Development of Adaptive Neuro-Fuzzy Inference System to Predict Concrete Compressive Strength

• Main Authors: Seyed Hakim, Seyed Jamaldin, Jamaludin, Norwati, Heng Boon, Koh, Mokhtar, Shahrul Niza, Ali Khalifa, Nasradeen, Jamellodin, Zalipah
• Format: Conference or Workshop Item
• Language: English
• Published: 2024
• Subjects: TA Engineering (General). Civil engineering (General)
• Online Access: http://eprints.uthm.edu.my/12475/1/978-981-99-6018-7_24; http://eprints.uthm.edu.my/12475/

Predicting the compressive strength of concrete is one of the complex problems in civil engineering because different parameters and factors must be considered. There is several research that have predicted the compressive strength of normal concrete using neuro-fuzzy systems. However, little research has been done to predict the strength of high strength concrete. Recently, machine learning techniques such as artificial neural networks (ANNs), fuzzy logic, and adaptive neuro-fuzzy inference system (ANFIS) are becoming extensively established in predicting complex problems. ANFIS has the advantages of both ANNs and fuzzy systems and is most suitable in engineering complicated applications. This study focuses on the development of ANFIS in predicting the compressive strength of high strength concrete. A total of 550 experimental datasets of concrete were used in this research. Each dataset was consisting of six input variables that were water, cement, fine and coarse aggregates, silica fume, and superplasticizer. The compressive strength of high strength concrete was considered as the output of the ANFIS model. In this study, 440 datasets were assigned as training datasets and 110 datasets were considered as testing sets to verify the ANFIS model. The mean square error (MSE) for the training set was 0.00573, and 0.00647 for the testing datasets. The ANFIS model was able to quickly predict the concrete compressive strength with high accuracy. Also, in this research, a sensitivity analysis was applied to study the contribution of input parameters to predict the compressive strength of concrete.