• Journal of information and communication convergence engineering
• /
• 제20권1호
• /
• pp.22-30
• /
• 2022

The reinforcement learning algorithm has proven its potential in solving sequential decision-making problems under uncertainties, such as finding paths to route data packets in wireless sensor networks. With reinforcement learning, the computation of the optimum path requires careful definition of the so-called reward function, which is defined as a linear function that aggregates multiple objective functions into a single objective to compute a numerical value (reward) to be maximized. In a typical defined linear reward function, the multiple objectives to be optimized are integrated in the form of a weighted sum with fixed weighting factors for all learning agents. This study proposes a reinforcement learning -based routing protocol for wireless sensor network, where different learning agents prioritize different objective goals by assigning weighting factors to the aggregated objectives of the reward function. We assign appropriate weighting factors to the objectives in the reward function of a sensor node according to its hop-count distance to the sink node. We expect this approach to enhance the effectiveness of multi-objective reinforcement learning for wireless sensor networks with a balanced trade-off among competing parameters. Furthermore, we propose SDN (Software Defined Networks) architecture with multiple controllers for constant network monitoring to allow learning agents to adapt according to the dynamics of the network conditions. Simulation results show that our proposed scheme enhances the performance of wireless sensor network under varied conditions, such as the node density and traffic intensity, with a good trade-off among competing performance metrics.

• 전기학회논문지
• /
• 제57권10호
• /
• pp.1869-1875
• /
• 2008

Recently, a direct data-driven synthesis of a proportional integral derivative(PID) controller for a linear time-invariant(LTI) plant was presented in [1]. The authors showed that a complete set of PID controllers achieving robust performance and stability can be calculated directly from frequency response(FR) data without an identified transfer function model. However, it is not convenient to use this method because it requires complicated numerical algorithms to find specific frequencies which are solutions of an identical equation. The method also requires determination of the boundary of the controller's parameters from a finite set of FR data. In this paper, we present the development of a user-friendly Matlab toolbox based on the method in [1]. This toolbox allows us to obtain a complete three-dimensional(3-D) graphical solution of PID controllers that meet multiple design objectives. Several examples are given to demonstrate the use of the toolbox.

• Journal of Computational Design and Engineering
• /
• 제3권1호
• /
• pp.71-90
• /
• 2016

Steel nitriding is a thermo-chemical process leading to surface hardening and improvement in fatigue properties. The process is strongly influenced by many different variables such as steel composition, nitrogen potential, temperature, time, and quenching media. In the present study, the influence of such parameters affecting physic-chemical and mechanical properties of nitride steels was evaluated. The aim was to streamline the process by numerical-experimental analysis allowing defining the optimal conditions for the success of the process. Input parameters-output results correlations were calculated through the employment of a multi-objective optimization software, modeFRONTIER (Esteco). The mechanical and microstructural results belonging to the nitriding process, performed with different processing conditions for various steels, are presented. The data were employed to obtain the analytical equations describing nitriding behavior as a function of nitriding parameters and steel composition. The obtained model was validated, through control designs, and optimized by taking into account physical and processing conditions.

• 한국공작기계학회논문집
• /
• 제16권4호
• /
• pp.29-35
• /
• 2007

The main objective of this paper is to propose the optimal design method of forging process using genetic algorithm. Design optimization of forging process was doing about one stage and multi stage. The objective function is considered the filling of die. The chosen design variables are die geometry in multi stage and initial billet shape in one stage. We performed FE analysis to simulated forging process. The optimized preform and initial billet shape was obtained by genetic algorithm and FE analysis. To show the efficiency of GA method in forging problem are solved and compared with published results.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2008년도 제39회 하계학술대회
• /
• pp.852-853
• /
• 2008

This paper proposes an optimal design method for the shape optimization of the permanent magnets (PM) of an in-wheel permanent magnet synchronous motor (PMSM) to reduce the cogging torque considering a total harmonic distortion (THD) and a root mean square (RMS) value of back-EMF. In this method, the Kriging model based on a design of experiment (DOE) is applied to interpolate the objective function in the spaces of design parameters. The optimal design method for the PM of an in-wheel PMSM has to consider multi-variable and multi-objective functions. The developed design method is applied to the optimization for the PM of an in-wheel PMSM.

• 대한전기학회논문지:전력기술부문A
• /
• 제52권10호
• /
• pp.584-592
• /
• 2003

This paper presents a fuzzy-GA method for the allocation and operation of dispersed generator systems(DGs) based on load model in composite distribution systems. Groups of each individual load model consist of residential, industrial, commercial, official and agricultural load. The problem formulation considers an objective to reduce power loss of distribution systems and the constraints such as the number or total capacity of DGs and the deviation of the bus voltage. The main idea of solving fuzzy goal programming is to transform the original objective function and constraints into the equivalent multi-objectives functions with fuzzy sets to evaluate their imprecise nature for the criterion of power loss minimization, the number or total capacity of DGs and the bus voltage deviation, and then solve the problem using genetic algorithm. The method proposed is applied to IEEE 12 bus and 33 bus test systems to demonstrate its effectiveness. .

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2002년도 추계학술대회 및 정기총회
• /
• pp.395-398
• /
• 2002

본 논문에서는 ‘다목적 함수 최적화 문제(Multi-objective Optimization Problem MOP)’를 풀기 위하여 유전자 알고리즘을 진화적 게임 이론 적용시킨 ‘내쉬 유전자 알고리즘(Nash GA)’과 본 논문에서 새로이 제안하는 공진화 알고리즘의 구조를 설명하고 이 두 알고리즘의 결과를 시뮬레이션을 통하여 비교 검토함으로써 ‘진화적 게임 이론(Evolutionary Game Theory : EGT)’의 두 가지 아이디어 -‘내쉬의 균형(Equilibrium)’과 ‘진화적 안정전략(Evolutionary Stable Strategy . ESS)’-에 기반한 최적화 알고리즘들이 다목적 함수 문제의 최적해를 탐색할 수 있음을 확인한다.

• 대한산업공학회지
• /
• 제29권3호
• /
• pp.213-221
• /
• 2003

In this paper, we consider a multi-objective part-machine grouping problem, in which part types have several alternative part routings and each part routing has a machining sequence. This problem is characterized as optimally determining part type sets and its corresponding machine cells such that the sum of inter-cell part movements and the sum of machine workload imbalances are simultaneously minimized. Due to the complexity of the problem, a two-stage heuristic algorithm is proposed, and experiments are shown to verify the effectiveness of the algorithm.

• 한국지반공학회논문집
• /
• 제18권1호
• /
• pp.71-78
• /
• 2002

본 연구에서는 다층지반에 근입된 흙막이 벽의 단계별 계측변위로부터 각 층의 지반물성을 추정하고 이로부터 차기단계의 거동을 예측하기 위한 역해석 기법을 제안하였다. 지반이 다수의 층으로 구성되어 있을 경우 찾아야 할 대상변수가 많아지게 되며, 대상변수가 많아질수록 역해석에 상당한 무리가 따르게 된다. 이러한 층별 지반물성을 효율적으로 추정하기 위하여 최하단층부터 순차적으로 대상변수들을 찾아가는 방법을 이용하였다. 역해석은 상당량의 반복계산이 필요하기 때문에 정해석 방법으로는 해석시간이 짧고 시공단계 별 해석이 가능한 탄소성보법을 사용하였다. 역해석 대상변수는 탄소성 하중-변위 곡선의 구성요소인 지반반력계수와 수평토압계수들을 취하였으며, 목적함수는 이상변위에 의한 오차를 최소화시키기 위하여 단계별 계측변위 증분과 해석변위 증분의 차이로 구성하였다. 목적함수를 최소화 시키는 대상변수들을 찾기 위한 최적화 수법으로는 제약순차선형계획 법을 이용하였다. 본 연구를 통하여 제안된 방법을 수치해석자료 및 현장계측자료를 이용하여 검증하였다.

• 한국산업정보학회논문지
• /
• 제27권4호
• /
• pp.37-45
• /
• 2022

본 논문에서는 대표적인 조합 최적화(combinatorial optimization) 문제인 다수 에이전트-다수 작업 할당 문제를 제시한다. 할당 문제의 목적은 각 작업의 달성률(achievement rate)의 합을 최대로 하는 에이전트-작업 할당을 결정하는 것이다. 달성률은 각 작업의 할당된 에이전트의 수에 따라 아래 오목 증가(concave down increasing)형태로 다루어지며, 본 할당 문제는 비선형(non-linearity)의 목적함수를 갖는 NP-난해(NP-hard) 문제로 표현된다. 본 논문에서는 할당 문제를 해결하기 위한 효과적이면서 효율적인 문제 해결 방법론으로 혼합 교차-엔트로피 알고리즘(hybrid cross-entropy algorithm)을 제안한다. 일반적인 교차-엔트로피 알고리즘은 문제 상황에 따라 느린 매개변수 업데이트 속도와 조기수렴(premature convergence)이 발생할 수 있다. 본 연구에서 제안하는 문제 해결 방법론은 이러한 단점의 발생 확률을 낮추도록 설계되었으며, 실험적으로도 우수한 성능을 보이는 알고리즘임을 수치실험을 통해 제시한다.