A GIS-based multi-criteria seismic vulnerability assessment using the integration of granular computing rule extraction and artificial neural networks

This study proposes multi-criteria group decision-making to address seismic physical vulnerability assessment. Granular computing rule extraction is combined with a feed forward artificial neural network to form a classifier capable of training a neural network on the basis of the rules provided by granular computing. It provides a transparent structure despite the traditional multi-layer neural networks. It also allows the classifier to be applied on a set of rules for each incoming pattern. Drawbacks of original granular computing (GrC) are covered, where some input patterns remained unclassified. The study was applied to classify seismic vulnerability of the statistical units of the city of Tehran, Iran. Slope, seismic intensity, height and age of the buildings were effective parameters. Experts ranked 150 randomly selected sample statistical units with respect to their degree of seismic physical vulnerability. Inconsistency of the experts' judgments was investigated using the induced ordered weighted averaging (IOWA) operator. Fifty-five classification rules were extracted on which a neural network was based. An overall accuracy of 88%, κ = 0.85 and R2 = 0.89 was achieved. A comparison with previously implemented methodologies proved the proposed method to be the most accurate solution to the seismic physical vulnerability of Tehran.