A Short Course on
Applied Evolutionary Optimization
Ohio Aerospace Institute (OAI) Cleveland, USA
15--18 June 2009
Aim of the Course:
Multidisciplinary design and optimization has become inevitable in obtaining cost-effective or high-performing design solutions in many engineering activities, due to the need of considering multiple disciplines each having its own multiple objectives, constraints and decision variables. The practical problems require handling different complexities, such as uncertainties in decision variables and parameters, nonlinear constraints, dynamic nature of problems, computationally demanding evaluation schemes. Although optimization methodologies using classical mathematically oriented methods have been around for the past 50 years or more, they have been mainly limited due to their inflexibilities in handling different complexities associated with practical multidisciplinary considerations mentioned above. On the other hand, optimization methodologies based on evolutionary principles, proposed in early sixties but gained popularity in practice since nineties, demonstrate tremendous potential to handle most of the practical complexities.
In this short course, we shall first introduce different optimization problem solving tasks that are commonly found in multidisciplinary analysis and design optimization problems. Thereafter, we shall briefly discuss some theoretical foundations of optimal solutions and then introduce basic solution principles using classical means. This will then provide a foundation for the need of more flexible optimization methodologies. Evolutionary optimization methodologies, starting with a genetic algorithm (GA), will then be introduced. Variations to the basic GA principles will then be discussed one by one to demonstrate the flexibility of GA methodologies for handling different complexities of optimization problems. Specifically, nonlinear constraint handling, meta-modeling techniques, large-scale problem solving, multi-modal and multi-objective optimization, robust and reliability-based optimization methodologies will be discussed in detail.
Schedule of the Course:
All presentation materials can be obtained by clicking on 'Lecture X'. Further reading materials, references to books and softwares are also provided. At the misc. hour, tutorials are assigned for the participants to solve by hand. The instructor of the course will be available for assistance.
| Date (Day) | 1st Session (1.5 hours) | 2nd Session (1.5 Hours) | 3rd Session (1.5 Hours) | Misc. (1 hour) |
|
15 June'09 (Monday) |
Introduction to Applied Optimization
|
Classical Optimization |
Introduction to Genetic Algorithms |
|
|
16 June'09 (Tuesday) |
Constraint Handling, Theoretical Foundation |
Real Parameter EAs EA code demonstration |
EAs for Complex Optimization Tasks Further Reading 1, 2 |
|
|
17 June'09 (Wednesday) |
Multi-Objective Optimization |
Advanced Topics in EMO |
Aerostructure Design (Meng-Sing Liou) |
NSGA-II code demonstration
|
|
18 June'09 (Thursday) |
EMO and Decision Making
|
Dynamic and Bilevel Optimization |
Recent Issues in EMO |
Wrap-up and Discussions |
For more papers on the above topics, please visit and download papers from http://www.iitk.ac.in/kangal/pub.htm
For codes implementing GAs and EMO algorithms (in C and C++), please visit and download codes from http://www.iitk.ac.in/kangal/soft.htm
Conduct of the Course:
Course Instructors:
Kalyanmoy Deb
Deva Raj Chair Professor
Department of Mechanical Engineering
Indian Institute of Technology Kanpur
PIN 028016, India
Email: deb@iitk.ac.in
URL: http://www.iitk.ac.in/kangal/deb.htm
Dr. Meng-Sing L:iou will deliver a lecture on applications of EAs to propulsion systems to aerospace vehicles.