A Fuzzy Model Approach to Stress-Strain Relationship of Concrete in Compression

In this study, a fuzzy model was developed to predict the stress-strain relationship of unconfined concrete in compression. For this purpose, a database including 91 data sets was used to train and test a neuro-fuzzy system which included two inputs and one output. The model can be utilized to predict the stress-strain relationships of the unconfined concrete standard cylindrical members tested by different researchers. After the training stage, a total of 64 if-then fuzzy rules were obtained. Lastly, performance of the model was evaluated and compared by the other models proposed by different researchers. The performance values such as root mean square error, relative error and average of predicted values to experimental values were calculated. These performance values showed that the constructed neuro-fuzzy model exhibited a high performance by means of prediction of stress-strain relationship of concrete.