ML-EHSAPP: a prototype for machine learning-based earthquake hazard safety assessment of structures by using a smartphone app

Abstract

The recent devastating earthquakes have caused severe physical, social, and financial damage worldwide and indicate that many existing buildings, especially in developing countries, are not designed to withstand seismic hazards and resulting vulnerabilities to earthquakes. It is difficult, time-consuming, and costly to investigate and inspect all the existing buildings in detail before an earthquake, especially in urban areas. Therefore, rapid methods for evaluating the vulnerability of buildings have attracted the interest of researchers. This paper investigates the seismic susceptibility through the combination of buildings’ geometrical attributes that affect the vulnerability of buildings and further extended into having foresight into the damage state of reinforced concrete (RC) buildings using artificial neural network (ANN). For this purpose, a multi-layer perceptron (MLP) network has been trained and optimized using two different databases of damaged buildings from the Nepal and Ecuador earthquakes. The results show the practicability and efficacy of the selected ANN approach for classifying actual damage grade based on structural damage, which the workflow can be followed and applied into other data from other regions for the detection of highly vulnerable buildings. A prototype of a smartphone app and its implementation was later introduced for data collection and vulnerability assessment purposes based on the proposed method.