engleski

Comparison of NEeutral Network Durability Models for Reinforced Concrete Structures

Data on relevant structural parameters gathered on eleven reinforced concrete structures in continental and three in Adriatic marine environment at various ages have been used for training artificial neural networks. Two separate models, for continental and marine environments have been developed. Data for modeling in continental environment were gathered at ten different ages of bridges and consists of 213 records. Data on marine environment were gathered at seven different ages of structures and consists of 124 records. The effects of the structure, properties of concrete and environmental conditions onto the degree of damage caused by steel corrosion have been investigated. The damages were classified into five categories based on the type of necessary remedial works. The artificial neural network for feature categorization was used as a tool for classification of damage and prediction of damage degree. This paper demonstrates the use of artificial neural networks in modeling the durability of reinforced concrete structures in marine and continental environment. A comparison of the two models regarding the different environments is given. The models are able to recognize and evaluate the effect of individual parameters on the damages. The developed models could be useful for planning the maintenance of investigated structures and design of remedial works.

reinforced concrete structure; durability; corrosion; damage categorization

nije evidentirano