| Extreme winds such as hurricanes and tornadoes can be extremely destructive and result in catastrophic property losses and the tragic loss of human lives. The need to predict damage and reduce the loss of life and property is becoming more and more important with every passing windstorm.; Artificial Neural Networks (ANN) provide a novel approach for representing the wind-induced damage prediction model. Modeled loosely after the biological neural networks of the human brain, ANN are generally used in situations where the interactions between the input and the output variable is too complicated for an analytical solution or where there is an insufficient understanding of the problem domain. Predicting wind-induced damage, however, is not a simple task due to the complex, subjective nature, and limited understanding, of the wind effects on buildings in extreme winds. This research concentrates on the investigation of the applicability of neural networks to wind-induced damage prediction, as well as the corresponding implementation issues. Even after years of post disaster windstorm damage investigations, consistent, complete and robust damage information is not available to train the ANN. Thus, synthetic data instead of real building damage information is being used. WIND-RITE{dollar}spcircler{dollar}, a knowledge based expert system for grading individual buildings in windstorms, is being used to provide the necessary damage information for the synthetic data.; This research shows that a feed forward multi-layer perceptron network with a backpropagation learning algorithm can be used effectively to model wind-induced damage predictions for residential buildings. As few as four hundred residential building samples are sufficient to train the network to learn the underlying relationships between the features of the building and its corresponding building damage grade. During training, the ANN model is able to learn the relationships between the input features and the resulting building damage grade effectively. It was also discovered that the ANN model is able to predict reasonably for samples it has yet to encounter.; The approach presented in this work can be used effectively for other building categories. In addition, when sufficient real wind-induced building damage information is available, this approach of using ANN will give a more realistic representation of the relationships between the building characteristics and the resulting wind damage. |
