Study On Chloride Ion Corrosion Resistance And Service Life Prediction Of Concrete Based On Neural Network

The deterioration of durability of concrete structures is becoming more and more serious.Especially in coastal areas,the structural failure caused by chloride ion erosion has attracted great attention.Therefore,it is of great significance to study the life prediction of concrete structures.The chloride diffusion coefficient of concrete is an important index to describe the resistance of concrete to chloride ion erosion,which directly affects the service life of concrete structures.However,the existing chloride diffusion coefficient prediction models of concrete are mostly single-factor models,considering fewer factors,resulting in a large difference between the calculated structural life and the actual life,so it is difficult to be popularized in practical engineering.In view of the above problems,the main research contents and achievements of this thesis are as follows:(1)Based on the 279 groups of experimental data of chloride diffusion coefficient of concrete measured by RCM method under no stress and 434 groups of experimental data of chloride diffusion coefficient of concrete measured by RCM method,NEL method and natural diffusion method under stress,the effects of water cement ratio,sand ratio,fly ash content,mineral powder content,silica fume content,stress ratio and age on chloride ion corrosion resistance of concrete were analyzed by single factor analysis,which laid a foundation for the establishment of the neural network prediction model for chloride diffusion coefficient of concrete(2)Taking water-cement ratio,sand ratio,fly ash content,mineral powder content,silica fume content and water reducing agent content as the input factors of BP neural network,and choosing chloride ion diffusion coefficient of concrete as the output result,the BP neural network prediction model of chloride ion diffusion coefficient of concrete under no stress was established,and the linear correlation coefficient of the network model prediction sample is 0.81696.The genetic algorithm was used to optimize the input factors of the network model.The results show that the BP neural network,which was re-established by using five factors such as water-cement ratio,sand ratio,fly ash content,mineral powder content and silica fume content,has better performance,and the linear correlation coefficient of the model predicted samples is 0.90393.(3)In order to avoid the neural network model falling into local optimum,genetic algorithm was used to optimize the weight and threshold of the BP neural network model for predicting chloride diffusion coefficient of concrete under stress-free action.The performance of the BP neural network model optimized by genetic algorithm was further improved.The average absolute error of the predicted output of the model is 13.74%,and the mean square error of the network model is 2.1961,which is significantly lower than the prediction model of chloride diffusion coefficient of concrete established by traditional regression method.(4)Taking water-cement ratio,sand ratio,fly ash content,mineral powder content,silica fume content,stress ratio,concrete age and the test methods of chloride ion diffusion coefficient as input factors,and the stress influence coefficient of chloride ion diffusion coefficient of concrete as output results,the BP neural network prediction model of chloride ion diffusion coefficient of concrete under stress was established.After optimization by genetic algorithm,the average absolute error of the predicted output of the network model is 16.80%.(5)Combining the BP neural network prediction model of chloride diffusion coefficient and DuraCrete model,the life prediction model of concrete structure under chloride ion erosion was established.The model can accurately show the effects of water cement ratio,mineral admixture,stress,chloride diffusion coefficient of concrete,surface chloride concentration,chloride diffusion coefficient attenuation index,concrete cover thickness on chloride diffusion rate and service life of concrete structure.