Application of machine learning in determination of the permissible level of traction current asymmetry

Approaches to solving the problem of determination of the permissible level of traction current asymmetry in railway transport that is required for the transition from service according to the standard value to the actual service are reviewed in the article. The presented solutions, which are based on the machine learning method, comply with the policy of the Russian Railways company in the field of transformation of the railway automation and telemechanics economy. Methods for collection of information from devices subject to the influence of traction current asymmetry are proposed to solve the problem. Method for calculation of the traction current asymmetry coefficient based on machine learning methods is selected and justified. The Lasso method is used as a machine learning method for solving the regression problem. This method allows to predict the required value with high accuracy, including cases of work with big data.