구조물의 동적 거동예측을 위한 다층퍼셉트론 최적화 방법 = Optimized Multilayer Perceptrons to Predict the Dynamic Response of Buildings

다국어 초록 (Multilingual Abstract)

The use of Artificial Intelligence (AI) to predict the behavior of any system is common these days. Use of the technology may be economical in terms of both time and cost. The database methods are robust and reliable. The current study is about the application of AI for predicting the dynamic responses of steel frame buildings. To obtain an optimized AI method, different multi-layer perceptron were trained and tested by varying the hyperparameters like number of nodes, layers, algorithms and functions. A multilayer perceptron with 2 hidden layers and 10 nodes showed the most accurate results in terms of regression and root mean square error.