Developing A Mathematical Model for Planning Repetitive Construction Projects By Using Support Vector Machine Technique

Civil and Environmental Engineering

Hatem Wadhah Amer¹  Erzaij Kadhim Raheim²  Burhan Abbas M.² 

[1] Middle Technical University, Baquba Technical Institute, Baghdad, Iraq.;University of Baghdad, Civil Engineering Department, Construction Project Management, Baghdad, Iraq.;

关键词: project management;    repetitive constriction project;    support vector machine;    planning and scheduling;   

DOI  :  10.2478/cee-2021-0039

来源: DOAJ

【 摘 要 】

Each project management system aims to complete the project within its identified objectives: budget, time, and quality. It is achieving the project within the defined deadline that required careful scheduling, that be attained early. Due to the nature of unique repetitive construction projects, time contingency and project uncertainty are necessary for accurate scheduling. It should be integrated and flexible to accommodate the changes without adversely affecting the construction project’s total completion time. Repetitive planning and scheduling methods are more effective and essential. However, they need continuous development because of the evolution of execution methods, essentially based on the repetitive construction projects’ composition of identical production units. This study develops a mathematical model to forecast repetitive construction projects using the Support Vector Machine (SVM) technique. The software (WEKA 3.9.1©2016) has been used in the process of developing the mathematical model. The number of factors affecting the planning and scheduling of the repetitive projects has been identified through a questionnaire that analyzed its results using SPSS V22 software. Three accuracy measurements, correlation coefficient (R), Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), were used to check the mathematical model and to compare the actual values with predicted values. The results showed that the SVM technique was more precise than those calculated by the conventional methods and was found the best generalization with R 97 %, MAE 3.6 %, and RMSE 7 %.

【 授权许可】

Unknown