Prediction of Concrete Compressive Strength Based on ISSA-BPNN-AdaBoost

Strength testing of concrete mainly relies on physical experiments, which are not only time-consuming but also costly. To solve this problem, machine learning has proven to be a promising technological tool in concrete strength prediction. In order to improve the accuracy of the model in predicting the compressive strength of concrete, this paper chooses to optimize the base learner of the ensemble learning model. The position update formula in the search phase of the sparrow search algorithm (SSA) is improved, and piecewise chaotic mapping and adaptive t-distribution variation are added, which enhances the diversity of the population and improves the algorithm's global search and convergence abilities. Subsequently, the effectiveness of the improvement strategy was demonstrated by comparing improved sparrow search algorithm (ISSA) with some commonly used intelligent optimization algorithms on 10 test functions. A back propagation neural network (BPNN) optimized with ISSA was used as the base learner, and the adaptive boosting (AdaBoost) algorithm was used to train and integrate multiple base learners, thus establishing an adaptive boosting algorithm based on back propagation neural network improved by the improved sparrow search algorithm (ISSA-BPNN-AdaBoost) concrete compressive strength prediction model. Then comparison experiments were conducted with other ensemble models and single models on two strength prediction datasets. The experimental results show that the ISSA-BPNN-AdaBoost model exhibits excellent results on both datasets and can accurately perform the prediction of concrete compressive strength, demonstrating the superiority of ensemble learning in predicting concrete compressive strength.