A good construction-site layout plan could directly or indirectly save construction cost and improve on construction efficiency; the larger of construction project or higher of construction overlapping is, the more important of construction-site layout is plan. Moreover, today?s constructive automatic technique reaches a specific degree; the application of automatic technique in construction-site layout plan has been accentuated. In general, the requirements of construction-site facilities depend on the size and type of construction project. However, a traditional view for arranging construction-site facilities by heuristic mostly causes a disappointing layout plan and makes uncompleted judgement. In this paper, it would formulate construction-site layout problem to combinatorial optimization one. Because it?s problematic model of mathematics is NP-complete problem, it would try to adopt self-learning neural network to solve construction-site layout problem. In order to test it's reliability; it will compare quality-efficiency with Random-searching and Annealing Neural Network method.