A hybrid optimized learning-based compressive performance of concrete prediction using GBMO-ANFIS classifier and genetic algorithm reduction

High performance concrete (HPC) is a type of concrete that cannot be produced using conventional methods. The exact percentage of materials used in the production of this concrete is one of the challenges facing civil engineers so that if ingredients are not in proportion, the strength of concrete is undermined. In the present study, attempts have been made to find an intelligent model to predict the quality of HPC. As a result of regression analysis, the automatic recognition would be affected by inferential estimation. Hence, to increase classification accuracy, first extracted feature is rearranged based on expectation-maximization clustering algorithm and then feature vector size is reduced using genetic algorithm. The proposed classification is adaptive neuro-fuzzy inference system, which is optimized by Gases Brownian Motion Optimization and able to predict outputs at an acceptable level in limited reiterations. The split ratio of data during learning and testing steps was 0.9 and 0.1, as measured by K-fold cross-validation method. Computation of criteria such as mean square error and mean absolute percentage error in the algorithm indicated the desirable performance of the proposed method.