Management Automation Solution
rolling stock of wagons, box-wagons,
containers and tank cisterns
Our product solves several problems: traffic forecasting, rolling stock management, visualization of the movement and location of rolling stock, dynamic pricing of tariffs. In addition, we predict repairs and uncouplings based on the forecast of wheel set wear.

The forecasting module solves the problems of predicting the time of movement of the rolling stock and operations on it. Our product gives a forecast of the time of the wagon in transit both before departure (according to the waybill) and in the movement of the wagon (based on operations). In addition to the actual travel time, our product predicts both the duration of individual operations, such as loading / unloading, and the likelihood of individual operations, such as uncoupling. Our solution also allows us to predict the available number of wagons at the station for loading on a given date.

The operator's workplace is a management cockpit for the functions of Movers and Logistics. The control module integrates data from various systems, internal rolling stock accounting systems, planning systems for the availability of rolling stock at stations for operations, contractual information at the client level. The control module enriches the actual data with the predictive data received from the forecasting module. The management module presents information in the form of management dashboards, taking into account the work processes of the corresponding functions. Dashboards allow not only to display information, but, above all, to make changes, thus managing the movement of rolling stock.

The visualization module displays summary information on the availability and movement of rolling stock in the form of a railway map. On the map, by analogy with traffic jam maps for cars, using colors, the state of traffic on different sections of the road is displayed in terms of the number of cars, and in terms of average speeds for different periods. In addition, the loading of stations, including port stations, the number and points of dropping wagons is displayed. Directly on the map, you can request routes and predicted travel times from A to B or along compound trajectories.
Technology in product development
Machine learning - using Python and the LightGBM library to build decision trees, as well as visualize live data on the location of wagons and drops
  • 1
    Historical data on wagon trips and their idle time at stations
  • 2
    Building machine learning models to predict given characteristics
  • 3
    Using real-time data to display current rail traffic
Time Forecast Module
rolling stock movements

  • Wagon travel time forecast
    End stations are the stations of departure and destination (A and B), and not intermediate stations along the way. The forecast of the time of arrival at the terminal stations is carried out by wagon.

    We solve the problem of predicting the travel time, since the arrival time = departure time + travel time.

    The sample is divided into groups according to the distances between the end stations. The greater the distance, the more accurately the predictive model predicts. To assess the accuracy, the MAPE metric is used - the average absolute error in percent
  • Wagon time forecast at the station
    Forecast of the wagon time at the stations of the beginning and end of the journey. This time consists of the unloading time of the current travel, the idle time at the station and the loading time for the next travel.

    Loading and unloading times can be described by operations on wagons, which will increase the accuracy of the model when using live operation data.

    The sample is divided into groups according to the average time at the stations. The more time a wagon spends on average at the station, the less accurate the model. This is due to the fact that it is most difficult to predict the demurrage of a wagon at the station.
    In total, for 96% of the stations, not only the average, but also the maximum error does not exceed 50%.
  • Metrics visualization module
    • Load of transport services
    • Throwing wagons and trains at stations
    • The speed of wagons on sections
    • The volume of wagons at a particular station
    Customizable visualization options:
    Detail: the whole map; to the railway section; to the railway; to the station
    Filters: loaded/empty wagons; by composition type
Wheel set wear forecast
The share of wagon repairs due to wheel set malfunction reaches 65%.
The uncoupling of a wagon for repairs often takes place at an unpredictable location for the company that owns the wagon. For repairs, there is not always infrastructure, spare parts (wheel sets, disks, axles)
Solution Benefits
  • 1
    Clarification of repair cost
    The system will allow you to know more precisely when it is worth repairing railcars and how much such repairs will cost. Knowing in advance where to transport the wagon, we can control the cost of parts for repairs.
  • 2
    Avoidance of Secondary Repair
    Avoidance of unnecessary repairs and quality control of repairs will reduce the cost of repeated repairs of wagons.
  • 3
    Preservation of deadlines and reputation
    When wagons are unlawfully uncoupled, not only deadlines are broken, but reputation suffers as well, if this ultimately affects the profits of the recipients of the cargo.
Connect with us
Our contacts
+7 967 215 7505