Deconstructability prediction for building using machine learning and ensemble feature selection techniques.

Area of Science:

• Construction Management
• Sustainable Engineering
• Artificial Intelligence in Civil Engineering

Background:

• The construction industry is a major consumer of resources and generator of waste globally.
• Circular economy principles are gaining traction to improve resource efficiency and unlock economic value through material reuse.
• Building deconstruction, the careful disassembly for component reuse, aligns with circular economy goals but requires efficient assessment.

Purpose of the Study:

• To address the limitations of manual building deconstruction assessments (time-consuming and costly).
• To develop a machine learning-based predictive model for assessing building deconstructability.
• To demonstrate the practical application of the developed model in a real-world deconstruction project.

Main Methods:

• Development of a machine learning model specifically designed for predicting building deconstructability.
• Utilization of ensemble feature selection techniques to identify key factors influencing deconstruction potential.
• Validation of the model's performance through its application in a case study of a building deconstruction project.

Main Results:

• Successful creation of a deconstructability predictive model using machine learning.
• Demonstration of the model's applicability and potential in a practical deconstruction scenario.
• The model offers a more efficient alternative to traditional manual inspection methods.

Conclusions:

• The developed machine learning model provides an effective solution for assessing building deconstructability.
• This predictive tool supports the adoption of circular economy strategies in the construction sector.
• The research facilitates greater resource efficiency and economic value recovery from buildings at end-of-life.