This paper presents a method of identifying closed-loop Hammerstein-Wiener systems, where a linear plant in the forward loop is stable or semi-stable. In the framework of the direct approach [17], an output error (OE) model is derived with the help of basis function expansions of nonlinearities. First an approximate linear state space model of the forward loop is identified using a subspace identification method. Then, initialized by the identified linear model, the mean square of the output error is optimized by using a gradient-based search. Some numerical results are included to show the applicability of the proposed method.

View full abstract

This paper investigates the problem of identifying errors-in-variables (EIV) models, where the both input and output measurements are corrupted by white noise, and addresses a new method to solve the problem. The identification problem of EIV models with unknown noise variances has been extensively studied and several methods have been proposed. To be further developed in terms of estimation accuracy, a generalized eigenvector method with no requirement of a priori knowledge about the noise variances is proposed by using the biased weighted least squares estimator. The proposed generalized eigenvalue problem can be derived by removing only the bias induced by the output noise, and thus the system parameter can be obtained as the generalized eigenvector without requiring the use of iterative identification procedure. Moreover, the bias compensation principle based algorithm, which is suitable for on-line implementation, is derived to solve the proposed generalized eigenvalue problem. The results of simulated examples indicate that the proposed approach provides good parameter estimates.

View full abstract

Theoretical properties of the ensemble Kalman filter (EnKF) and smoother (EnKS) for nonlinear non-Gaussian state-space models are provided. I first show that the EnKF is an approximation algorithm for the linear optimal filter. I next define the moment-matching linearization (MML) and from this viewpoint, I prove that the EnKF is more accurate than the Gaussian filter for both discrete-time and continuous-time cases. I also define the linear optimal smoother (LOS) using the MML. Then, the EnKS algorithm that approximates the LOS is derived and its difference from the existing EnKS is clarified.

View full abstract

We consider discrete-time nonlinear stochastic systems and investigate useful recursive procedures for estimating the states of these systems. Since mathematical models derived by engineers are not free from modeling errors in practice, it is an important task to investigate state estimation methods which work well for stochastic systems with unknown disturbances. So, in this paper, we deal with nonlinear stochastic systems with unknown disturbances and investigate state estimation methods which satisfy disturbance decoupling property and can be applicable to nonlinear systems. Numerical simulations are given to show usefulness of the proposed method over the standard extended Kalman filter.

View full abstract

While H_∞ filtering theory for stochastic continuous-time systems driven by Poisson processes has been presented by B. Song et al.([21]), H_∞ smoothing theory for the systems and the relationship between H_∞ filtering and smoothing have not been yet fully investigated. In this paper, we study the H∞ state estimation (filtering and smoothing) problems for a class of linear continuous-time systems driven by Wiener and Poisson processes on the finite time interval. In order to derive H_∞ state estimators, we adopt unified stochastic variational approach as [15, 17] which has not been found in previous work by any other researchers.

View full abstract

In this paper, the author investigates numerically a class of two-dimensional fuzzy random sets(abbreviated as FRSs), which is one case of those proposed by the author[1] as models of vague capricious perceptions of random phenomena. First, the basic results concerned with two-dimensional FRSs as vague capricious perceptions of random phenomena are reviewed. Expectations of two-dimensional FRSs, their estimators, and distances between expectations and their estimators are also reviewed briefly. Finally, estimates for expectations of two-dimensional FRSs and their estimation errors are numerically examined by simulation studies, when their level sets are given by disk-form ones at each levels.

View full abstract

This paper presents a novel learning automaton, β- type, which consists of 2-state Bayesian estimators. The β-type learning automaton is presently among the fastest learning automata known, which was proposed in our earlier works. However, compared with the β- type learning automaton and the conventional learning automata, the β-type learning automaton deteriorates from the viewpoint of memory usage and other resources, for example, since computational and energy resources of some applications are limited, such as the wireless sensor networks, reducing memory footprint and performance optimization are very important issues. So, in this study, we propose the β-type learning automaton with minimum resources, 2-state Bayesian estimators. Then, the efficiency of proposed β-type learning automaton is shown through several simulation results under some random environments.

View full abstract

A stochastic averaging consensus algorithm is considered for a multi-agent system over a noisy undirected network with multi-input/multi-output (MIMO) linear symmetric agents. The convergence of the algorithm is investigated, which gives an explicit relation between the number of iterations and the closeness of the agreement, i.e., a stopping rule. The result is illustrated through a numerical example.

View full abstract

In this paper, we study optimal LQ tracking control problems for linear discrete-time hybrid systems. We adopt an approach to optimize system states and mode distributions simultaneously. We consider optimization problems for averaged systems and averaged performance indices throughout mode distributions.

View full abstract

In this paper, we study optimal tracking control problems with preview for linear continuous-time hybrid systems. We adopt an approach to optimize system states and mode distributions simultaneously. We consider optimization problems for averaged systems and averaged performance indices throughout mode distributions. On the contrary to the previous work [12], in this paper, we consider to utilize any preview information of reference signals.

View full abstract

In this research, the linear exponential quadratic Gaussian (LEQG) control proposed by Jacobson[1] is introduced into a control problem of Two-Wheeled Inverted Pendulum (TWIP) subject to random disturbances. The purpose of this research is to develop one of suitable controller for the TWIP subject to large random disturbances from the road environment. The mathematical model of the TWIP with random disturbance should be treated as a nonlinear system with random disturbances. The random disturbances are assumed to act at the bottom of the right and left wheels from the surface of the rough road. In this research, such random disturbances are modeled by the white Gaussian noises, mathematically. In order to deal with the nonlinearity of the mathematical model of the TWIP, the gain scheduling control technique is introduced for the LEQG controller. The performance of the controller for the stochastic TWIP system is demonstrated by performing several numerical simulation studies in a couple of random disturbance conditions, i.e. a large variance of random disturbance and a small variance of one. Furthermore, three types of reference trajectories were tried to trace in the numerical simulations.

View full abstract

In industrial control process, saving energy and produce high quality products is demanded. A cerebellar model articulation controller (CMAC) based performance-driven (PD) PID controller is proposed to tune its gains for both transient and steady state, so that the above requirement achieves. For a conventional CMAC, a high learning accuracy is obtained by sacrificing its generalization ability. Therefore, a hierarchical clustering CMAC (HC-CMAC) is proposed. Compare with conventional CMAC, the proposed HC-CMAC enables each weight table different number of labels. In this case, the feature of CMAC such as partial learning ability and generalization ability are remained, simultaneously, the weight tables with less labels increase generalization ability and the weight tables with more labels increase learning accuracy. In some industrial control processes, if a machine participates not only one production process, a few reference signals may be set to the machine due to different conditions. If a controller tunes its gains for different situations, it costs a large calculation time, hence, generalization ability for similar reference signals of a controller is important. Based on above considerations, a HC-CAMC PD PID controller is proposed and some simulation results to show the effectiveness of proposed method are demonstrated in the paper.

View full abstract

This paper presents the design of a data-driven adaptive control system based on output feedback strictly passive (OFSP) with a feedforward input for nonlinear systems in discrete-time domain. It is well recognized that an adaptive control system based on OFSP conditions can achieve asymptotic stability via a static output feedback. However, these conditions are very restrictive on most realistic systems. Therefore, a parallel feedforward compensator (PFC) is introduced to alleviate such restrictions, but remains a steady state error which is removed by utilizing the feedforward input. Besides, the data-driven approach is one of effective control strategies for nonlinear systems in the process control, and the control parameters can be updated rapidly at each equilibrium point. However, there is no research work that had been done in terms of applying the data-driven approach in a adaptive control system based on OFSP. Therefore, once the robust stability of the adaptive control system based on OFSP can be guaranteed, the adaptive gains can be optimized by the data-driven approach such that the output performance is able to be improved. The proposed scheme is verified through an numerical example, the results of which demonstrate the effectiveness.

View full abstract

This paper considers the distributed power management of a microgrid consisting of consumers, PV generators, and batteries. The objectives of the microgrid management are (1) the supply-demand balancing, (2) the low dependency on the main grid based on the local-production-local-consumption policy, and (3) the reservation of battery charges to prepare for faults or islanded operations. For the microgrid management, it is important to explicitly take account of the prediction of PV power generation because it heavily depends on weather conditions. By extending the result due to Takeda and Takaba (Trans. SICE, vol. 54, no. 2, 2018), we propose a distributed stochastic model predictive control method using the predicted PV powers. In the proposed method, each battery management system (BMS) attached to a battery determines the power flows from and into the battery in a distributed manner so that the above objectives are achieved in the local area around the BMS. We verify the effectiveness of the proposed method by numerical simulations.

View full abstract

In this research, a water level model for after condenser in a spray ash desalination system is proposed. The model is constructed by introducing a white Gaussian process to an existing model in the conventional research, where the mean and the variance of the white Gaussian process are determined by using experimental data. In order to verify the behavior of the proposed model, numerical simulations were conducted.

View full abstract

For the shape sensing of long-range cables, we study a shape sensing method that utilizes the strain data from optical fibers installed around the cable. We estimate the shape by solving the equation of a moving frame whose coefficient contains the curvature and twist rates obtained by the strain data. The estimation error process is formulated by a stochastic differential equation, and the error variance is obtained by solving the variance equation. The obtained theoretical error variance indicates that the shape estimation error becomes maximum when the curve is straight, and the general curve's error is smaller than that for the straight line. We validated these results by numerical simulation.

View full abstract

In this paper, we propose a continuous-time (CT) loop filter for minimizing the quantization noise in a Delta-Sigma (ΔΣ) modulator. We use the impulse-invariant transformation to obtain a CT loop filter from a discrete-time (DT) loop filter. We consider a non-ideal CT ΔΣ modulator which is affected by the excess loop delay (ELD). The ELD is known to affect the stability and noise shaping performance of the CT ΔΣ modulator. Our objective is to design the CT modulator whose stability is not affected by the presence of ELD. A design example for a lowpass CT ΔΣ modulator based radio frequency (RF) receiver is provided for simulations. We also compare the mean squared error (MSE) performance of our modulator with the conventional design.

View full abstract

The author previously applied extended Kalman filter for estimating the posture of depth sensor attached to a walking person or a mobile vehicle, where the posture of the sensor was included in the state vector. However, it cannot adapt to situations where the initial value of extended Kalman filter was not adequate. In various textbooks of Kalman filter, the initial value of the filter algorithm has not been discussed extensively. However, it is very important, especially for the model when the domain of the state vector has several groups and different starting point leads to different results. In this paper, the initial value setting of the algorithm is proposed. It is simple, but it has proven to work well and has a good property to adapt to various situations. By defining the reset conditions differently, it is shown that the system can detect different surfaces.

View full abstract

In this paper, we propose hybrid methods of Tone-Injection (TI) and Clipping-and-Filtering (CF) for suppressing the non-linear distortion of OFDM. TI utilizes the genetic algorithm to search for an effective solution. The proposed methods are evaluated in terms of Peak-to-Average-Power Ratio (PAPR), Bit-Error-Rate (BER), and out-of-band radiation. The effectiveness of the proposed methods is demonstrated in the simulation.

View full abstract

In this paper, we investigate several methods to resolve integer ambiguities in Global Navigation Satellite Systems (GNSS) carrier phase positioning applied by Kalman filter. Especially, we focus on integration methods of Least squares AMBiguity Decorrelation Adjustment (LAMBDA) [1, 2] method and Kalman filter. They are compared from theoretical and experimental points of view, and their characteristics and performance are investigated. We conduct experiments using real receiver data, and evaluate the integer ambiguity resolution performance. Based on the results, we show a guideline to resolve ambiguities with Kalman filter.

View full abstract

In this paper, we present a new relative positioning algorithm based on GNSS Regression measurement models (GR models) by using multiple antennas based on alternative use of the positions of reference receivers. We show the algorithms of estimating all unknown antenna positions by applying the double difference (DD). Then we derive updating algorithms by using true positions of the reference receivers. Finally, we carried out the experiments using real GNSS data and show the positioning results for the the proposed method as well as the conventional method.

View full abstract

We analyze the time series of exchange rates and gold price, USD/JPY, EUR/USD, USD/CHF, GBP/USD and XAU/USD, in the period before and after the bankruptcy of Lehman Brothers by using relative noise contribution. Relative noise contribution detects the influences among multiple time series. As a result, we found the exchange rates and gold price are basically driven by their own influence. However, when the affairs happened, they sometimes were driven by other rates or price. In addition, the reaction to the affairs is changed before and after the bankruptcy of Lehman Brothers. These may result in the difference of trader's ideas between before and after of the bankruptcy. This result implies that the framework has a potential to find some relationship among the markets and participant's behavior.

View full abstract

At present, the unprecedented cholera outbreak occurs in Yemen and the strategization to control cholera transmission becomes imperative. Hence, infection prevention and control of epidemic are cited as one of the important social problems. In the vector-borne diseases such as malaria and dengue fever, there exists time delay caused by an incubation period in the virus development in the vectors on the transmission of disease. It should be noted that there is possibility of getting reinfected in the infectious disease such as malaria. Moreover, in the realistic spread of the infectious disease, environmental change and individual difference cause some kinds of random fluctuations in the infection, the recovery rates and the vaccination effect. Taking these facts into consideration, we propose two types of the stochastic delayed infectious models with reinfection. Since the spread of infection has reference to the stability of the disease-free steady state (DFS) of the stochastic infectious models, we analyze the stability of the DFS by using the stochastic Lyapunov theorem.

View full abstract

Biological signals recorded in the nervous system contains rich information on the motions which have been applied for various prosthetic applications. It is essential to extract appropriate feature quantities from the signals to estimate the intention of the subject for voluntary movements. We have diverted the line spectrum pair (LSP) representation, which had been originally developed for speech coding, as a feature quantity for the biological signals directly recorded from the living body. In this paper, 2 examples of utilizing LSP for different biological signals are demonstrated. While the human myoelectric (EMG) signals have shown the ability of the LSP representation to estimate the hand shapes, the neural local field potentials (LFP) of a rat were used to detect its intention to walk.

View full abstract

In recent years, more and more people have been getting married at a later age, and this trend is only going to increase in the near future. As a result, many infertility treatments have been researched upon and developed. One such treatment is in-vitro fertilization, wherein a follicle is directly absorbed from the ovary of the woman and impregnated with a sperm cell under a microscope. Of the total number of follicles in a woman's body, 20-30% are usually observed to contain vacuoles, which do not stimulate vitelline growth. As such, in most cases, one cannot judge whether a follicle has an ovum or a vacuole unless we examine it under a microscope. This study proposes a method capable of distinguishing follicles on the basis of local movement.

View full abstract

This paper discusses “attention” in cognitive function by analyzing hemodynamics and potentials while performing modified paced visually serial addition test (mPVSAT). The test needs higher-order attention i.e. selective attention, which is evaluated by event related hemodynamics and event related potentials in our study.

View full abstract

Recently, many researchers have been studying the Brain Computer Interface (BCI). As one of its input signals, the EEG activity when users do motor imagery is analyzed in frequency domain to be detected ERD or ERS. Currently, however, these analysis methods require users' long-term training to detect them and get higher EEG recognition accuracy[1]. In this paper, to avoid these problems, we consider the analysis method for the EEG activity in time domain when subjects do 4 limbs motor imagery. Further, by the feature extraction relating to Movement-Related Cortical Potential(MRCP), we assess the relationship between features and recognition accuracy.

View full abstract

This research attempts the eye movement detection for the communication of people with disabilities by employing the electrooculographic (EOG) signals. Eye movements for the left and the right are detected by capturing the peak which appears in the waveform. Then, the judgment for left or right eye movement is conducted based on four parameters about the detected peak. To confirm the effectiveness of the eye movement detection method considered in this research, experiments were performed.

View full abstract

In this research, a nurse call system for patients with movement disorders by eye movements extraction based on image processing is developed. In the system of the previous research, to extract the human intention, three kinds of iris positions were detected based on image processing. However, the system has a problem that the detection of eye movement with blink did not work well. In this research, a communication system using eye movement extraction based on image processing with blink detection function is developed. The blink contained in images from camera is detected by using the information about the differences of consecutive images. The accuracy of the proposed method was evaluated through experiments.

View full abstract

We have been developing the pronunciation training system, in which visual instruction about the correct articulation are shown. These instructions are produced based on the formants of speech signal and show the shape of articulator to the learners. They work effectively, but more comprehensible instruction on how to move the articulator is required. We think that the facial muscles that move the articulator are useful to such the instruction. In this report, we attempt to construct lips movement model representing the relationship between the facial muscles activity and the lips movement and discuss the application of this model to pronunciation training.

View full abstract

Japanese elementary and junior high schools have traditionally provided calligraphy classes to students. However, as the number of teachers decreases, so too does the number of calligraphy classes, making it difficult for students to improve their skills due to an inadequate amount of class time. Hence, a system capable of automatically evaluating students' progress on writing characters would enable them to independently bolster their calligraphy skills. Many methods of teaching calligraphy have already been proposed [1]-[3]. However, these require special devices which prove difficult to use. Therefore, we suggest an offline system is best as it is easy to use. In this study, ‘offline' means easily digitized visualization with the use of a smartphone or scanner. This study's participants were elementary school students capable of writing correctly.

View full abstract