Loading...

A Machine Learning Model for Simplified Damage and Loss Assessment of the Buildings After an Earthquake

Mohammadian, Mohammad Reza | 2020

1099 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53403 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Mohtasham Dolatshahi, Kiarash
  7. Abstract:
  8. This study presents a Machine Learning framework for post-earthquake structural damage state and economic loss estimation. The framework is proposed to rapidly determine whether a structure is safe or unsafe to be occupied following a seismic event, to predict the detailed structural response, and finally to evaluate the economic loss of the buildings. An extensive database is developed by analyzing twelve low-rise, mid-rise, and high-rise eccentrically braced frames (EBFs) under 44 scaled ground motion records through incremental dynamic analysis (IDA) to train the Machine Learning algorithms. Machine Learning algorithms are employed to classify the structural damage state of the buildings and predict engineering demand parameters (EDPs) including peak story drift ratio (SDR), peak floor acceleration (PFA), and peak shear-link rotation patterns. The predicted response demands are used to compute the probabilistic damage state of each component in each story and the economic loss associated to each specific seismic event. In order to reach reliable predictive models, Support Vector Machine (SVM) and Artificial Neural Network (ANN) are used for classification and regression, respectively. Coefficient of determination (R2) of the predictive models for engineering demand parameters are between 49 to 92 percent, and the accuracy of classification for structural damage state are between 77 and 88 Percent. A case study on one of the structures is finally conducted to clarify the application of the framework. The results especially show that the proposed framework can be used as a simplified and accurate model for the loss estimation of the buildings right after the earthquakes.
  9. Keywords:
  10. Seismic Damage ; Machine Learning ; Post-disaster Inspection ; Structural Failure ; Peak Floor Acceleration ; Eccentrically Braced Frame (EBF) ; Story Drift ; Story Drift Ratio Pattern Prediction

 Digital Object List