共同住宅 프로젝트의 初期 工事費 豫測을 위한 神經網 學習에 遺傳子 알고리즘을 適用한 모델에 관한 硏究

The purpose of this study was to propose a model of neural networks training by genetic algorithms for predicting preliminary cost estimates about apartment projects at the early project stage. In previous studies, neural network model is superior to the regression model in cost estimation. But it has a problem that could be trapping local minima. Therefore, this study applied genetic algorithms to train the weights of neural networks. The result of the research revealed that training neural networks using genetic algorithms is effective in estimating the preliminary costs of apartment projects at the early project stage.