• Journal of the Institute of Convergence Signal Processing
• /
• v.4 no.3
• /
• pp.49-54
• /
• 2003

In this paper, we propose a controller with the self-organizing neural network compensator for compensating PID controller's response. PID controller has simple design method but needs a lot of trials and errors to determine coefficients. A neural network control method does not have optimal structure as the parameters are pre-specified by designers. In this paper, to solve this problem, we use a self-organizing neural network which has Back Propagation Network algorithm using a Gaussian Potential Function as an activation function of hidden layer nodes for compensating PID controller's output. Self-Organizing Neural Network's learning is proceeded by Gaussian Function's Mean, Variance and number which are automatically adjusted. As the results of simulation through the second order plant, we confirmed that the proposed controller get a good response compare with a PID controller. And we implemented the of controller performance hydraulic servo motor system using the DSP processor. Then we observed an experimental results.

• KIPS Transactions on Computer and Communication Systems
• /
• v.7 no.4
• /
• pp.103-110
• /
• 2018

Services that enhance user convenience by using various IoT devices are increasing with the development of Internet of Things(IoT) technology. Also, since the price of IoT sensors has become cheaper, companies providing services by collecting and utilizing data from various sensors are increasing. The smart IoT home system is a representative use case that improves the user convenience by using IoT devices. To improve user convenience of Smart IoT home system, this paper proposes a method for the control of related devices based on data analysis. Internal environment measurement data collected from IoT sensors, device control data collected from device control actuators, and user judgment data are learned to predict the current home state and control devices. Especially, differently from previous approaches, it uses deep neural network to analyze the data to determine the inner state of the home and provide information for maintaining the optimal inner environment. In the experiment, we compared the results of the long-term measured data with the inferred data and analyzed the discrimination performance of the proposed method.

• The KIPS Transactions:PartB
• /
• v.9B no.5
• /
• pp.587-598
• /
• 2002

In multicast environment, the main objective of group key management is to provide security services to group communications by sharing a single group key among all the members of the group and subsequently encrypting and decrypting all the communication messages exchanged among the members of the group. Up to now, there has been no effort to develop group key management mechanism that considers the rate of users' join/leave operations. Hence, in this research, we propose group key management mechanisms that consider the rate of user's join/leave operations. We also define a new tree structure called variable tree which is much more flexible than full regular trees and show that variable trees are more efficient than full regular trees for group key management. Especially, we propose an algorithm that minimizes the necessary number of rekey messages according to the rate of join and leave operations. We also shows that if the rate of leave operation is greater than 50%, then the tree structure with degrees 2 or 3 are the optimal structures.

• KIEAE Journal
• /
• v.17 no.5
• /
• pp.69-76
• /
• 2017

Purpose: This study aimed at developing an Artificial Neural Network (ANN) model for predicting the amount of cooling energy consumption of the variable refrigerant flow (VRF) cooling system by the different set-points of the control variables, such as supply air temperature of air handling unit (AHU), condenser fluid temperature, condenser fluid pressure, and refrigerant evaporation temperature. Applying the predicted results for the different set-points, the control algorithm, which embedded the ANN model, will determine the most energy efficient control strategy. Method: The ANN model was developed and tested its prediction accuracy by using matrix laboratory (MATLAB) and its neural network toolbox. The field data sets were collected for the model training and performance evaluation. For completing the prediction model, three major steps were conducted - i) initial model development including input variable selection, ii) model optimization, and iii) performance evaluation. Result: Eight meaningful input variables were selected in the initial model development such as outdoor temperature, outdoor humidity, indoor temperature, cooling load of the previous cycle, supply air temperature of AHU, condenser fluid temperature, condenser fluid pressure, and refrigerant evaporation temperature. The initial model was optimized to have 2 hidden layers with 15 hidden neurons each, 0.3 learning rate, and 0.3 momentum. The optimized model proved its prediction accuracy with stable prediction results.

• Journal of Korean Society of Steel Construction
• /
• v.19 no.4
• /
• pp.395-402
• /
• 2007

Probabilistic risk assessment was conducted on a hybrid cable-stayed bridge consisting of a steel-composite plate girder and a concrete girder with a long span, designed using the working stress design and strength design methods. The component reliabilities of the bridge's cables, pylons, girders, and steel-concrete conjunction were evaluated using the AFOSM(Advanced First Order Second Moment) algorithm and the simulation technique at the critical sections, based on the maximum axial force, shear, and positive and negative moments of the selected sections. For the analysis of system reliability, the hybrid cable-stayed bridge consisting of cables, pylons, and plate girders was modeled into combined failure modes, and for system reliability, the probabilities of failure and reliability index of the structural system were evaluated. Based on the results of this study, the critical failure modes of the hybrid cable-stayed bridge based on the bridge's structural characteristics are suggested, and the efficiency of the partial ETA technique for use in the risk assessment method was confirmed.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2012.05a
• /
• pp.637-637
• /
• 2012

본 연구에서는 내수침수 저감을 위하여 효율적(effective)인 우수저류조 설치에 따른 침수저감효과 극대화 방안을 제시하고자 한다. 여기서 효율성(effectiveness)은 침수저감량의 극대화 측면과 비용의 최소화 측면 두 가지로 구분된다. 최적 방재 시설물의 설치는 단순 설치비용 대비 저감량이 가장 큰 안을 제시하는 것은 의미가 없으며 일정 기준 이상의 방재성능을 발휘하면서 주어진 예산안에서 최적안을 찾아야 하므로 비용의 최소화 측면과 침수 저감량, 즉 맨홀에서의 월류 저감량을 최대화 하는 두 가지의 목적을 동시에 달성해야 한다. 따라서 본 연구에서는 다목적 최적화 알고리즘의 적용을 통하여 우수저류조 최적 설치지점을 선정하는 기법을 제안하였다. 본 연구에 적용한 다목적 최적화 방법으로는 목적함수의 최적해 탐색 효용성 측면에서 우수하다고 평가되고 있는 유전자 알고리즘을 적용하였다. 다목적 최적화의 경우 해의 우열을 판단하기 위한 적합도 함수는 실제 각 목적함수의 적합도 값(real fitness value)이 아닌 해의 상대적인 우열(dominance or non-dominance)에 따라 부여되는 등급(rank)에 의해서 해의 우열이 결정되며 여기서는 Fonseca and Fleming(1993)이 제안한 Ranking method를 적용하여 적합도를 결정하였다. 한편 도시 우수관망의 해석 및 우수저류조 설치에 따른 월류량 분석을 위하여 미 환경청(US Environmental Protection Agency; EPA)에서 제공하고 있는 EPA-SWMM 5.0 engine을 사용하였으며 최적화 알고리즘의 구성을 위하여 Visual C++와 SWMM DLL을 연동하여 사용하였다. 연구 대상유역은 인천 청라지구(3공구)를 대상으로 기법의 적용성을 검토하였으며 저류지 설치에 따른 비용함수는 EPA(2002)에서 제안한 저류지 체적대비 공사비용을 원화로 환산한 후 청라지구의 공시지가를 고려하여 결정하였다. 최적화 기법의 적용 결과 저류지 설치비용에 따라 최대로 월류량을 저감시킬 수 있는 우수저류조 최적 설치위치의 조합(Pareto-front)을 결정할 수 있었다.

• Journal of Advanced Marine Engineering and Technology
• /
• v.19 no.1
• /
• pp.60-70
• /
• 1995

In the field of marine transportation the energy saving is one of the most important factors for profit. In order to reduce the fuel oil consumption the ship's propulsion efficiency must be increased as much as possible. The propulsion efficiency depends upon a combination of an engine and a propeller. The propeller has better efficiency as lower rotational speed. This situation led the engine manufacturers to design the engine that has lower speed, longer stroke and a small number of cylinders. Consequently the variation of rotational torque became larger than before because of the longer delay-time in the fuel oil injection process and an increased output per cylinder. As this new trends the conventional mechanical-hydrualic governors for engine speed control have been replaced by digital speed controllers which adopted the PID control or the optimal control algorithm. But these control algorithms have not enough robustness to suppress the variation of the delay-time and the parameter pertubation. In this paper we consider the delay-time and the perturbation of engine parameters as the modeling uncetainties. Next we design the controller which has zero offset in steady state engine speed, based on the two-degree-of-freedom control theory and $\mu$-synthesis. Thd validity of the controller is investigated through the response simulation. We use a personal computer and an analog computer as the digital controller and the engine (plant) part respectively. And, we certify that the designed controller maintains its performance even though the engine parameters may vary.

• Journal of the korean society of oceanography
• /
• v.32 no.4
• /
• pp.163-170
• /
• 1997

Some CMP gathers acquired from shallow marine seismic reflection survey in offshore Korea do not show the hyperbolic trend of moveout. It originated from so-called swell effect of source and streamer, which are towed under rough sea surface during the data acquisition. The observed time deviations of NMO-corrected traces can be entirely ascribed to the swell effect. To correct these time deviations, a residual statics is introduced using Genetic Algorithms (GA) into the swell correction. A new class of global optimization methods known as GA has recently been developed in the field of Artificial Intelligence and has a resemblance with the genetic evolution of biological systems. The basic idea in using GA as an optimization method is to represent a population of possible solutions or models in a chromosome-type encoding and manipulate these encoded models through simulated reproduction, crossover and mutation. GA parameters used in this paper are as follows: population size Q=40, probability of multiple-point crossover P$_c$=0.6, linear relationship of mutation probability P$_m$ from 0.002 to 0.004, and gray code representation are adopted. The number of the model participating in tournament selection (nt) is 3, and the number of expected copies desired for the best population member in the scaling of fitness is 1.5. With above parameters, an optimization run was iterated for 101 generations. The combination of above parameters are found to be optimal for the convergence of the algorithm. The resulting reflection events in every NMO-corrected CMP gather show good alignment and enhanced quality stack section.

• Journal of Biomedical Engineering Research
• /
• v.17 no.4
• /
• pp.449-458
• /
• 1996

Most of the conventional electrocardiowaphs foil to detect signals other than P-QRS-T due to the limited SNR and bandwidth. High-resolution electrocardiography(HRECG) provides better SNR and wider bandwidth for the detection of micro-potentials with higher frequency components such as vontricular late potentials(LP). We have developed a HRECG using uncorrected XYZ lead for the detection of LPs. The overall gain of the amplifier is 4000 and the bandwidth is 0.5-300Hz without using 60Hz notch filter. Three 16-bit A/D converters sample X, Y, and Z signals simultaneously with a sampling frequency of 2000Hz. Sampled data are transmitted to a PC via a DMA-controlled, optically-coupled serial communication channel. In order to further reduce the noise, we implemented a signal averaging algorithm that averaged many instances of aligned beats. The beat alignment was carried out through the use of a template matching technique that finds a location maximizing cross-correlation with a given beat tem- plate. Beat alignment error was reduced to $\pm$0.25ms. FIR high-pass filter with cut-off frequency of 40Hz was applied to remove the low frequency components of the averaged X, Y, and Z signals. QRS onset and end point were determined from the vector magnitude of the sigrlaIL and some parameters needed to detect the existence of LP were estimated. The entire system was designed for the easy application of the future research topics including the optimal lead system, filter design, new parameter extraction, etc. In the developed HRECG, without signal averaging, the noise level was less than 5$\mu$V$_rms RTI$. With signal averaging of at least 100 beats, the noise level was reduced to 0.5$\mu$V$_rms RTI$, which is low enough to detect LPs. The developed HRECG will provide a new advanced functionality to interpretive ECG analyzers.

• Journal of KIISE:Software and Applications
• /
• v.30 no.11
• /
• pp.1054-1061
• /
• 2003

Q-Learning, a representative algorithm of reinforcement learning, experiences repeatedly until estimation values about all state-action pairs of state space converge and achieve optimal policies. When the state space is high dimensional or continuous, complex reinforcement learning tasks involve very large state space and suffer from storing all individual state values in a single table. We introduce Q-Map that is new function approximation method to get classified policies. As an agent learns on-line, Q-Map groups states of similar situations and adapts to new experiences repeatedly. State-action pairs necessary for fine control are treated in the form of rule. As a result of experiment in maze environment and mountain car problem, we can achieve classified knowledge and extract easily rules from Q-Map