Keywords: damage evaluation, building vulnerability, AI expert system.

After an earthquake it is necessary to assess as faster as possible the damage degree of buildings that have suffered moderate and severe damages. Some of them can be a hazard for the community and others could be used as temporary shelters for people who have lost their housing or that have been evacuated from unsafe buildings. In order to carry out the damage evaluation process properly it is important to involve surveyors with vast experience in these works and in structural engineering. Nevertheless, when a major earthquake occurs, the damage in the area can be so extended that it is not possible to make all evaluations by experts only. Voluntary professional engineers with little or without experience at all, who are not familiarized with earthquake damages, ought to carry out the most part of these evaluations. Although the existing evaluation guidelines present good different level damage descriptions it is ordinary to have the tendency of non-expert inspectors to aggravate or to underestimate the real level of damage at all.

The information that is handled in building damage evaluation is subjective and incomplete. In addition the damage levels are usually linguistic qualifications such as light, minor, moderate, average, severe, etc. Then, a building damage post-earthquake evaluation expert system, using artificial intelligence (AI) neural networks and fuzzy sets techniques has been developed. This tool allows learning from experts, using the Kohonen algorithm, and the use of linguistic variables to perform the building damage evaluation by non- experts that participate in a massive survey of buildings.

This model considers the different possible damages in structural and architectural elements and the potential site seismic effects in the ground. It also takes into account the preexisting conditions that can make the building more vulnerable, such as the quality of the construction materials, plant and height irregularities and bad structural configurations. For the model development, several building damage evaluation guidelines were taken into account from different countries, namely Mexico, Japan, United States, Italy, Macedonia (old Yugoslavia), the methodology used after the earthquake of 1999 in the coffee growing area in Colombia, and the methodology developed for Bogotá City also in Colombia.

The number of input neurons or variables in the model is not constant. It depends on the class of the structural system that will be evaluated, and on the importance of the different groups of variables selected for the evaluation. The number of neurons of the input layer for the structural elements group changes according to the class of building. The considered structural elements depend on the structural system that studies in each case. The architectural elements include partitions walls, elements of facade and stairs. The problems of landslides and liquefaction of the ground are contemplated in the ground conditions, as well as the pre-existent conditions, they are described before. In some cases the ground and pre-existent conditions can or not be considered, this depend of the influence in the building state, this is specially used in extreme cases.

Three layers neural network have been designed, using a non-supervised learning algorithm, and has been calibrated using the information of 1999 Colombian coffee growing area earthquake. The artificial neural network obtains indexes for the state of the different elements included in the evaluation, Structural elements, non-structural elements, ground conditions and preexistent conditions. The next step is to obtain results for the global state, reparability and habitability of the building; this process uses a fuzzy rule base.

Based on the damage level of the structural and non-structural elements and the state of the ground and pre-existent conditions, the habitability and the reparability of the building are assessed. The global level of building damage is estimated with the structural and non- structural damage results. The global building state is determined by taking into account the rule base on ground conditions, and thus the habitability of the building as outcome of the building state. These remarks or recommendations also include security measures to take into account by the building owners or occupants. The building reparability also depends on other fuzzy rule base: the pre-existent conditions. They contribute to define the technical and economical feasibility of building restoration. Then, for each building evaluation the system aids to make two critic decisions that are basic in the emergency response phase after the occurrence of a strong earthquake.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £82 +P&P)