defter*
defter / katalog / ME 518
ME 518

Data-Based Control Systems

Classical control starts from a known model of the plant; this course is about what to do when you don't have one and have to let measured data fill that gap. You'll spend the term identifying input-output and state-space models from noisy experiments, compressing them with DMD/DMDc, and wrapping the result in MPC or reinforcement-learning controllers — typically in Python and MATLAB, building toward a project. It sits at the intersection of signals, optimization, and ML, and is increasingly how robotics, process, and aerospace groups close the loop when first-principles modeling runs out.

Credit3ECTS5FacultyFaculty of EngineeringBölümMechanical Engineering

Değerlendirme 100% — 5 adım

20%
30%
5%
25%
20%
Midterm Midterm 20%
Final Final 30%
In-class attendance Attendance 5%
Project Term project 25%
Homework HW - evaluated via quizzes 20%

Önerilen kaynaklar 1 kitap

📖
Önerilen
Data Driven Science & Engineering Machine Learning
Dynamical Systems, and Control
Steven L. Brunton · J. Nathan Kutz

Haftalık müfredat 14 hafta

Hafta 1
Motivation, review of classical and state-feedback control
Hafta 2
Measurement, noise models, data preprocessing
Hafta 3
System identification: FIR/ARX, least squares
Hafta 4
System identification: N4SID
Hafta 5
Model validation
Hafta 6
DMD fundamentals
Hafta 7
DMD with control, DMD-C
Hafta 8
Model Predictive Control (MPC) fundamentals
Hafta 9
Data-driven MPC implementation (Python, Matlab)
Hafta 10
Robustness and stability of data-driven controllers
Hafta 11
Reinforcement Learning fundamentals
Hafta 12
Reinforcement learning examples, MPC vs RL
Hafta 13
Case studies
Hafta 14
Project presentations

Ders notları — henüz yok

ME 518 için defter ekibi henüz not yazmadı.

İlk dosyayı sen atarsan — not, slayt, geçmiş sınav, çözüm, cheat-sheet, ne varsa — defter ekibi öğrenci paylaşımlarından bu dersin notlarını yazar. Drive linki / PDF / ZIP, hepsi olur.

← katalog

⚠️ FZ engelleyen şartlar

Course Learning Outcomes: Course Learning Outcome Assessment Acquire and preprocess experimental data for control design Midterm Final Term project HW - evaluated via quizzes Construct and validate linear input output and state space models using system identification techniques Midterm Final Term project HW - evaluated via quizzes Extract reduced order predictive models with Dynamic Mode Decomposition (DMD/DMDc). Midterm Final Term project HW - evaluated via quizzes Formulate and implement Mode

Hocalar 0 bu dönem · 1 geçmiş

Geçmişte ders veren (1 kişi)
Yıldıray Yıldız