Machine learning-based prediction for maximum displacement of seismic isolation systems

This study developed machine learning (ML) models for predicting the peak lateral displacements of seismic isolation systems subjected to earthquakes. Six ML methods were employed for this purpose: linear, ridge, least absolute shrinkage and selection operator (lasso), artificial neural network (ANN), support vector machine, and random forest (RF). The isolated superstructure was assumed to be rigid, and the base isolation system was assumed to be firmly anchored to a bedrock foundation. A total of 234,360 data points were generated using OpenSees to train and test the predictive models. The input variables included three parameters representing the behavior of isolation systems (i.e., the ratio of the initial to post-yield stiffness, normalized characteristic strength, and post-yield period) and the average spectral accelerations at five selected periods (i.e., 1, 2, 3, 4, and 5 s) for the considered ground motions. The RF model exhibited the best performance among the six models. The coefficients of determination (R-2) for the RF model were 0.9930 and 0.9498 for the training and testing datasets, respectively. In addition, the ANN model exhibited the second-best performance. The normalized characteristic strength was found to be the most influential variable in the investigation of the individual input variables for the prediction. Finally, a practical analysis of a hypothetical three-story base-isolated moment frame building was performed to demonstrate the effectiveness and limitations of the developed RF model in predicting the maximum displacements of the isolation systems. The prediction results were compared with those from a previous study and with the equivalent linear force procedure in American Society of Civil Engineers (ASCE) 7-16. A graphical user interface was created for the developed RF model for easy access by engineers. This study represents a pioneering step for the application of ML to the estimation of the responses of seismic isolation systems in practical design.

Automated summary

This study developed machine learning models to predict the peak lateral displacements of seismic isolation systems. The random forest model showed the best performance, and the normalized characteristic strength was found to be the most influential variable. Furthermore, a practical analysis and comparison demonstrated the effectiveness and limitations of the developed model.