Prediction And Analysis Of Asphalt Mixtures Fatigue Based On ICA Optimized Neural Network

The fatigue failure of asphalt pavement is one of the main forms of asphalt pavement damage,and it is also considered to be the origin of the basic theory and design of asphalt pavement structure.Through a lot of studies show that the fatigue performance of asphalt mixture is affected by many factors.There is a big difference between the indoor fatigue test results and the actual fatigue performance.Therefore,it is necessary to study the fatigue performance prediction of asphalt mixture under the condition of many factors.Based on the gradation of asphalt mixture,combined with the existing research results and the application of neural network,this paper proposes the use of optimized neural network model(the combination of imperial competition optimization algorithm and neural network)to the asphalt mixture under the influence of many factors.Fatigue performance was predicted and analyzed.The research in this paper is mainly as follows:(1)By consulting the relevant literature,the factors affecting fatigue performance,the method of making test pieces,the method of fatigue test and the selection of relevant control parameters were compared and analyzed,and the specific test variables and test schemes based on the optimized neural network prediction model were determined.(2)The basic principles,characteristics and their advantages and disadvantages of BP neural network,RBF neural network and Empire competition algorithm are expounded.Four neural network prediction models of BP,RBF,ICA-BP and ICA-RBF were established by using MATLAB software.(3)Through the MATLAB software,four neural network prediction models of BP,RBF,ICA-BP and ICA-RBF were run to predict the fatigue performance of asphalt mixture,and the relevant parameters of the program were debugged to achieve the best prediction results.The optimal prediction model(ICA-BP)is obtained by error analysis of the output of the four prediction models.(4)The optimal performance model(ICA-BP)was used to predict the fatigue performance of asphalt mixture.The weight matrix and weight vector were obtained.Finally,the fatigue performance prediction formula of asphalt mixture was obtained.Through the scientific analysis of the weight matrix,the degree of influence of various fatigue performance factors is obtained.(5)Based on the optimal prediction model and the conventional BP neural network prediction model,the relationship between the number of samples and the prediction error is studied,and the variation between the number of samples and the prediction error is obtained.Finally,the application of the model was described,which based on the optimal neural network model.In conclusion and prospect section,the author summarized the work and showed shortcomings of this paper,and discussed the work in future.