An integrated risk assessment and prediction framework for fire ignition sources in smart-green multi-unit residential buildings

Abstract

Fire risk in smart-green multi-unit residential buildings (SG-MURBs) is found to be causing a significant threat to public safety, property, and the environment. Therefore, this study developed a framework to assess and predict the potential ignition source factors that cause fire incidents (ISFs-FIs) in SG-MURBs. The framework consists of four steps: identification of potential ISFs-FIs; benchmarking; development of ANN model to predict ignition sources-related human factors and fire origin; and development of ANN model for fire impacts prediction. The ANN Simulink models were generated to simulate ANN models in Simulink or deploy with Simulink Coder tools using MATLAB. The developed framework was applied to seven cities in British Columbia (BC). The results identified forty key potential ISFs-FIs. The levels of relative frequency, total dollar loss, and risk of ISFs-FIs were determined and then benchmarked to identify the most critical factors. The results indicated that Vancouver, BC, has a very high fire risk level with 58%. Statistically significant (p < 0.005) correlation was found with a negative correlation (R = − 0.83) between elderly people and fire frequency, whereas a positive correlation (R = − 0.87) between younger people and fire frequency. Finally, two ANN prediction models are developed with R2 predictive abilities of 72% and 99%, respectively. These results assist decision-makers in enhancing fire prevention strategies accordingly.