• 대한전기학회논문지
• /
• 제34권5호
• /
• pp.173-178
• /
• 1985

The control of a linear system with random coefficients is discussed here. The cost function is of a quadratic form and the random coefficients are assumed to be completely observable by the controller. Stochastic Process involved in the problem by the controller. Stochastic Process involved in the problem formulation is presented to be the unique strong solution to the corresponding stochastic differential equations. Condition for the optimal control is represented through the existence of solution to a Cauchy problem for the given nonlinear partial differential equation. The optimal control is shown to be a linear function of the states and a nonlinear function of random parameters.

• 품질경영학회지
• /
• 제16권2호
• /
• pp.1-9
• /
• 1988

This paper considers a machine maintenance problem. The machine's condition is partially known by observing the machine's output products. This problem is formulated as an infinite horizon partially observable Markov decison process to find an optimal maintenance policy. However, even though the optimal policy of the model exists, finding the optimal policy is very time consuming. Thus, the intends of this study is to find ${\varepsilon}-optimal$ stationary policy minimizing the expected discounted total cost of the system, ${\varepsilon}-optimal$ policy is found by using a modified version of the well-known policy iteration algorithm. A numerical example is also shown.

• Asian-Australasian Journal of Animal Sciences
• /
• 제11권3호
• /
• pp.265-267
• /
• 1998

Cattle in Alaska seemed to be tolerant to low blood selenium (Se) although arctic winter energy demands might exacerbate oxidative damage to tissues lacking protection by Se. The thermogenic properities of brown adipose tissue (BAT) and the long cold adaptation period for Alaskan cattle indicates that they might develop BAT. Eighteen mature beef cows with whole blood Se of< 19 ppb were used to examine possible relation-ships among Se deficiency, tissue lesions, and observable BAT. All cows were wintered on a diet low in Se and nine cows were given supplemental Se provided by intraruminal bolus. Blood Se was elevated (p < .05) from November through February by supplemental Se, but body weight and back fat thickness were unaffected (p < .10). Tissues were taken from two Se deficient cows, two cows supplemented with Se, and two cows given large doses of Se (300 mg by injection) four weeks before slaughter. Histopathological examination of 187 samples of fat (7 to 55 from each cow) showed no observable amount of BAT. Examination of other tissues showed no lesions attributable to Se deficiency.

• 산업기술연구
• /
• 제24권A호
• /
• pp.99-107
• /
• 2004

This paper considers reinforcement learning control with the self-organizing map. Reinforcement learning uses the observable states of objective system and signals from interaction of the system and the environments as input data. For fast learning in neural network training, it is necessary to reduce learning data. In this paper, we use the self-organizing map to parition the observable states. Partitioning states reduces the number of learning data which is used for training neural networks. And neural dynamic programming design method is used for the controller. For evaluating the designed reinforcement learning controller, an inverted pendulum of the cart system is simulated. The designed controller is composed of serial connection of self-organizing map and two Multi-layer Feed-Forward Neural Networks.

• Journal of information and communication convergence engineering
• /
• 제7권4호
• /
• pp.539-544
• /
• 2009

An estimation of 3D velocity field including occluded parts without maxing tracer to the fluid had not only never been proposed but also impossible by the conventional computer vision algorithm. In this paper, we propose a new method of three dimensional optical flow of the fluid based on physical model, where some boundary conditions are given from a priori knowledge of the flow configuration. Optical flow is obtained by minimizing the mean square errors of a basic constraint and the matching error terms with visual data using Euler equations. Here, Navier-Stokes motion equations and the differences between occluded data and observable data are employed as the basic constrains. we verify the effectiveness of our proposed method by applying our algorithm to simulated data with partly artificially deleted and recovering the lacking data. Next, applying our method to the fluid of observable surface data and the knowledge of boundary conditions, we demonstrate that 3D optical flow are obtained by proposed algorithm.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2001년도 ICCAS
• /
• pp.69.1-69
• /
• 2001

In this paper, observability properties of a multiantenna GPS measurement system for the estimation of errors in INS are presented. It is shown that time-invariant INS error models are observable with measurements from at least three GPS antennas on the vehicle. There is at least one unobservable mode with two antennas. There are three unobservable modes with one antenna. It is also shown that time-varying INS error models are instantaneously observable with measurements from three GPS antennas. A numerical simulation results are given to verify the effectiveness of the multiantenna measurement system on the INS error estimation. In the simulation, a GPS measurement system is considered in which a trade-off between computational load and accuracy of estimation is achieved.

• 품질경영학회지
• /
• 제22권4호
• /
• pp.59-78
• /
• 1994

In some observational studies, we have often random censoring model. However, the data available may be partially observable censored data consisting of the observed failure times and only those nonfailure times which are subject to follow up. In this paper, we present an extension of the problem of partially parametric estimation of the survival function to such partially observable censored data. The proposed estimator treats the observed failure times nonparametrically and uses a parametric model only for those nonfailure times which are subject to follow-up. We discuss the motivation and construction of the proposed estimator and investigate the limiting properties of the proposed estimator such as asymptotic normality. Also, when the assumed parametric model is exponential, the asymptotic variance of the estimator is obtained. Furthermore, an example is given to compare the proposed estimator with the modified Kaplan Meier(MKM) estimator. From the results, it is shown that the relative efficiency of the proposed estimator is higher than that of the MKM estimator in the follow-up study with increasing time.

• 천문학회지
• /
• 제46권2호
• /
• pp.65-74
• /
• 2013

We probe the feasibility of high-frequency radio observations of very rapid flux variations in compact active galactic nuclei (AGN). Our study assumes observations at 230GHz with a small 6-meter class observatory, using the SNU Radio Astronomical Observatory (SRAO) as an example. We find that 33 radio-bright sources are observable with signal-to-noise ratios larger than ten. We derive statistical detection limits via exhaustive Monte Carlo simulations assuming (a) periodic, and (b) episodic flaring flux variations on time-scales as small as tens of minutes. We conclude that a wide range of flux variations is observable. This makes high-frequency radio observations-even with small observatories-a powerful probe of AGN intra-day variability; especially, those which complement observations at lower radio frequencies with larger observatories like the Korean VLBI Network (KVN).

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2000년도 제15차 학술회의논문집
• /
• pp.54-54
• /
• 2000

For its synergistic relationship, an integrated SDINS/GPS system has been adopted in many navigation areas. As an application of SDINS/GPS integration, the in-flight alignment process of a SDINS utilizing GPS carrier phase measurements is introduced and analyzed via an observability analysis using nul1 space method. A measurement model of double-differenced GPS carrier phase measurements is newly derived in order to be used with a SDINS error model. Also, conditions for determining the complete observability of a SDINS/GPS system are suggested and proved. Consequently, it is shown that the system is not completely observable in case of one basel me. With one baseline aligned with y-axis of body frame, pitch error and x-axis accelerometer bias are unobservable states. Also shown is that al1 states are completely observable when sequential maneuver is performed. Above results are confirmed by a covariance analysis.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2000년도 제15차 학술회의논문집
• /
• pp.489-489
• /
• 2000

Recently, Reinforcement Learning(RL) methods have been used far teaming problems in Partially Observable Markov Decision Process(POMDP) environments. Conventional RL-methods, however, have limited applicability to POMDP To overcome the partial observability, several algorithms were proposed [5], [7]. The aim of this paper is to extend our previous algorithm for POMDP, called Labeling Q-learning(LQ-learning), which reinforces incomplete information of perception with labeling. Namely, in the LQ-learning, the agent percepts the current states by pair of observation and its label, and the agent can distinguish states, which look as same, more exactly. Labeling is carried out by a hash-like function, which we call Labeling Function(LF). Numerous labeling functions can be considered, but in this paper, we will introduce several labeling functions based on only 2 or 3 immediate past sequential observations. We introduce the basic idea of LQ-learning briefly, apply it to maze problems, simple POMDP environments, and show its availability with empirical results, look better than conventional RL algorithms.