A Quantitative Multi-objective Management Model Based On Genetic Algorithms In Construction Project

With the wide application of information technology in construction project management and improvement of management technique, quantitative management of construction project becomes possible. Time, cost and quality are three basic objectives in construction project management. Success of construction project depends on multi-objective management. The three objectives are inconsistent but unified, and they are interacting with each other, as makes it more difficulty to manage project quantitatively. Establishing an effective quantitative multi-objective management model and making trade-off between time, cost and quality will provide decision-maker powerful support.As a search optimal method, genetic algorithms have been applied in construction optimization problems in recent years. The characters of good function of search and high efficiency make genetic algorithms more suitable to solve multi-objective optimization problems in construction project. Genetic multi-objective optimization theory provides several methods to solve such problems, in which Pareto ranking is an effective one. And convergence and equality of solutions is controlled through fitness sharing technique, which makes solutions converge to optimal trade-off surface. Therefore, Pareto optimal solutions are obtained.In this thesis, a more scientific and reasonable quality measurement system is built based on research results about construction project quality. Secondly, according to qualitative research about multi-objective management and research on time-cost optimization in construction project, a time-cost-quality trade-off model based on genetic algorithms and multi-objective theory is built. Then, a computation procedure is designed based on Baker's Pareto ranking method and fitness sharing technique, and the evaluation process is illustrated in detail. At the end, a case study is given to verify the feasibility and practicability of the model. A computer program is designed to calculate solutions, through which the weakness of the model is presented and several aspects of improvement are proposed.