| Summary: | 碩士 === 國立高雄應用科技大學 === 土木工程與防災科技研究所 === 103 === Construction projects are getting more complex and bigger day by day, also the workload is increased. If early planning is done more properly, it would avoid unnecessary project delays and the cost over-run, make the construction phase more smooth. For construction projects, implementation of early planning is very important.
This research investigate the relationship of project pre-planning efforts and project performance. 105 domestic construction projects data are collected. By using the artificial intelligence methods to establish relevant models and analysis, PDRI scores are set as input variables for each model, the output variable is project performance (schedule, costs). Artificial Intelligence Models are trained to predict construction project performance. The results show that the predicted building project "cost" performance for adaptive neural fuzzy inference model is slightly better than the support vector machine and the neural network. Models the prediction accuracies (mean absolute percentage error) for ANNs, SVMs, and ANFIS models are 4.78%, 4.84%, and 4.96% respectively. The predicted building project "schedule" performance for adaptive neural fuzzy inference system model is slightly better than the support vector machine and the neural network. Models the prediction accuracies (mean absolute percentage error) for ANNs, SVMs, and ANFIS models are 6.32%, 6.37%, and 7.14% respectively.
The research found that adaptive neural fuzzy inference system can effectively enhance the prediction of the construction project performance using PDRI scores.
|
|---|
