Strength Prediction Of Rubber Steel Fiber Reinforced Concrete Based On Meta-Heuristic Algorithm

In this paper,the strength prediction of rubber steel fiber reinforced concrete is studied.With the development of scientific research,the traditional prediction methods cannot meet the practical needs in accuracy and batch processing,so the neural network is applied more and more widely.However,because of the initial threshold and weight randomly generated by a single BP neural network,it is easy to lead to low operation efficiency and local optimal solution.Therefore,a new meta-heuristic algorithm,particle swarm optimization(PSO)algorithm,is used to optimize the BP neural network,and the global optimal initial weights and thresholds are obtained,and then the training and strength prediction of the BP neural network are carried out.By testing 50 groups of standard cubes,the following conclusions are obtained:1.Steel fiber content within 0.75%,steel fiber can improve the compressive strength of concrete,but the steel fiber content continues to increase,the compressive strength increase is not obvious.The incorporation of rubber particles results in a large decrease in the strength of concrete.2.In terms of workability,rubber particles have little effect on the fluidity of concrete,but it will reduce the cohesion and water retention of concrete.Steel fiber will significantly reduce the workability of the mixture.3.The analysis shows that the maximum relative error of PSO-BP neural network is 0.20421 and the average relative error is 0.077849.The maximum relative error of single BP neural network is 0.27797 and the average relative error is 0.13857.PSO-BP neural network has a 4%advantage in data fitting.In summary,the PSO-BP neural network has a good effect in predicting the strength of rubber steel fiber concrete.