An Artificial Neural Network (ANN) Model Proposal for Cost Minimization and Cost Estimation Based on Building Dimensions for Reinforced Concrete Duplex Villa in Preliminary Design
Loading...
Journal_Current_Construction_Issues_CEPPIS_2020-65-90.pdf (791.63 KB)
Authors
Date
2020-11
Journal Title
Journal ISSN
Volume Title
Publisher
BGJ Consulting
Abstract
The primary aim of this study is to create connections / graphs showing the change of different design parameters and unit and average costs for reinforced concrete duplex houses. In this way, optimum cost-effective designs can be achieved. Another objective is to realize a cost estimation model based on a limited number of design parameters. Such a model will contribute to time and time savings in estimating low error rate cost to the preliminary design phase. Study; A model based on Artificial Neural Network (ANN) has been established for the purpose of preliminary estimation of the construction costs of 115 duplex villas. The data set, which was formed by using the existing data, was entered as data into ANNs structured in single and multi-layer, feed-forward, consultant learning features. These residential structures; basement floor areas, ground floor areas, first floor areas, building total areas, building heights, exterior façade areas, exterior façade areas, number of bathrooms, number of wc, number of kitchens, total wet areas, number of balconies, total balcony areas, rooms numbers and hall numbers were used as main evaluation criteria (input vectors) to the network. The cost values of each structure were used as output vectors. Learning, information storage and generalization features of this method; The performance of cost estimates in terms of proximity to reality is investigated. The solutions found by ANN are compared with Unit Price Based Cost (BFY) and Regression Analysis (RA) methods; error rates of cost estimates were evaluated. According to the results obtained, the estimated values obtained from the ANN created are closer to reality and applicable than the RA data. As a result of the study, it is understood that ANN modeling approach can be used successfully in the prediction stage of the costs of duplex reinforced concrete residential buildings. With the help of graphs showing the relationship between building dimensions and unit costs, optimum values were determined to reach minimum cost. These graphs can be utilized in the preliminary design phase of similar types of structures.
Description
Keywords
Citation
Uğur, L.O., Sağaz, M. (2020). An Artificial Neural Network (ANN) Model Proposal for Cost Minimization and Cost Estimation Based on Building Dimensions for Reinforced Concrete Duplex Villa in Preliminary Design. Journal of Current Construction Issues. CIVIL ENGINEERING PRESENT PROBLEMS, INNOVATIVE SOLUTIONS - Challenges Facing the Construction Industry: 65-90.