Construction-Site Layout Using Annealed Neural Network

Abstract

Construction-site layout is an important construction planning activity. The impact of good layout practices on money and timesaving becomes more obvious on larger construction projects. In this study, we formulate the problem as a combinatorial optimization problem. Construction-site layout is delimited as the design problem of arranging a set of predetermined facilities on a set of predetermined sites, while satisfying a set of constraints and optimizing an objective. In this paper, the annealed neural network model, which merges many features of simulated annealing and the Hopfield neural network is employed to solve the problem, and a program written in C, called SitePlan, is built on a personal computer to implement the algorithm. In addition, a strategy to set a reasonable initial temperature in the simulated annealing procedure is proposed, the effects of various parameters in annealed neural network are examined, and two case studies are used to illustrate the practical applications and to demonstrate this model's efficiency in solving the construction-site layout problem.