Fuzzy Logic Approach in Determination of Strength in Concrete

Abstract

The aim of this thesis is to address capabilities in the prediction of compressive strength of concrete to affect quality control in construction. To comprehend this, a compressive strength predicting model using the principles of fuzzy logic set theory had been employed. The model put into use ‘fuzzy logic’ as a tool to predict the compressive strength of concrete on a given day. Data collected from previous researches and laboratory work had been put into use in the model construction and testing. The input variables of water/binder ratio, cement content, water content, and fly ash percentage and the output variable of 28-day cement compressive strength were fuzzified by the use of triangular membership functions and Gaussian membership functions which were deployed for the fuzzy subsets. The prediction of the 28-day cement strength data by the developed fuzzy model proved to be quite satisfactory. The training and testing of 4 different models were done. The Minimum average percentage error levels in the fuzzy model were seen to be as low as (3%) in the case of Model 3. A comparative study of the different models (all 3 Triangular and 1 Gaussian) had been done. The results indicated that the application of the fuzzy logic algorithm was quite satisfactory when a triangular membership function with decreased subset range was used. The outputs of the Triangular and Gaussian models were almost similar.