Estimation of compressive strength of concrete cement using random vector functional link networks: a case study

Estimation of compressive strength (CS) of concrete is an active area of research in the domain of manufacturing engineering. The concrete ingredients are disseminated arbitrarily over the whole concrete matrix and the relationships among them are a complex nonlinear system, thus challenging to model. Soft computing techniques are found better to statistical methods applied for prediction of CS. However, sophisticated prediction models are still missing and need to be explored. Random vector functional link network (RVFLN) is an efficient algorithm with low time complexity and is able to handle complex domain problems without compromising on accuracy. It assigns input-hidden layer weights and bias randomly without further modification and computes output layer weights iteratively by gradient methods or non-iteratively by least square methods. Though it is applied on solving various engineering problems, its application on CS prediction of concrete is scarce. The learning ability, structural simplicity, and computational efficiency of RVFLN motivated us to investigate its efficiency on compressive strength prediction. In this article, we develop a RVFLN-based forecast for estimation of CS concrete. A publicly available dataset from UCI repository is used to develop and access the performance of the model. For comparative analysis, few other models such as FLANN, MLP, RBFNN, MLR, and ARIMA are also developed and used for the forecasting considering samples with curing ages at 14, 28, 56, and 91 days. All the models are evaluated in terms of MAPE, ARV, U of Theil’s statistics (UT), NMSE, and execution time. Considering the four-sample series, RVFLN attained an average MAPE of 0.5341, ARV of 0.0982, UT of 0.3239, and NMSE of 0.1188 which are smaller than other models. The study revealed that RVFLN is quite capable in modeling and predicting the CS data. Outcomes of comparative studies and statistical significance tests are in favor of RVFLN-based forecasting.