• International Journal of Control, Automation, and Systems
• /
• 제2권3호
• /
• pp.263-278
• /
• 2004

This paper reviews dynamic programming (DP), surveys approximate solution methods for it, and considers their applicability to process control problems. Reinforcement Learning (RL) and Neuro-Dynamic Programming (NDP), which can be viewed as approximate DP techniques, are already established techniques for solving difficult multi-stage decision problems in the fields of operations research, computer science, and robotics. Owing to the significant disparity of problem formulations and objective, however, the algorithms and techniques available from these fields are not directly applicable to process control problems, and reformulations based on accurate understanding of these techniques are needed. We categorize the currently available approximate solution techniques fur dynamic programming and identify those most suitable for process control problems. Several open issues are also identified and discussed.

• 대한기계학회논문집A
• /
• 제27권6호
• /
• pp.864-870
• /
• 2003

In recent engineering, the designer has become more and more dependent on the computer simulations such as FEM(Finite Element Method) and BEM(Boundary Element Method). In order to optimize such implicit models more efficiently and reliably, the meta -modeling technique has been developed for solving such a complex problems combined with the DACE(Design and Analysis of Computer Experiments). It is widely used for exploring the engineer's design space and for building approximation models in order to facilitate an effective solution of multi-objective and multi-disciplinary optimization problems. Optimization of a train suspension is performed according to the minimization of forty -six responses that represent ten ride comforts, twelve derailment quotients, twelve unloading ratios, and twelve stabilities by using the Kriging model of a train suspension. After each Kriging model is constructed, multi -objective optimal solutions are achieved by using a nonlinear programming method called SQP(Sequential Quadratic Programming).

• International Journal of Control, Automation, and Systems
• /
• 제6권3호
• /
• pp.401-412
• /
• 2008

This paper discusses a design methodology of cooperative path planning for dynamical multi-agent systems with spatial and temporal constraints. The cooperative behavior of the multi-agent systems is specified in terms of the objective function in an optimization formulation. The path of achieving cooperative tasks is then generated by the optimization formulation constructed based on a differential flatness approach. Three scenarios of multi-agent tasking are proposed at the cooperative task planning framework. Given agent dynamics, both spatial and temporal constraints are considered in the path planning. The path planning algorithm first finds trajectory curves in a lower-dimensional space and then parameterizes the curves by a set of B-spline representations. The coefficients of the B-spline curves are further solved by a sequential quadratic programming solver to achieve the optimization objective and satisfy these constraints. Finally, several illustrative examples of cooperative path/task planning are presented.

• 한국경영과학회:학술대회논문집
• /
• 한국경영과학회 2006년도 추계학술대회
• /
• pp.513-524
• /
• 2006

The ammunition allocation problem is a Multi-objective optimization(MOO) problem, maximizing fill-rate of multiple user troops and minimizing transportation time. Recent studies attempted to solve this problem by the prior preference articulation approach such as goal programming. They require that all the preference information of decision makers(DM) should be extracted prior to solving the problem. However, the prior preference information is difficult to implement properly in a rapidly changing state of war. Moreover they have some limitations such as heavy cognitive effort required to DM. This paper proposes a new ammunition allocation model based on more reasonable assumptions and uses an interactive MOO method to the ammunition allocation problem to overcome the limitations mentioned above. In particular, this article uses the GDF procedure, one of the well-known interactive optimization methods in the MOO liter-ature, in solving the ammunition allocation problem.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.1296-1299
• /
• 1996

This paper proposes a design automation method for the flight control system of an aircraft based on optimization. The control system design problem which has many specifications is formulated as multi-objective optimization problem. The solution of this optimization problem should be considered in terms of Pareto-optimality. In this paper, we use an evolutionary algorithm providing numerous Pareto-optimal solutions. These solutions are given to a control system designer and the most suitable solution is selected. This method decreases tasks required to determine the control parameters satisfying all specifications. The design automation of a flight control system is illustrated through an example.

• 한국수자원학회논문집
• /
• 제38권9호
• /
• pp.737-746
• /
• 2005

최적화란 목적함수가 최대 또는 최소가 되도록 하는 결정변수를 찾아가는 절차이다. 기존의 많은 연구자들은 최적해의 효율적인 탐색과정에 집중한 반면 최적화의 시작점이라 할 수 있는 목적함수 구성을 위한 연구는 상대적으로 미진한 것이 사실이다. 따라서 본 연구에서는 국내외에서 빈번히 사용되고 있는 가중평균법을 사용하여 tradeoff를 고려한 목적함수와 절대우선순위를 위한 가중값을 적용한 목적함수를 구성하여 표본추계학적 동적계획법을 통해 산정한 최적운영률을 비교하였다. 그 결과 절대우선순위를 위한 가중값을 적용한 경우가 보다 실제 저수지운영과 부합하는 결과를 나타내는 것을 확인할 수 있었다. 따라서 국내 다목적댐 운영계획에 보다 적합한 목적함수를 구성하기 위해서는 절대우선순위를 위한 가중값을 부여하여 목적함수를 구성하는 것이 타당한 것으로 판단된다.

• 한국경영과학회지
• /
• 제15권2호
• /
• pp.1-15
• /
• 1990

A mathematical definition of the cluster is suggested. A nonlinear 0-1 integer programming formulation for the multi-dimensional entity clustering problem is developed. A heuristic method named MDEC (Multi-Dimensional Entity Clustering) using centroids and the binary partition is developed and the numerical examples are shown. This method has an advantage of providing bottle-neck entity informations.

• 한국품질경영학회:학술대회논문집
• /
• 한국품질경영학회 1998년도 The 12th Asia Quality Management Symposium* Total Quality Management for Restoring Competitiveness
• /
• pp.680-695
• /
• 1998

The objective of this study is to optimize the responses of the creep feed ceramic grinding process simultaneously by an off-1ine multicriteria optimization methodology. The responses considered as objectives are material removal rate, flexural strength, normal grinding force, workpiece surface roughness and grinder power. Alumina material was ground by the creep feed grinding mode using superabrasive grinding wheels. The process variables optimized for the above objectives include grinding wheel specification, such as bond type, mesh size, and grit concentration, and grinding process parameters, such as depth of cut and feed rate. A weighting method transforms the multi-objective problem into a single-objective programming format and then, by parametric variation of weights, the set of non-dominated optimum solutions are obtained. Finally, the multi-objective optimization methodology was tested by a sensitivity analysis to check the stability of the model.

• 한국경영과학회지
• /
• 제8권1호
• /
• pp.41-59
• /
• 1983

The purpose of this paper is to present a method of multi-objective, stochastic optimization in water resources system which investigates the development of potential non-normal deterministic equivalents for subsequent use in multiobjective stochastic programming methods, Given probability statement involving a function of several random variables, it is often possible to obtain a deterministic equivalent of it that does not include any orginal random variables. A Stochastic trade-off technique-MSTOT is suggested to help a decision maker achieve satisfactory levels for several objective functions. This makes use of deterministic equivalents to handle random variables in the objective functions. The emphasis is in the development of non-normal deterministic equivalents for use in multiobjective stochastic techniques.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2003년도 ICCAS
• /
• pp.142-145
• /
• 2003

Many control applications using Neural Network need a priori information about the objective system. But it is impossible to get exact information about the objective system in real world. To solve this problem, several control methods were proposed. Reinforcement learning control using neural network is one of them. Basically reinforcement learning control doesn't need a priori information of objective system. This method uses reinforcement signal from interaction of objective system and environment and observable states of objective system as input data. But many methods take too much time to apply to real-world. So we focus on faster learning to apply reinforcement learning control to real-world. Two data types are used for reinforcement learning. One is reinforcement signal data. It has only two fixed scalar values that are assigned for each success and fail state. The other is observable state data. There are infinitive states in real-world system. So the number of observable state data is also infinitive. This requires too much learning time for applying to real-world. So we try to reduce the number of observable states by classification of states with Self-Organizing Map. We also use neural dynamic programming for controller design. An inverted pendulum on the cart system is simulated. Failure signal is used for reinforcement signal. The failure signal occurs when the pendulum angle or cart position deviate from the defined control range. The control objective is to maintain the balanced pole and centered cart. And four states that is, position and velocity of cart, angle and angular velocity of pole are used for state signal. Learning controller is composed of serial connection of Self-Organizing Map and two Multi-layer Feed-Forward Neural Networks.