Parametric cost estimating of highway projects using neural networks

• Main Author: Samir Ayed, Amr
• Format: Thesis
• Language: English
• Published: Memorial University of Newfoundland 1997
• Subjects: Newfoundland
• Online Access: https://research.library.mun.ca/8011/; https://research.library.mun.ca/8011/1/Ayed_AmrSamir.pdf; https://research.library.mun.ca/8011/3/Ayed_AmrSamir.pdf

Contractors' experience on previous projects can undoubtedly be considered as an important asset that can help preventing mistakes and also increases the chances of success in similar future encounters. Construction cost data collected from past projects may be used to support cost estimating at different stages of a project's life cycle. At early stages of a project, parametric cost estimate is performed when detailed project information is lacking. The usable historical data at this level pertain to the characteristics of past projects (e.g., location, size, complexity), their construction environment (e.g., market, weather, year), in addition to the associated costs spent. The large number of these factors in addition to other external political, environmental, and technological risks, represent a complex problem in establishing accurate cost estimating models and have thus contributed to the inadequacy of traditional cost estimating techniques. -- This thesis uses a non-traditional estimating tool, Neural Networks, to provide an effective cost-data management for highway projects and accordingly develops a realistic cost estimating model. Neural Networks are techniques based on advances in Artificial Intelligence branch of computer science. They have recently been used as a new information management tool in many construction applications to provide an effective cost estimating tool for highway construction cost data. In the present study, the characteristic factors that affect the cost of highway construction have been identified and actual cases of highway and bridge projects constructed in Newfoundland during the past five years have been used as the source of cost data. The structure of a Neural Network template has been formed on a spreadsheet and three different techniques, Backpropagation training, Simplex Optimization and Genetic Algorithms, have been utilized to determine the optimum Neural Networks model. The resulting optimum model has been coded on Microsoft Excel in a user-friendly program to ...