Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Design problem optimization with multi-objective evolutionary algorithms
Complex design challenges involve conflicting objectives and require robust optimization techniques. They commonly arise in engineering, building design, robotics, drug design, and energy systems, among others, where balancing competing criteria is essential. Sunshade optimization is also a complex...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English English |
| Published: |
Department of Computer Science
2026
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | Complex design challenges involve conflicting objectives and require robust optimization techniques. They commonly arise in engineering, building design, robotics, drug design, and energy systems, among others, where balancing competing criteria is essential. Sunshade optimization is also a complex design problem as it has many conflicting objectives. Sunshades significantly influence a building's thermal performance, daylight quality, occupant comfort, and energy usage. However, traditional sunshade designs typically focus on a limited set of objectives, often ignoring broader considerations such as cost efficiency and outside-view obstruction. This thesis addresses that gap by implementing and comparing two advanced multi-objective evolutionary algorithms—Multi-Objective Covariance Matrix Adaptation Evolution Strategy (MOCMA-ES) and the Non-Dominated Sorting Genetic Algorithm II (NSGA-II)—to optimize sunshades across five key objectives: thermal comfort, energy consumption, Useful Daylight Illuminance (UDI), cost, and outside-view obstruction. A single-room office model was used as a test bed, with parameterized sunshades simulated through Honeybee, EnergyPlus, and Radiance. Experiments were conducted in four distinct climate zones—Cape Town (moderate), Nairobi (hot), Colombo (hothumid), and Oslo (cold)—to ensure broad applicability. Both algorithms consistently outperformed traditional, manually designed sunshades in reducing thermal discomfort and energy usage while also improving UDI. Gains in cost and view preservation were more modest, primarily because minimal overhang sunshades can already be inexpensive and unobtrusive. Statistical tests indicated no systematic performance advantage of one algorithm over the other; NSGA-II tended to produce larger Pareto fronts, whereas MOCMA-ES explored a broader range of objective values. The main contribution of this research is the use of two advanced multi-objective evolutionary algorithms to optimize sunshade designs based on five key objectives, tested in four climate zones representing both the northern and southern hemispheres, as well as regions below and above the equator, demonstrating clear advantages over traditional, manually designed sunshades in achieving a balanced trade-off among competing performance criteria. |
|---|