Improving the Performance of a Fuzzy Logic Model in Seismic Damage Prediction using a Guided Adaptive Search-based Particle Swarm Optimization ALGORITHM

Abstract

This paper proposes a fuzzy logic model to improve the accuracy of seismic damageability simulations for buildings. The Rapid Visual Screening (RVS) method is often used to evaluate seismic damages in buildings due to its speed and simplicity, but it can be subject to human error and other uncertainties. The proposed model uses fuzzy logic to address these uncertainties and build a more robust simulator for estimating the seismic damage state. To fine-tune the hyperparameters of the fuzzy model, the Guided Adaptive Search-based Particle Swarm Optimization (GuASPSO) algorithm is used, which has been shown to be efficient and effective. The model is applied to simulate the damageability of reinforced concrete buildings damaged in the 2017 Sar-Pol-Zahab earthquake in Iran, and the results are compared to those obtained using two popular meta-heuristic optimizers, the PSO and GWO algorithms. The results demonstrate that the GuASPSO algorithm outperforms the other two in terms of performance metrics in the training, validation, and total data sets. The proposed model is a significant step toward more accurate and practical seismic damageability simulations.

Keywords

• rapid visual screening
• seismic damageability
• reinforced concrete structures
• optimization
• guaspso
• fuzzy logic
• self-organizing map