Prediction of the Compressive Strength of Self-compacting Concrete containing Rice Husk Ash using Data Driven Models

The construction and maintenance of structural pavement was a high-cost problem in last decade. The mechanical properties of self compacting concrete (SCC) required important factors .From its mechanical properties, the compressive strength (CS) is necessary to investigate experimental and computational intelligence analysis in construction materials. Developing models with accurate estimation for this key property caused to saving costs and time and producing an optimal blend. Because of the many advantages, using of SCC in structures is increasing. Construction of precast-prefabricated components, with the use of concrete has also recently been considered. Concrete properties have significant role in precast-prefabricated girders behavior. Exact prediction of these properties is the base of member’s analysis and design. The main purpose of this study is presents new formulation to estimate the compressive strength of self-compacting concrete containing rice husk ash (RHA) using robust variant of genetic programming, namely gene expression programming (GEP) method. To evaluate the performance of the GEP-based proposed model, prediction was also done using classical data driven methods named artificial neural network (ANN) and multiple linear regression (MLR) models.  A large and reliable experimental database containing the results of 156 compressive strength of SCC incorporating RHA is collated through an extensive review of the literature. The performance of proposed models of CS is then assessed using the database, and the results of this evaluation are presented using selected performance measures. New expressions for the estimation of CS of SCC are developed based on the database. To evaluate the modeling performances of the proposed GEP models for CS, different statistical metrics were used. Correlation coefficient (R), root mean square error (RMSE), mean absolute error (MAE) were used as the measure of precision. The results showed that the models developed using the aforementioned methods have accuracy over 90 percent in prediction of CS of SCC. The results of testing datasets are compared to experimental results and their comparisons demonstrate that the GEP model (R=0.94, RMSE= 4.308 and MAE=4.916) outperforms ANN (R=0.92, RMSE= 5.136 and MAE=5.624) and MLR (R=0.89, RMSE= 8.212 and MAE=9.472). Proposed models have a strong potential to predict compressive strength of self compacting concrete incorporating rice husk ash with great precision. The importance of different input parameters is also given for predicting the compressive strengths at various ages using gene expression programming. Performed sensitivity analysis to assign effective parameters on compressive strength indicates that cementitious binder content is the most effective variable in the mixture. The assessment results present that the performance of the proposed models are in close agreement with the experimental results. Moreover, the new GEP-based formulation provides improved estimates of the compressive strength of SCC compared to ANN and MLR models. The proposed design equation can readily be used for pre-design purposes or may be used as a fast check on deterministic solutions.