• Korean Management Science Review
• /
• v.29 no.1
• /
• pp.1-14
• /
• 2012

The goal of the multi-reservoir operation planning is to provide an optimal release plan that maximize the reservoir storage and hydropower generation while minimizing the spillages. However, the reservoir operation is difficult due to the uncertainty associated with inflows. In order to consider the uncertain inflows in the reservoir operating problem, we present a Stochastic Dynamic Programming (SDP) model based on the markov decision process (MDP). The objective of the model is to maximize the expected value of the system performance that is the weighted sum of all expected objective values. With the SDP model, multi-reservoir operating rule can be derived, and it also generates the steady state probabilities of reservoir storage and inflow as output. We applied the model to the Geum-river basin in Korea and could generate a multi-reservoir monthly operating plan that can consider the uncertainty of inflow.

• Journal of Korean Society of Transportation
• /
• v.26 no.1
• /
• pp.127-135
• /
• 2008

Generally, optimum shortest path algorithms adopt single attribute objective among several attributes such as travel time, travel cost, travel fare and travel distance. On the other hand, multi-objective shortest path algorithms find the shortest paths in consideration with multi-objectives. Up to recently, the most of all researches about multi-objective shortest paths are proceeded only in single transportation mode networks. Although, there are some papers about multi-objective shortest paths with multi-modal transportation networks, they did not consider transfer problems in the optimal solution level. In particular, dynamic programming method was not dealt in multi-objective shortest path problems in multi-modal transportation networks. In this study, we propose a multi-objective shortest path algorithm including dynamic programming in order to find optimal solution in multi-modal transportation networks. That algorithm is based on two-objective node-based label correcting algorithm proposed by Skriver and Andersen in 2000 and transfer can be reflected without network expansion in this paper. In addition, we use non-dominated paths and tree sets as labels in order to improve effectiveness of searching non-dominated paths. We also classifies path finding attributes into transfer and link travel attribute in limited transit networks. Lastly, the calculation process of proposed algorithm is checked by computer programming in a small-scaled multi-modal transportation network.

• Journal of Electrical Engineering and Technology
• /
• v.9 no.2
• /
• pp.415-422
• /
• 2014

This paper presents a new approach to solve the multi area unit commitment problem (MAUCP) using an evolutionary programming based particle swarm optimization (EPPSO) method. The objective of this paper is to determine the optimal or near optimal commitment schedule for generating units located in multiple areas that are interconnected via tie lines. The evolutionary programming based particle swarm optimization method is used to solve multi area unit commitment problem, allocated generation for each area and find the operating cost of generation for each hour. Joint operation of generation resources can result in significant operational cost savings. Power transfer between the areas through the tie lines depends upon the operating cost of generation at each hour and tie line transfer limits. Case study of four areas with different load pattern each containing 7 units (NTPS) and 26 units connected via tie lines have been taken for analysis. Numerical results showed comparing the operating cost using evolutionary programming-based particle swarm optimization method with conventional dynamic programming (DP), evolutionary programming (EP), and particle swarm optimization (PSO) method. Experimental results show that the application of this evolutionary programming based particle swarm optimization method has the potential to solve multi area unit commitment problem with lesser computation time.

• International Journal of Control, Automation, and Systems
• /
• v.2 no.3
• /
• 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.

• Industrial Engineering and Management Systems
• /
• v.13 no.1
• /
• pp.29-42
• /
• 2014

Extensive research has been conducted on supplier evaluation and selection as a strategic and crucial component of supply chain management in recent years. However, few articles in the previous literature have been dedicated to the use of fuzzy inference systems as an aid in decision-making. Therefore, this essay attempts to demonstrate the application of this method in evaluating suppliers, based on a comprehensive framework of qualitative and quantitative factors besides the effect of gradual coverage distance. The purpose of this study is to investigate the applicability of the numerous measures and metrics in a multi-objective optimization problem of the supply chain network design with the aim of managing the allocation of orders by coordinating the production lines to satisfy customers' demand. This work presents a dynamic non-linear programming model that examines the important aspects of the strategic planning of the manufacturing in supply chain. The effectiveness of the configured network is illustrated using a sample, following which an exact method is used to solve this multi-objective problem and confirm the validity of the model, and finally the results will be discussed and analyzed.

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.25 no.3
• /
• pp.44-50
• /
• 2017

This paper proposes arrival scheduling and advisory generation algorithms which can be used in the terminal airspace with Point-Merge procedures. The proposed scheduling algorithm consists of two steps. In the first step, the algorithm computes aircraft schedules at the entrance of the Point-Merge sequencing legs based on First-Come First-Served(FCFS) strategy. Then, in the second step, optimal sequence and schedules of all aircraft at the runway are computed using Multi-Objective Dynamic Programming(MODP) method. Finally, the advisories that have to be provided to the air traffic controllers are generated. To demonstrate the proposed algorithms, the simulation was conducted based on Jeju International Airport environments.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.171-171
• /
• 2020

Decision making strategies should consider both adaptiveness and robustness in order to deal with two main characteristics of climate change: non-stationarity and deep uncertainty. Especially, robust strategies are different from traditional optimal strategies in the sense that they are satisfactory over a wider range of uncertainty and may act as a key when confronting climate change. In this study, a new framework named Robust Stochastic Dynamic Programming (R-SDP) is proposed, which couples previously developed robust optimization (RO) into the objective function and constraint of SDP. Two main approaches of RO, feasibility robustness and solution robustness, are considered in the optimization algorithm and consequently, three models to be tested are developed: conventional-SDP (CSDP), R-SDP-Feasibility (RSDP-F), and R-SDP-Solution (RSDP-S). The developed models were used to derive optimal monthly release rules in a single reservoir, and multiple simulations of the derived monthly policy under inflow scenarios with varying mean and standard deviations are undergone. Simulation results were then evaluated with a wide range of evaluation metrics from reliability, resiliency, vulnerability to additional robustness measures. Evaluation results were finally visualized with advanced visualization tools that are used in multi-objective robust decision making (MORDM) framework. As a result, RSDP-F and RSDP-S models yielded more risk averse, or conservative, results than the CSDP model, and a trade-off relationship between traditional and robustness metrics was discovered.

• Korean Journal of Hydrosciences
• /
• v.7
• /
• pp.61-75
• /
• 1996

The purpose of this study is to develop an efficient model for the least cost design of multi-site detention systems. The IDP (Incremental Dynamic Programming) model for optimal design is composed of two sub-models : hydrologic-hydraulic model and optimization model. The objective function of IDP is the sum of costs ; acquisition cost of the land, construction cost of detention basin and pumping system. Model inputs include channel characteristics, hydrologic parameters, design storm, and cost function. The model is applied to the Jung-Rang Cheon basin in Seoul, a watershed with cetention basins in multiple branching channels. The application results show that the detention system can be designed reasonably for various conditions and the model can be applied to multi-site detention system design.

• Journal of Korea Water Resources Association
• /
• v.38 no.9 s.158
• /
• pp.737-746
• /
• 2005

Optimization is a process that searches an optimal solution to obtain maximum or minimum value of an objective function. Many researchers have focused on effective search algorithms for the optimum but few researches were interested in establishing the objective function. This study compares two approaches for the objective function: one allows a tradeoff among the objectives and the other does not allow a tradeoff by assigning weights for the absolute priority between the objectives. An optimization model using sampling stochastic dynamic programming was applied to these two objective functions and the resulting optimal policies were compared. As a result, the objective function with no tradeoff provides a decision making process that matches practical reservoir operations than that with a tradeoff allowed. Therefore, it is more reasonable to establish the objective function with no a tradeoff among the objectives for multi-purpose dam operation plan in Korea.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• 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.