چكيده به لاتين
In recent years, many studies have been conducted on the geometric deterioration of railway track with the aim of predicting failures. However, the discussion of deterioration modeling needs further study.
Due to the fact that a number of railways in the country have a mixed traffic load and so far the effect of mixed loading has not been seen in the developed degradation models, so the need to develop a geometric degradation model with mixed traffic load is essential. This modeling will help to investigate the effect of different load combinations on line degradation, to calculate the degradation rate of the desired track in terms of mixed cumulative traffic load.
To develop the deterioration model, Esfarayen-Naqab block has been studied and mlp artificial neural network in MATLAB environment has been used.
Input data for model development, with the help of information related to the average speed of trains and the number of trains passing the desired block in 1398, has been generated randomly. Three models, Sato, Shenton and iwnicki, have been selected to generate data related to the dependent parameter. Based on the three models, line settlement (mm) was determined as a dependent parameter of the developed model. Input data including: cumulative traffic load, percentage of passenger trains, percentage of freight trains, percentage of empty trains and average speed of trains were placed in three models and their output was used as the target parameter. Finally, using the input of three models as the input of the final model and their output as the target parameter and with the help of a neural network, the desired model was developed.
Thousand data have been produced for training and testing the model, of which 70% has been used for training and 30% for model testing. The best neural network function is a network with a hidden layer and five neurons in the hidden layer, which is trained by the Levenberg-Marquardt method and the sigmoid tangent functions in the hidden layer and the linear function in the output layer are used as the activity function. The performance of this model is 0/7747 and 3/43×〖10〗^(-4)for the coefficient of determination and the mean squared error, respectively. The coefficient of determination is related to the developed model and each of the three models Sato, Shenton and Iwniki is 0/6835, 0/8344, 0/8754, respectively. After developing the model to investigate the effect of mixed traffic load on line deterioration, 10 different loading combinations were created. The numbers in parentheses from left to right are the percentage of passenger, freight and empty train, respectively; (70,5,25), (35,50,10), (15,30,55), (35,35,30), (50,45,5), (10,50,40), (50 , 10,40), (80,5,15) (10,5,85), (5,90,5), the result of loading combinations from right to left (0/275, 0/952, 0/568 0/679, 0/933, 0/802, /0190, 0/292, 0/141, 0/99). The result shows that the combination of different loads has a different effect on the geometric deterioration, in the meanwhile the freight train causes more deterioration.
