one-dimensional difference equation, is discussed. Methods for constructing chaotic systems with the invariant density, which is piece-wise uniform, are demonstrated.

View full abstract

The chaos synchronization system used for secrecy communication is considered. The compound system which consists of subsystems is chaos-synchronized. In order to increase the secrecy of models, state of synchronization section is constituted by fuzzy model. Subsystems with the same chaotic dynamics are asynchronous at first. In consideration of the non-linear feedback control is applied to the chaos synchronization control. This control input can be made small by use of the unstable periodic regions. It is important that the mathematical model of the chaos system is not known. However the model known well is used. Using the state of the chaos synchronization section, bifurcation parameter is changed and chaos modulations are performed. The chaos state of the encryption section is used for encryption of information signals. The chaos dynamics of the encryption section is generated using chaos neuron dynamics.

View full abstract

In this paper, based on a stationary forward rate model, a principal component analysis of real market data of forward rates is presented. We focus on the study of consistency of a forward rate model by comparing the sample average with the theoretical average implied by the arbitrage-free hypothesis.

View full abstract

The aim of this research is to give a brief survey of framework to evaluate/quantize the transparency of a firm. We present a modification of Merton's model for corporate debt, where we assume that the process of the firm value is not observable but is strongly correlated with the sum of prices of stock and bonds which is observable and tradable. By a filtering argument, an explicit credit spread formula is obtained. Using the formula, ”market implication” of transparency of a firm can be quantified. A numerical example is presented to illustrate the procedure.

View full abstract

We consider jump diffusion processes with compound Poisson process whose jump range has normal distribution or double exponential distribution and also their Bernoulli approximations. In this paper, we shall estimate the parameters of these models by using MCMC-based Bayes formula. As an application, we shall make model selection with respect to Nikkei financial data in virtue of the EIC-criterion.

View full abstract

SLAM is a hot topic in robotics community. It uses range sensors and aquires the distances to various directions as the sensor moves and changes its direction, so that it can acquire the environmental landscape and estimate the sensor's position/angle simultaneously. In this paper, we will explain the detail of FastSLAM by Montemerlo, and propose a modification of the algorithm.

View full abstract

The author generalizes the cloth simulation methods about 3DCG animation based on the mathematically rigorous formulation. The motion of the general dimensional surface (manifold) is simulated by the Newtonian equation which is derived for the continuum surface model ( not for the discrete approximation model ) based on the variational principles. This Newtonian equation become a feedback system. If suitable stochastic process is set as the goal trajectory of the feedback system, we can generate virtual interesting 3DCG animations of the surface motions.

View full abstract

It is supposed that the elements in circulation of rumors mainly consist of importance and vagueness in psychology and marketing fields. In this paper, the influences of vagueness included in circulation of rumors are modeled and analyzed. Rumors including vagueness element are modeled by local influence in agent-based simulations and agents with vague characters are represented by probability. Simulation studies show that the agents with vagueness perform with some randomness and that the action results of these agents have some regularity.

View full abstract

Hodgkin-Huxley model for nervous signal transmission is known as multi-variable strongly non-linear model. We consider Hodgkin Huxley model as the stochastic process to estimate the state variables and the model parameters from the measurement data with noise. We discretize the model using the local linearization method and apply Kalman filter to this model. So that the trend input current and the model parameters are estimated simultaneously utilizing the maximum likelihood method.

View full abstract

In this paper, we propose a modified Gaussian Sum filtering method and apply it to land-vehicle INS(Inertial Navigation System)/GPS(Global Positioning System) integrated navigation as well as In-Motion Alignment systems. Then we show the experimental results of INS/GPS integrated system by using simulated data, with assuming the various error sources such as the errors of GPS measurement, INS sensors and various running trajectories.

View full abstract

In this paper, we present an ionospheric estimation method based on GNSS Regression models (abbreviated by GR models) and show to derive the ionospheric delay (or advance) local model in the sky over Japan using the Gps Earth Observation NETwork (GEONET) data provided from the Geographical Survey Institute (GSI) of Japan. By showing the GR models for GNSS observables in the case of known positions (in the reference stations), we derive a method to estimate the ionospheric delays (or advances) as well as the so-called integer ambiguities contained in L1 and L2 band carrier-phase data at the reference stations. Then we emphasize the importance to estimate integer ambiguities of their own receivers at the reference stations.

View full abstract

Kinematic Global Positioning System (GPS) is a positioning method that uses carrier phase data to output a highly precise position. In kinematic GPS, the position accuracy might be degraded significantly by cycle slip, i.e., a sudden jump in the carrier phase observation by an integer number of cycles. Methods for detecting cycle slip include a method that uses statistical tests of carrier phase innovations. This method has the merit that the noise of the innovation is small and does not affect the detection performance; however, it has the disadvantage that the movement of the land vehicle degrades the detection performance and causes either undetection and mis-detection. In this study, a dynamic model that includes jerk (i.e., the rate of change of acceleration) is proposed, and based on this model, a cycle slip detection method that corresponds to the movement of a land vehicle is then developed. This method precisely predicts the position of a land vehicle, even when it accelerates and decelerates, and improves the performance of the cycle slip detection. Moreover, the performance of this method is experimentally evaluated using observation data collected with a car.

View full abstract

This paper introduces a new application of a passive radar system using a GPS satellite.Radar is a system which uses reflection signals in order to detect and pursue targets. Passive radar makes use of reflection signals such as commercial broadcasts and communications signals to detect targets and to estimate parameters. However, passive radar systems are not always designed to detect and process the echoes from targets illuminated by transmitters. The system would lose the availability if the receivers couldn't detect these signals.We are concerned with the estimation method for a target position by one echo from a target illuminated by a GPS satellite. When we could receive a reflection signal from a GPS satellite we can use constraint conditions which are the relative delay between a direct signal and reflection signals and the rate of change of phase differences. We can obtain the reflection position by solving simultaneous equations. The equations are presented as the equation relating the relative delay between a direct signal and reflection signals, and the equation relating differential of relative delay.In this paper, we describe proposed method under the assumption that the target is a vertical plane. In addition, we apply proposed method to observation data and evaluate the availability of proposed method.

View full abstract

An algorithm for music transcription system that listens to sounds and displays notes of the corresponding tones has been proposed. The algorithm features a two-stage processing that observes note length at first, and then, pitch names. The former stage effectively detects break points caused by starting of next note or frequency hop at pitch name changing with dyadic wavelet transforms. The latter stage analyzes component of each pitch name on the interval of the adjoining break points with Gabor wavelet. The algorithm is applied adaptive state estimation technique to cope with tones including considerable harmonics.

View full abstract

In Brain Computer Interface research, it is common to use linear method to analyze human's brain signals. For instance, AR modeling method may be used in the process of extracting features from the EEG (Electroencephalogram) signals. Some studies used statistical pattern recognition method based on AR model to discriminate the EEG signals (for example see [1] and [5]). On the other hand, the studies of Freeman (1991) suggested that Chaos underlies the ability of the brain to respond flexibly to the outside world. Therefore, some nonlinear modeling method may be used to analyze brain signals.In this study, we applied a non-linear modeling method based on quasi-AR model to extract features from EEG signals recorded during right hand, left hand, and right foot motor imagery.

View full abstract

The paper proposed pattern recognition method based on ICA (Independent Component Analysis) for EEG (electroencephalogram) signals during right and left hand motor imagery. ICA can separate unknown source signals from their mixture signals if they are statistically independent. Some effective features for pattern recognition appear in the separated signals with scaling adjuster. In this paper, we try to discriminate EEG signals during left and right hand motor imagery based on ICA. As a result, we obtained the prospect concerning the construction of BCI system with the reliable pattern recognition method for discrimination of motor imagery.

View full abstract

We have developed the real-time speech visualization system called “KanNon” [1], which supports speech communication of hearing-impaired people. The KanNon system presents informations of the speech such as loudness, pitch, sound spectrogram and characters by speech recognition system in real-time.Furthermore we have attempted to develop the vowels recognition system which recognizes phonemes of speeches. For this purpose, in this paper, we develop a method for distinction of phoneme sections and we propose two discrimination indices of dividing phonemes. Finally, we show experimental results of discrimination of phoneme sections as well as the recognition of phonemes.

View full abstract

A Radial Basis Function (RBF) Group Method of Data Handling (GMDH)-type neural network algorithm proposed in this paper is applied to the medical image recognition of abdominal X-ray CT images. The optimum neural network architecture for the medical image recognition is automatically organized using RBF GMDH-type neural network algorithm and the regions of abdominal organs such as the liver, stomach and spleen are recognized and extracted accurately.

View full abstract

In this paper, we consider the problem of motion recovery in the theory of machine vision under perspective stereo vision. The object moving in space may be occluded from the cameras for some observation time intervals. First we show that the so-called motion parameters are identifiable generically in a normal stereo settings, where the object is visible from the two cameras simultaneously at least in a nonzero time interval. Then by developing an experimental stereo vision system and using extended Kalman filter for identification, we examine the validity experimentally.

View full abstract

Scene analysis is reformulated within satellite-roadway-vehicle network. For integrating multi-viewpoint imagery, randomness underlying natural complexity is extracted and associated with fractal version of maneuvering affordance. Scale shift due to perspective projection is evaluated for separating the image of not-yet-identified object from the maneuvering affordance.

View full abstract

In this paper, we consider a blind image deconvolution method for linear motion blurred images. Point spread function (PSF) of the linear motion blur has two parameters, i.e., motion length and motion direction. The parameters are estimated using the modified discrete Radon transform and cepstral analysis. The estimated PSF parameters are then used in the image deconvolution process. To restore the blurred images, we first model the 2-D degradation process into 1-D form along the blurring path. Therefore, the new 1-D blurred image model is constructed for each row of the blurred image in the 2-D degradation process. The corresponding PSF matrix is then inverted to get the true image. Finally, experimental results show our proposed idea.

View full abstract

In this study, three dimensional medical images of the lungs, pulmonary vessels and bronchial trees are recognized and extracted by artificial neural networks. Two neural networks trained using the back propagation algorithm are applied to medical image recognition. First, the neural network recognizes the outline of the lungs and outputs the lung regions and 3-dimensional images of the lungs are displayed. Then, second neural network recognizes outline of the pulmonary vessels and bronchial trees and outputs the regions of them and 3-dimensional images are displayed. These pulmonary vessels and bronchial trees are divided into five lobes by 3-dimensional region growing method. These image processing proposed in this study can be easily applied to another medical image such as MRI image.

View full abstract

In this study, we applied the radial basis function (RBF) neural network to recognition of 3-dimensional medical images of the head. This neural network has the three layered architecture that is constructed with the input layer, the hidden layer and the output layer. The regions of the brain, the skull and the blood vessel, were recognized and extracted by the RBF neural networks. These image processing were carried out for all slices of multi-detector row computed tomography (MDCT) images and 3-dimensional images of the brain, the skull and the blood vessel regions of the head were displayed with volume rendering software. It is shown that the RBF neural network is useful for the 3-dimensional medical image recognition of the many kinds of organs such as the brain, the skill and the blood vessels.

View full abstract

This fundamental research is concerned with the crack detection of a concrete pillar combined with a concrete hexahedron, which is a model of the base pile of a bridge pier. Stationary waves are generated in the pillar and the hexahedron by hitting the head of the concrete hexahedron with a hammer. An acceleration pickup on the hexahedron observes the stationary waves generated inside the concrete pillar through the concrete hexahedron. The proposed method measures not only the length of the pillar but also the depth of the crack in the pillar by modeling the sensor output as an output of a linear dynamic system with unknown parameters and applying a maximum likelihood method.

View full abstract

The equations of Brillouin distributed sensing using an optical fiber cannot be solved analytically due to nonlinear interactions between light waves and phonons. In a previous paper, an asymptotic solution was obtained using a perturbation method. However, since the solution does not depend on the length of the optical fiber nor the distance to the measurement positions, it is insufficient for performance analysis of long-range measurements. Moreover, the higher order terms in the perturbation expansion include secular terms. In this paper, a multiscale expansion method in the singular perturbation theory is used to solve the nonlinear equations for the performance analysis of long-range measurements. An asymptotic solution is obtained that contains the pump depletion effect.

View full abstract

Numerical computations of large-scale linear programming problems often contain instability owing to modelling errors and accumulation of round-off errors. So the question arises whether the effect of these errors increase or decrease with the size of problems. We show that the effect of the random errors in the original data to the optimum usually has a trend to decrease as the number of variables increases. This result is stated in terms of the law of large numbers in probability theory.

View full abstract

It seems that hi-tech pollution and many other difficult modem problems come up from slighting or reducing many kinds of complicated relationship among various environmental factors including even ethical or cultural faces to a secondary position and giving priority to only utility beyond trueness over any other everything. In this paper, to solve these problems, we first pay attention to the criterion of “Relationism-First” that once after investigating at the first stage of study many environmental factors and the mutual correlations being latent among them as possible at the same time and in the same ring of study (for trueness), then our specified interesting cases for engineering application should be considered (for effectiveness). In the previous paper, by taking care of light and shade( that is, utility and risk) as a method for mutual intersubjective analysis, an extended correlation analysis for only two environmental factors has been applied on trial. In this paper, another extended correlation analysis available to more actual fluctuation limited within the finite amplitude interval is newly introduced. Furthermore, as a principle experiment for the proposed method, by applying it to the contrasted two environmental factors: magnetic field (related to risk) and sound (related to utility) around VDT and cellular phone before and after attachment of Tecno AO (active bio-controller as some magnetic oscillator), the proposed method is experimentally confirmed, too.

View full abstract

In this paper, coal selection for a coal fired boiler in a thermal power station is investigated. Available data of actual used coal types are too small compared with properties of coal type. Classification of coal properties is, hence, introduced and the coal properties are consolidated into several categories through the factor analysis. Since applicability of coal is evaluated by using the multiple regression analysis where the explanatory variables are the factor scores for respective categories. The proposed selection method is applied to a boiler in a thermal power plant.

View full abstract