This page is no longer being updated and remains online for informational and historical purposes only. The information is accurate as of the last page update.
For questions about page contents, contact the Communications Division.
Diana Hasegan ’10 (Tirgu Mures, Romania) is pursuing a B.S. in civil engineering and an A.B. in economics and business. She is currently performing research to develop a system which will help architects and engineers work better together with Anne Raich, assistant professor of civil and environmental engineering. The following is a first person account of Hasegan’s experience.
Genetic algorithms, neural networks, multi-objective optimization — If these terms sound unfamiliar to you, then you feel the same way I was feeling four months ago. But now something has changed. During this summer, I was fortunate enough to work with Professor Raich in the civil and environmental engineering department on a research project entitled “User Guided Conceptual Design and Multi-Objective Optimization Using Genetic Algorithms of Large Span Roof Trusses.
I must admit that when I first heard the title of the research I felt a little scared because I did not even know where to start, or in what direction to read it. But I was curious and excited about learning and Professor Raich has done an amazing job guiding my learning process.
The basic principle behind this project is helping engineers to come up with better solutions for the design of large span roof trusses. These are the structural frameworks that support the roof of a building. The civil engineer has to input in the computer program some initial information regarding the expected outcome such as span, height, and load of the roof. Then the program runs the first round of computational methods, which are genetic algorithms, and comes up with a number of feasible and optimal designs for the input parameters. After that, the engineer is asked to input their preferences by selecting specific trusses out of those generated. After this step, the program runs multiple cycles to come up with better solutions for the roof truss design that take into consideration the preferences of the user.
The first step of this project was reading books and research papers to familiarize myself with the concepts and learn the basics. The concept of genetic algorithms, the first concept I learned, was surprisingly simple, yet profound. Genetic algorithms are a series of procedures applied to a set of data that follow the natural process for reproduction of living organisms. A population, or number set, of data is created randomly by the computer. Then the concept of survival of the fittest is applied. Each member (piece of data) from that population is evaluated according to some pre-established criteria and the best members are selected for crossover.
In this next step, the genetic information of the best members is interchanged to create children sets of data. After this, mutation is applied just like in genetics and some information in the data is randomly changed to create diversity in the population. This is an apparently simple concept that we all witness in our everyday life, but in its core structure, it can be applied to solve structural optimization problems, such as the one we are working with.
The second part of the project involved neural networks. This concept was inspired by the human brain and the networks that neurons form when storing information in memory. Just as humans learn colors, shapes, or letters and are able to identify them when presented with an image or a text, a computer algorithm is able to simulate the activity of the brain to recognize the preferences of the engineer with regard to the truss designs.
I found it exciting to be able to read and understand engineering research papers. Even more exciting than that was gaining hands-on experience with programming simple problems as well as implementing changes in the program that we were actually testing.
The summer EXCEL experience was extremely rewarding as it helped me to better understand the demands of the research world and introduced me to a specific domain of conceptual design in civil engineering. My experience on this project has not concluded with the end of the summer, but will continue into the new semester.