A Conceptual Framework for Building Cost Estimate using Artificial Intelligence: Convolution Neural Network (CNN) and Wolf Pack Algorithm

Abstract

The accuracy of cost estimates often depends on the availability of information from the development of building drawings. As a result, this leads to limited information for producing accurate cost estimates for building. Typically, the conceptual cost estimates have a wide accuracy range, which means estimated costs can deviate from actual between -50% to +100% When budgeted costs are inaccurate, the construction costs of a project can be severely impacted. Traditional cost estimation methods in construction projects often rely on expert judgment, historical data, and basic mathematical models, which can be subjective, time-consuming, and prone to errors.  Therefore, leveraging on Artificial Intelligence (AI) technologies offers potential to improve the efficiency of cost estimation for construction projects. The aim of the research is to develop an innovative conceptual method for estimating building costs using artificial intelligence. This research adopts a subfield of AI, which are Convolutional Neural Network (CNN) and the Wolfpack Algorithm. The research objective of this paper is to propose a conceptual research framework on AI based cost estimation for construction project and to show that convolutional neural networks are not only suitable for image processing but also for dealing with large amounts of data in quantity surveying accuracy. The research methodology adopted uses an experimental approach where simulation modeling is generated to determine optimal cost estimation for a low-rise building project. A total of 29 residential building and carparks cost data such as estimated and actual costs are input in the model. 20 of the data are used for simulation modelling whereas another 9 data are used for testing the model. The results of this research are expected to break the long-standing perception that convolutional neural networks are only suitable for processing images, and the results show that convolutional neural networks combine Wolfpack Algorithm can reduce the error value of construction cost estimation.