Journal Article

Experimental Study on Evaluation of Organization Collaboration in Prefabricated Building Construction

Lian, Jinlai, Chen, Kaizhe, Bao, Dingjing, Wan, Shuai, Lei, Ying, Chen, Yuan

2025

Overview

Prefabricated buildings have become important in the transformation and upgrading of the construction industry due to their advantages, including high efficiency, energy conservation, low cost, and environmental friendliness. To further promote the wide application of prefabricated construction, the improvement of construction organization design has become an urgent problem to be solved. Therefore, this study developed a new evaluation method for prefabricated construction collaboration. The proposed evaluation system was built based on the combination of knowledge- and data-driven approaches, i.e., a dual-driven evaluation method. The knowledge-driven part of this evaluation system used an evaluation model based on the analytic hierarchy process (AHP), while the data-driven part used a prediction model based on the BO-XGBoost algorithm to verify the validity of the AHP-based model. To demonstrate the effectiveness of the proposed dual-driven evaluation system, we conducted a case analysis using the data of 204 construction cases obtained from digital simulation platform experiments. The results of the AHP-based evaluation model showed that there was a significant disparity in construction collaboration levels in this case study, with a large proportion of low-level collaboration cases. This indicated that there was a lack of proper collaboration in project management, component production, and on-site assembly, reflecting the urgent need for improvement in collaboration efficiency. Regarding the data-driven analysis, the BO-XGBoost prediction model was built based on the AHP-based evaluation results. It was found that the prediction accuracy of the BO-XGBoost model was as high as 98.1%, indicating that the proposed AHP-based model was scientific and effective. Moreover, the BO-XGBoost model was compared with the random forest, support vector machine, and logistic regression prediction models. The BO-XGBoost model outperformed the other three prediction models in terms of accuracy, precision, recall rate, and F1 score. The proposed dual-driven evaluation system provided a new perspective for the scientific evaluation of prefabricated construction collaboration. The findings of this study contributed to enhancing the project management optimization capability of smart construction sites.

• Publisher: MDPI AG
• Subject: Accuracy / Algorithms / Analytic hierarchy process / Artificial intelligence / Building information modeling / Buildings / Case studies / China / Collaboration / Construction / construction collaboration evaluation / Construction industry / Construction sites / Data analysis / Decision making / Deep learning / Digital simulation / digital simulation platform / Effectiveness / Energy conservation / Evaluation / Knowledge / Knowledge management / Logic / Machine learning / Methods / Neural networks / Optimization / Performance evaluation / Prediction models / Prefabricated buildings / prefabrication construction / Project management / Regression analysis / Support vector machines / XGBoost