Modeling Tower Crane Operations Using Group Method of Data Handling and Genetic Algorithms
Because of the existence of multi-variables in a versatile site environment, construction activities are difficult to predict and model. The use of traditional mathematical models to portray the linear and non-linear relationship between variables and outcome fails to optimize the number of inputs, resulting in under-or over-fitting the data. Overfitting can occur when the predicted field is redundant Under-fitting occurs when the resulting model fails to match patterns of interest in the data. The approach of this paper is to integrate the Group Method of Data Handling (GMDH) which helps in screening out the variables and the Genetic Algorithm (GA) that serves to optimize the prediction model. The productivity data collected from tower crane operations for high-rise building construction was used to demonstrate the superiority of the approach. It shows that the GA optimization aided with polynomial model developed from GMDH performs better than the Artificial Neural Networks (ANN) model in term of the number of trials required reaching the near-optimal solution. It is believed that the continuous nature of the polynomial model facilitates the GA search process to reach the optimal solution earlier.
CM. Tarn K. K. Chan
Department of Building & Construction, City University of Hong Kong. Hong Kong
国际会议
2003 International Conference on Construction & Real Estate Management(2003 建设与房地产管理国际会议)
哈尔滨
英文
292-295
2003-11-20（万方平台首次上网日期，不代表论文的发表时间）