علي خزائي

Accurate estimation of the duration of construction activities is a key factor in the success of construction projects. According to studies, to date, various methods have been used to estimate the duration, each of which, despite its various advantages, has been associated with challenges. This study was conducted with the aim of creating a tool to improve the estimated duration of the main activities of the construction industry (focusing on the concreting activity of slabs). To achieve this goal, in this study, artificial neural networks (ANN) have been used as the core of computing and modeling. For this purpose, through literature review and interviews with experts, factors that can specifically affect the duration of concrete slab activity have been identified. In the next step, based on the combination of factors, quantitative data related to concreting of three different types of slabs have been collected from contracting companies active in this field in Iran. Then, using the collected data, four different neural networks (single-layer MLP networks, two-layer MLP, RBF and GMDH) were designed, trained and tested to estimate the concreting time of three types of slabs. Finally, using the ensemble learning method, the results of the designed networks are combined to increase the accuracy of the estimation. Also, in order to better compare the proposed method, the results were eva‎luated with the estimation results of PERT scheduling method. The result of this method showed that the error rate of the proposed method is about 3% and also compared to previous works and PERT method has a better performance and significantly reduces the error in estimating the duration of activity.

تخمين , فعاليت هاي صنعت ساخت , شبكه عصبي مصنوعي , يادگيري گروهي

estimating , activities of the construction industry , artificial neural networks , ensemble learning

Ali Khazaee

Dr. Parviz Ghoddousi