MACHINE LEARNING METHODOLOGIES TO ASSES DEBRIS EXTENT AFTER EARTHQUAKE

Each significant seismic event is an opportunity to assess the performance of our built environment. After such events, a large quantity of varied information is often collected within a short time period. Therefore, a quick and reliable method for detecting damage to buildings is required for disaster response with the intent of reducing both human and economic losses. In the last few years, the increasing advancement of technologies and Computer Science has permitted the adoption of Machine Learning techniques in various fields, included Civil Engineering and Urban Resilience. One of the most influence parameters in the evaluation of seismic loss is the extension of the debris and their associated effects on critical infrastructure. Debris accumulation can result in disruption of the road network and compromise rescue operations. This implies an overall increase in the average number of people who have difficulty evacuating, with a significant risk that some residents cannot evacuate at all. Starting from the newest Machine Learning algorithms and using Python programming language, the purpose of this study is to collect different features and data from the main seismic events that occurred in the past several decades and use them to forecast the extension of building-rubble after earthquakes. This type of model would ultimately allow civil protection agencies to plan their rescue operations while reducing the risk of getting stuck by extended rubble.