Predicting the required duration for construction activities using Artificial Neural Networks

• Main Author: Golizadeh, Hamed
• Format: Thesis
• Language: English
• Published: 2013
• Subjects: TH Building construction
• Online Access: http://eprints.utm.my/id/eprint/33824/5/HamedGolizadehMFKA2013.pdf; http://eprints.utm.my/id/eprint/33824/; http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:70128?site_name=Restricted Repository
• Institution: Universiti Teknologi Malaysia

The duration of a construction project is a key factor to consider before the project starts, as it can determine the success or failure of the project. Difficulties in estimating the duration of activities that can also lead to error if manually estimate it. The main purpose of this study is to develop a model to estimate the duration of construction’s major activities in the structural part of concrete frame of buildings. In this study, available methods and models have been investigated and this is achieved through reviewing the previous literatures. It is argued that using Artificial Neural Network (ANN) is the most proper method to achieve the aim of this study. Consequently, through literature investigation and experts interviewing, those factors which can critically influence the activity duration have been opted. Four different buildings in two different regions of Malaysia are selected as case for the project. Finally, the collected data and variables implemented into the models and nine ANN models have been trained, tested and validated. Contractors and firms can utilize these models in the planning phase of their project to avoid the errors made by human beings and producing more accurate estimations of activity durations.