검색 상세

적응형 시뮬레이션을 위한 범용 시뮬레이션 환경

초록/요약

Simulation is widely used in many fields, such as engineering, defense, and commerce, in that it can simulate various situations of reality according to the user's purpose. To properly simulate the actual system, it is necessary to select the structure design and simulation environment following the characteristics of the target system. This process is based on existing modules and simulation engine in relation to simulation configuration, so the process of designing new simulation is limited by the characteristics of the selected module and simulation engine. Also, when a new module is used, it is designed not to be supported by the engine itself but to load and use the module for the model itself. The process of building simulation is divided into modeling and simulation. The user performs a modeling process by abstracting the real-world system to be simulated. The simulation is performed by constructing the target system through combination of the models derived through the modeling process and executing it. Therefore, the simulation engine just did its job by running the models. This study aims at devising a generalized simulation environment that can adapt to the environment. Simulation results based on simulation scenarios for specific situations are used as data to analyze the target system. Most of the tasks for analyzing the simulation are configured to derive the optimal route or the result, which makes it easier to construct an adaptive simulation environment through this study. To construct an adaptive simulation environment, this study integrates the reinforcement learning module with the engine so that the user can configure the adaptive simulation using the corresponding module. In addition, the engine proposed in this study enables users to access and utilize the simulation time management by storing and retrieving the snapshot for the simulation specific phase through its DB operation. This allows the user to design a higher level of simulation. In the case study, the features of the adaptive simulation environment are introduced through the simulation examples constructed using the proposed simulation engine.

more

목차

I. 서론 1
1. 연구의 필요성 1
2. 이론적 배경 6
2.1. DEVS(Discrete Event System Specifiaction) 형식론 6
2.1.1. 이산 사건 시스템 6
2.1.2. DEVS 모델링 방법론 8
2.1.3. 원자 모델 9
2.1.4. 결합 모델 11
2.2. 강화 학습 알고리즘 12
2.2.1. 강화 학습 개요 12
2.2.2. 마르코프 결정 과정(Markov Decision Process) 13
3. 관련 연구 15
3.1. 기존 시뮬레이션 엔진 15
3.1.1. DEVSim++ 15
3.1.2. Matlab & Simulink 18
II. 본론 19
1. 개요 19
2. mlDEVS 형식론 19
2.1. 기본 모델(Base Model) 20
2.2. 모델링 규칙 21
3. mlDEVS 모델의 시뮬레이션 21
3.1. 시뮬레이터 알고리즘 21
3.2. 학습 알고리즘 22
4. 모델 구현 23
4.1. 기본 모델 구현 23
4.1.1. 클래스 구조 23
4.1.2. SimObject 클래스 23
4.1.3. SimBaseModel 클래스 24
5. 시뮬레이터 구현 26
5.1. 시뮬레이터 구성 26
5.2. 시뮬레이션 제어 26
5.3. 시뮬레이션 이벤트 메시지 제어 31
5.4. 시뮬레이션 스냅샷 저장 및 불러오기 기능 32
III. 사례연구 35
1. 실험 개요 35
2. 실험 설계 35
3. 실험 과정 및 결과 분석 37
IV. 결론 42
V. 참고문헌 44

more
반출 Meta View 목록