ETD

• machine learning algorithms
• neural network
• regression
• seismic risk prediction
• structural engineering

Over half of the world's population lives in cities, which is expected to rise to 68% by 2050. With urban areas overgrowing worldwide, the need to accurately predict seismic risks for buildings becomes increasingly urgent. This thesis combines tools from structural engineering with artificial intelligence to enhance our understanding of earthquake impacts on buildings, forecasting the seismic risk and potential damages in response to earthquakes. We first used a finite element method (FEM) tool to calculate seismic risks. However, FEM tools are time-consuming, requiring several hours to analyze only one building. The primary objective was thus to substitute the FEM with machine learning algorithms to achieve comparable results more quickly. A machine learning-based regressor was developed, and its performance was evaluated through error analysis, facilitating the identification of potential inputs influencing regressor errors. In addition, a web application was designed and developed to visualize earthquake effects on surrounding buildings, integrating the results of seismic risk prediction. This tool offers an intuitive interface for urban planners and engineers to assess seismic risk and plan infrastructure resilience effectively. This study shows the effectiveness of the artificial intelligence-based approach in seismic risk analysis and building damage prediction, providing a more efficient and accessible solution for seismic risk management in urban areas.