系統建模與模擬
Systems Modeling & Simulation
| 節 | 週二 |
|---|---|
A 18:30–19:20 | 系統建模與模擬 EE210 3 節連堂 |
B 19:30–20:20 | |
C 20:30–21:20 |
* 根據陽明交大上課時間表所列
本課程聚焦動態系統的建模、模擬、估測與最佳化,涵蓋以下核心內容: 1. 建模方法:基於物理機理與數據融合驅動的動態系統建模方法,包含時域/頻域參數化建模、機器學習與深度神經網路建模,以及其對應的數學函式解析與理論推導。 2. 系統分析:時間演化系統的數學描述、機電工程系統控制器設計,以及模型模擬與驗證。 3. 智能演算法:無模型強化學習、機器學習分類分群與回歸演算法之應用。 本課程透過數學建模與計算模擬技術,培養動態系統分析與優化之核心能力,強調理論推導與實務應用的整合。學生將掌握從物理現象抽象化至數學模型,再到電腦模擬的完整方法論,並學習機器學習於動態系統建模與模擬的進階應用技術。 課程大綱 (1) 時域/頻域/神經網路系統建模之基本觀念 ( 2) 工程或物理現象與系統模型數學函式之間的關係 (3) 機器學習(Machine Learning)動態建模、模擬與分類 (4) 強化學習(Reinforcement Learning)建模、模擬與控制 The focus of this course is on modeling, simulation, estimation, and optimization of dynamic systems. Concepts taught in this course include physics-based and empirical modeling, dynamic modeling, machine learning and model-free reinforcement learning. For this purpose, the methods to mathematically describe time-evolving systems, the method of system modeling and design of controller for mechanical and electrical engineering systems, machine learning classification and regression are applied. 課程目標: 培養學生使用基於物理和機器學習的模型動態系統的理論和應用。 能夠利用不同建模方法給予系統適當的動態描述。 使學生具備分析工程問題的能力,運用工程知識和工程模擬程序解決問題。 學生將能夠收集和分析時間序列數據,以構建數據驅動的系統控制策略。 學生將能夠制定和執行一個項目,該項目利用機器學習中的課程主題和針對新應用程序的優化方法。 To train students in the theory and application of modeling dynamical systems using physics and machine learning. Students will be able to use different modeling methods to give appropriate dynamic description to the system. Enable students to have the ability to analyze engineering problems, and use engineering knowledge and engineering analysis simulation programs to solve problems. Students will be able to collect and analyze time-series data to construct data-driven system control strategies. Students will be able to formulate and execute a project that utilizes course topics and optimization methods for new applications in machine learning.
微積分、工程數學。
課後作業及平時學期課中表現 Exercise & Attendance : 20% 期中考 Mid. Quiz: 40% 期末考 Final quiz: 40%
- L1: Introduction to Dynamic System Modeling
- L2: Time Domain Model of Systems
- L3: Frequency Domain Model of Systems
- L4: Neural Network Modeling, Simulation, and Control
- L5: Machine Learning Modeling
- L6: Reinforcement Learning, Simulation, and Control
- L7: Modeling and Simulation of Flight Control System
| 週次 | 主題 |
|---|---|
| 第 1 週 | Introduction of Dynamical Systems Modeling and Simulation |
| 第 2 週 | Introduction of Dynamical Systems Modeling and Simulation -Introduces a few Controlled Dynamical Systems: 1. 光學物鏡致動器建模,模擬,與控制 2. 機器學習系統建模與模擬 (a) 結合時變情緒軌跡界面之機器學習自動音樂情緒追蹤系統 (b) 以機器學習演算之風機葉片損傷偵測建模與模擬 (c) 強化學習在無人飛行器之建模,模擬,與飛行控制 3. 基於反應輪的衛星姿態穩定平台系統建模,模擬,與控制 4. 溫度控制系統建模,模擬與PID Neural Network控制 |
| 第 3 週 | Introduction of Dynamical Systems Modeling and Simulation -Introduces a few Controlled Dynamical Systems: 1.光學物鏡致動器建模,模擬,與控制 2. 機器學習系統建模與模擬,案例簡介: (a) 結合時變情緒軌跡界面之機器學習自動音樂情緒追蹤系統 (b) 以機器學習演算之風機葉片損傷偵測建模與模擬 (c) 強化學習在無人飛行器之建模,模擬,與飛行控制 3. 基於反應輪的衛星姿態穩定平台系統建模,模擬,與控制 4. 溫度控制系統建模,模擬與PID Neural Network控制 |
| 第 4 週 | Time Domain Modeling 系統時域建模與模擬 *Input/Output (Unit-pulse response) representation of System *Convolution Model of CTS and DTS *Differential Equation Model of CTS *Difference Equation representation of DTS *Least-Squares Error Modeling of DTS *State Space Modeling of Systems |
| 第 5 週 | Frequency Domain Modeling 系統頻域建模與模擬 |
| 第 6 週 | Frequency Domain Modeling 系統頻域建模與模擬 |
| 第 7 週 | Neural Network Modeling and Simulation 神經網路系統建模與模擬 |
| 第 8 週 | Neural Network Modeling and Simulation 神經網路系統建模與模擬 |
| 第 9 週 | Machine Learning Modeling 機器學習建模與模擬 -Supervised Machine Learning Modeling 監督式機器學習建模,模擬,與應用實例 |
| 第 10 週 | -Unsupervised Machine Learning Modeling 無監督式機器學習建模,模擬,與應用實例 |
| 第 11 週 | 期中考 |
| 第 12 週 | Reinforcement Learning Modeling, Simulation, and Control 強化學習建模,模擬,與控制 >Markov Decision Processes ; >Dynamic Programming ; >Monte-Carlo method ; >Temporal Difference Method ; >Actor-Critic method ; >Policy Gradient Method; ... |
| 第 13 週 | Reinforcement Learning Modeling, Simulation, and Control 強化學習建模,模擬,與控制-應用實例: 多軸組態無人飛行器之強化學習系統建模,模擬,與控制 Reinforcement Learning for Multi-Rotor Configuration UAV |
| 第 14 週 | Reinforcement Learning Modeling, Simulation, and Control 強化學習建模,模擬,與控制-應用實例: 基於強化學習軌跡生成的無人飛行器全方位自主垂直棲息系統建模,模擬,與控制 Omni-Directional Autonomous Perching of Unmanned Aerial Vehicle using Reinforcement Learning Trajectory Generation, Modeling, and Simulation |
| 第 15 週 | Flight System Modeling, Simulation, and Control 飛行系統建模,模擬,與控制 |
| 第 16 週 | 學期考試 |
David L. Smith., ”Introduction to Dynamic Systems Modeling for Design”, Prentice Hall. Simon Haykin. "Neural Networks and Learning Machines", Pearson. StevenL. Brunton, J. Nathan Kutz, "Data-Driven Science and Engineering machine Learning, Dynamical Systems, and Control", 2nd edition, Cambridge
- 地點
- EE448
- 聯絡方式
- chengstone@nycu.edu.tw