Indian Journal of Science and Technology

Abstract

Objectives: The main purpose of this study is to introduce modern technique to using artificial neural network for predicting the cost of structure works for highway project at the feasible study phase. Methods: Multi-layer perceptron trainings utilized back-propagation algorithm was used. In this study, the feasibility of ANNs approach for modeling these cost characters was inspected. A lot of problem in relation to ANNs construction such as internal parameters and the effect of ANNs geometry on the performance of ANNs models were inspected. Information on the relative importance of the variable’s affecting on the cost parameters predictions was given and mathematical equations in order to estimating the cost of structure works for highway project were determined. Findings: One model was developed for the prediction the structure works cost of highway project. Data and information utilized in this model was collected from Stat Commission for Roads and Bridges in republic of Iraq. ANNs model have the ability to predict the cost for structure works for highway project with very good degree of accuracy equal to 93.19% and the coefficient of correlation (R) was 90.026%, Applications: Neural network has shows to be a promising approach for use in the initial phase of highway projects when typically only a limited or minus data and incompleted information set is ready for cost analysis.

Keywords: Artificial Neural Network, Cost of Structure Works, Coefficient of Correlation, Highway Project, Predicating