Machine learning applications in predicting structural failures and earthquake damage

Abstract

Living beings have faced natural disasters throughout history, shaping evolution and life itself. Though prediction remains challenging, technological advances have improved accuracy. As human structures grow taller and more complex, their strength is constantly tested by nature’s powerful forces. Amongst many examples of powerful forces of nature, earthquake is perhaps the most dangerous due to its devastating impact, unpredictability and scale. Populated countries like Bangladesh, India, China, Japan etc. are all extremely vulnerable to earthquake damage and mass devastation. Bangladesh combined with low elevation, population density, building/structural density etc. is especially vulnerable to structural failures and earthquake damage. Hence, the best way to prevent such disasters is to sort out points of weakness in buildings to predict potential structural failures to save lives and properties. Earth quake damage is often brutally devastating by causing mass destruction, death and huge financial damage. This research paper explores how machine learning can help with the prediction of structural failures and earthquakes using various sensors like seismograph, tilt meter etc. These sensors can gather data which can be processed by LSTM or Long Short-Term Memory machine learning algorithm to give output via EEWS and also provide real-time monitoring. This qualitative research explores the effectiveness of machine learning in terms of structural failure prediction and earthquake damage for a densely populated country like Bangladesh.