Machine-Learning Algorithms in the Service Life Prediction of Facility Management: Approach in Southern Chile

Abstract

Concerning preventive maintenance plans for heritage timber buildings, computational methods are pioneering knowledge for the implementation of new preservation approaches in heritage structure management. In this context, fuzzy logic and random forest methodologies manage both data obtained from professional experts and data obtained in situ from the buildings themselves. This kind of digital procedure can harmonize the outcomes of building assessments because slight variations in the evaluation of input parameters do not produce a significant dispersion over the model’s output. Preventive conservation strategies require cooperation among qualified experts who examine multidisciplinary knowledge related to heritage properties. Thus, new digital protocols and procedures that help decision makers prioritize preventive interventions and avoid corrective actions are paramount in minimizing the irreparable loss of properties. The main aim of this research is a new approach to two computational management systems: fuzzy logic and random forest. The outcomes of this study will be useful to stakeholders who are responsible for the maintenance of heritage buildings, as this methodology reduces the probability of failure and uncertainty during decision-making. The instruments derived will establish mitigation strategies oriented toward proactive future maintenance programs for heritage timber buildings in southern Chile.