Fracture performance prediction of polyvinyl alcohol fiber-reinforced cementitious composites containing nano-SiO2 using least-squares support vector machine optimized with quantum-behaved particle swarm optimization algorithm

Abstract Based on least-squares support vector machine optimized using a quantum particle swarm optimization algorithm (QPSO-LSSVM), a prediction model was established to predict the fracture properties of polyvinyl alcohol fiber-reinforced cementitious composites (CCs) containing nano-SiO2 (PVA-CCNS) and improve its accuracy and effectiveness. Nineteen groups of measured data obtained from fracture performance tests on a three-point bending notched PVA-CCNS beam were selected for analysis and prediction. The prediction results of the QPSO-LSSVM model were compared with those of the least-squares support vector machine optimized with the particle swarm optimization algorithm, least-squares support vector machine and back-propagation neural network models. The simulation analysis results indicated that the goodness of fit ( R 2 ) values of the fracture energy, initial fracture toughness and unstable fracture toughness were 0.790, 0.940 and 0.950, respectively, for the QPSO-LSSVM prediction model. In addition, the fitting degree between the measured and predicted values of the QPSO-LSSVM prediction model was better than those of the other three models. The higher accuracy, better convergence, and robustness of the QPSO-LSSVM model than the other three models proves that the QPSO-LSSVM model is an optimal method for predicting the fracture performance of CCs. The proposed model can guide the mix proportion design of CC mixtures.