Research on the Application of Machine Learning in Predictive Maintenance of Building Structures

With the acceleration of urbanization, the number of large buildings has been increasing, which in turn has led to structural issues such as vibrations and settlements. These could potentially serve as safety hazards. Traditional building maintenance methods predominantly rely on periodic inspections, which might result in resource wastage and the inability to detect certain issues promptly. Machine learning (ML), with its capability to extract patterns from data and predict structural problems, offers new opportunities for predictive maintenance (PdM). This paper discusses three methods of PdM and delves deeply into ML and deep learning (DL) techniques, such as Logistic Regression (LR), Support Vector Machine (SVM), Decision Tree (DT), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN). These techniques provide powerful tools for building health monitoring, not only enabling real-time surveillance of various states and potential problems of the building but also forecasting possible future damages or failures by analysing historical data. Compared to traditional methods, ML offers enhanced accuracy and predictive capabilities, significantly reducing maintenance costs and time. However, the integration of these techniques also introduces new challenges, such as the complexity of data collection, processing, and analysis, as well as model training and optimization. This paper aims to deeply explore how ML can play a role in PdM of building structures, its potential advantages, and challenges that might be encountered in practical applications.