An artificial neural network for the prediction of the strength of supplementary cementitious concrete

Abstract

Supplementary materials (SM) for cement replacement became more feasible in the previous decade due to their pozzolanic strength and durability properties. The strength variation according to the age of the binding material is a critical subject for SM concrete. The time of water curing is critical in order to maintain the pozzolanic reaction in SM concrete, which assists in the development of strength in cementitious properties.In this study, the laboratory results of concrete specimens were assessed for various mix designs, and the obtained corresponding strengths were also predicted with ANN techniques. The difference between experimental and ANN predicted values was marginally low. Thus, the ANN model applicability emphasised the productive use in predicting the strength of supplementary materials like fly ash or any similar pozzalanic mineral admixture. This research proposes an ANN-based technique for predicting the strength of fly ash added SM concrete. Typical experimental data is used to build, train, and test the artificial neural network (ANN) model. With ANN and input parameters, a total of 324 distinct data points for SM concrete were utilised to estimate SM concrete strength. Various combinations of layers, number of neurons, and learning rate were examined during the training phase. When the root mean square error (RMSE) reached or remained below 0.001, the training was stopped, and the findings were verified using a test data set. With respect to the relative error provided for trained model data, the results achieved were typically below 10% for compressive strength and below 5% for split tensile strength. The ANN models predict concrete strength with excellent accuracy, and the findings show that utilising ANNs to predict concrete strength is both practicable and advantageous.

Friction stir-welding of AZ31B Mg and 6061-T6 Al alloys optimization using Box-Behnken design (BBD) and Artificial Neural network (ANN)

“…The artificial network is also called a neural network are the heart of deep learning algorithm and a subset of machine learning and used to improve the quality of mechanical research [38], [39]. The learning algorithm is being developed through ANN.…”

Predicting strength of concrete by ensemble technique

“…It is not easy to strike a balance between cost and quality when considering the quantity of each suitable concrete material to use, as determining the C.S of concrete takes a lot of time and work. Scientists have spent the better part of a decade creating artificial methods for picking the most effective strength prediction techniques [15] to help them save time and money in the lab. Complex concrete mixtures are difficult to locate and predict.…”

“…Recently, researchers have been putting a lot of effort into developing prediction scenarios for a variety of mechanical features in concrete with the use of tearing technologies like artificial intelligence (AI) and machine learning (ML) [15,17]. Using methods such as supervised learning, it is possible to estimate a great many parameters (W/C, SCBA%, FA, CC, CA ), although with varying degrees of accuracy in the regression, classification, clustering, and reinforcement learning [22].…”