Keywords: architectural design, knowledge model, soft computing, artificial neural networks, genetic algorithm.

Architecture/engineering is an applied science where many lessons can be learned from post-occupancy evaluation of existing structures. The design engineers/architects should be able to derive from each previous design some qualitative values, especially on users approval regarding building's quality as to assure a successful design. During this process, they are quite often challenged with linguistic qualitative soft data needing to be interpreted and integrated into their design decision-making processes. They should know much about their customer's desires, requirements, and especially customer's preferences when it comes to specific design issues. The post-occupancy evaluation has gained great importance to form an extensive knowledge base for proper use in future projects. This very much holds true in residential construction industry where customers have various preferences as far as safety and comfort issues are concerned.

This paper demonstrates an Artificial Neural Network (ANN) and Genetic Algorithm (GA) based knowledge model of customer's preferences regarding comfort and safety issues in a large residential multi-storey flat housing scheme.

For the study a 320 flat residential housing consisting of total 20 four-story buildings in Poona, India was taken into consideration. Post occupancy evaluation was performed, in which the existing residents were asked different questions regarding safety and comfort of the structure, and their ideal expectations for each of them. A questionnaire consisting of total 45 questions regarding comfort and safety issues of the existing building and residents' opinion about the same and their ideal expectations were collected. The data in the form of a structured questionnaire regarding comfort and safety issues was collected. A five-point scale was used to depict the range of importance from least to most for each issue.

A General Regression Neural Networks (GRNN) model was trained and evaluated in order to determine the best representative response for each question. The questionnaire dealing with various issues related to the safety and comfort were grouped into various grouping for GA optimization, and creating various scenarios to improve safety and comfort for the studied housing complex.

1

S. Durmisevic, O. Ciftcioglu and S. Sariyildiz, "Knowledge modelling of soft data in architectural design", proceedings of the CIB-W78 International Conference IT in Construction in Africa, Section 14, 1-13, 2001.

2

N. Evaboumwan and C.J. Anumba, "Towards a concurrent engineering model for design and build projects", The structural engineer, 74, 73-75, 1996.

3

M. Haque, V. Karandikar, "A Study on Comfort and Safety in a Residential Housing Complex: A Genetic Algorithm Knowledge Model", CD- Proceedings of the Second International Conference on Information Systems in Engineering and Construction (ISEC-02), Florida, 2002.

4

"NeuroShell 2 User's Manual", Ward Systems Group, Inc., Frederick, MD, USA, 1996.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £82 +P&P)