Prediction Of Bearing Capacity Of Reinforced Concrete Members After High Temperature Based On GA-BP Neural Network

Fire is a common disaster in human society.Among them,building fires have the most frequency and the heaviest losses.In the event of a fire in a building,high temperature will weaken the strength performance of building materials such as steel bars and concrete,which will reduce the bearing capacity of reinforced concrete(RC)members and will lead to the destruction of the reliability of the building structure.In order to repair the building structure,it is necessary to clarify the residual bearing capacity of RC members after high temperature.However,there are many factors that affect the bearing capacity of RC members after high temperature.When calculating the bearing capacity of RC members after high temperature,both the test method and the theoretical calculation method are more complex.It is particularly important to introduce new methods and technologies to predict the residual bearing capacity of RC members after high temperature.With the development of the era of Big Data,BP(back propagation)neural network and genetic algorithms(GA)as branches of intelligent algorithms are widely used in the research field because of their strong non-linear data mapping ability and fitting ability,And it has achieved the ideal effect in the traditional civil engineering industry.Therefore,this paper uses GA-BP neural network to propose a method for quickly predicting the bearing capacity of reinforced concrete members after high temperature.The main research content and results of the force prediction model are as follows:(1)Appropriate thermal parameters and material strength reduction coefficients of steel and concrete after high temperature are selected,and a method for calculating the bearing capacity of RC members after the high temperature is proposed based on the improved section method.Based on the ABAQUS finite element analysis software,according to the temperature field theory,select the parameters of the component high-temperature simulation,analyze the temperature field of the reinforced concrete component,determine the temperature field of the component section at different temperatures,and obtain the strength reduction coefficient of the steel and concrete materials.Finally,the calculation method of the bearing capacity of RC members after high temperature is determined by the improved section method.(2)The high temperature test and static loading test of RC beam after high temperature are carried out to obtain the temperature field of beam members and the bending test data of RC beam,which are compared with the temperature field simulation and the theoretical calculation value of bearing capacity to verify the feasibility of the temperature field model and the bearing capacity calculation method based on the improved section method.Referring to the literature,aiming at the stress model of RC beam shear bearing capacity,RC column axial compression bearing capacity and eccentric compression bearing capacity after high temperature,using the experimental data in the literature to compare the theoretical calculation value,the effectiveness of the calculation method is proved.(3)Based on GA-BP neural network,the bearing capacity prediction model of RC members after high temperature is established.Through the analysis and calculation,the parameters needed for the establishment of BP neural network and the operation of genetic algorithm are determined.The main factors affecting the bearing capacity of RC members after high temperature are selected as the input parameters of GA-BP neural network,and the bearing capacity of RC members after high temperature calculated by the improved section method is taken as the output parameter.The prediction model of bearing capacity of RC members after high temperature based on GA-BP neural network is obtained.(4)The prediction effect and evaluation of the bearing capacity prediction model of RC members after high temperature based on GA-BP neural network.GA-BP neural network and BP neural network are respectively used to predict the bearing capacity of RC members after high temperature,and the results are compared with the theoretical values proved to be effective by experiments.The accuracy of the neural network is evaluated by comparing the relative error,absolute error and fitting ability between the predicted values and the theoretical values of the two neural networks.The prediction results show that the prediction accuracy and stability of the prediction model for the bearing capacity of RC members after high temperature based on the GA-BP neural network are higher.