Bayesian network for earthquake damage risk modeling and management

Many researches have been done for earthquake forecast. However, a risk management model is also needed for estimating earthquake damage so that governments and the public can be prepared for prevention. This dissertation introduces a method of establishing prediction models for earthquake damage based on Bayesian Network. By collecting useful information and data of historical earthquakes, the BN model structure can be built from prior knowledge. Then the BN model is trained for decision making and prediction and the conditional probability tables are determined. When new earthquake occurs, the prediction model can be used to roughly forecast the overall damage in fatalities and economic losses once the data of earthquake is available. Overall, this method of modelling and risk prediction is reliable and valuable in earthquake damage estimation. Key word: Modelling; Bayesian Network; Earthquake Damage Prediction.Master of Science (Computer Control and Automation