Predictive analytics in the planning of a contingent of track fitters in automated railway transport management systems

Abstract

One of the important efficiency indicators in the organization of the current railway track maintenance is the high-quality execution of technological processes in compliance with typical time standards. Track work of the track distances is carried out based on the indicators of the track upper structure condition, taking into account traffic safety and optimal infrastructure maintenance costs. Planning and organizing of track work, first of all, depend on the main participants of technological processes that is the track fitters, their qualifications and categories for the types of work to be performed. At the same time, specialists of the distance of the track are engaged in planning the contingent of workers and, in many respects, the quality of the work depends on their knowledge and qualifications, the conclusions they draw based on the reporting data from the systems. The article discusses the issues of planning a contingent of track fitters to fulfill the current content of the track, taking into account existing problems, as well as the possibility of improving corporate automated systems through the introduction of digital technologies – predictive analytics, big data, machine learning. The planning of track work is carried out in the unified corporate automated infrastructure management system based on data on the movement of personnel from the unified corporate automated human resource management system. There are disagreements in the systems on the categories of track fitters, because the list number of workers is taken into account in the unified corporate automated human resource management system, without taking into account temporary transfers. This results in a difference between the average category of workers in the systems, also due to the fact that tools for calculating staff turnover, forecasting traffic and the number of track fitters are not implemented, which leads to the complexity of contingent planning, especially for an annual period. Predictive analytics in combination with mathematical models of non-obvious dependencies and descriptive analysis will allow systems to create predictive analytics of contingent movement and identify hidden problems that can affect the effective planning of travel work.