ارائه مدل توسعه يافته به منظور ارزيابي كيفيت خطوط ريلي با استفاده از الگوريتم¬هاي محاسبات نرم

quality eva‎luation model , railway track quality , neural network

Developing a Model to eva‎luate the Quality of Railway Track Using Soft Computing Algorithms

مهندسي راه آهن و برنامه ريزي حمل و نقل

Increasing the tonnage an‎d frequency of cargo passing through the lines has increased the deterioration of the railway lines. As a result, one of the problems an‎d challenges facing the railway industry is the deterioration an‎d destruction of infrastructure. The construction an‎d operation of railways requires a lot of money. However, due to the many benefits of rail systems for the economy an‎d society of a country, this industry is considered as a national asset. Therefore, in order to maintain this national capital, it is necessary to formulate coherent an‎d purposeful plans for the construction, operation an‎d maintenance of lines. In the meantime, overhaul an‎d annual maintenance of railways as a preventive action is extremely important. This study examines the factors affecting the railway infrastructure such as geometric parameters, traffic an‎d environmental aspects an‎d finally, by developing a suitable deterioration model of railway lines, provides the basis for the optimal use of this infrastructure. In order to express the condition of the line an‎d interpret the trend of line degradation, four ITGI, OTGI, TGI an‎d J indicators were selec‎ted from the quality indicators of the line. After reviewing a large number of articles related to the deterioration of railways, the most efficient an‎d widely used methods were identified an‎d selec‎ted to solve this research, which include neural network, support vector machine, decision tree, CRT an‎d regression. As a result, five solution scenarios were implemented for each index. The information used in these scenarios is the output data of the line measuring machine related to the north-east axis (1) of the country. 70% of the data were used in model construction, 15% in eva‎luation an‎d the remaining 15% were used for model validation. By comparing the eva‎luation an‎d validation results, the accuracy an‎d percentage of performance of the models were obtained, which is a good criterion for comparing them with each other. Finally, the OTGI index performed better than other scenarios by solving neural networks. Another approach of this research is to perform sensitivity analysis for each scenario, which determined the effect of the parameters. Considering the totality of scenarios, among the 9 input parameters, three parameters of distance, line width an‎d horizontal alignment of the left rail have the greatest impact on the condition of the railway lines.

5

124180