Monitoring of Transport Mechanisms and Containers at Steel Production Plant

In 2021-2022 Mallenom Systems developed a subsystem for Data-Center Automatika LLC which monitors transport mechanisms at the steel production plant. This subsystem became part of the dispatching and optimization system for steel melting production process at NLMK PJSC.

Mallenom Systems’ solution, based on neural network algorithms for image analysis, controls the movement of steel ladles and cranes, pouring of cast iron and movement of iron ladles in the converter workshops of the enterprise. It also generates relevant data on their location and condition. 

The obtained data are used to solve the challenges of internal logistics at the enterprise.

During the project implementation, the following challenges of tracking containers and mechanisms associated with unique conditions of steel manufacturing have been successfully resolved:

• Failure of video cameras to focus and shift in their positioning due to production vibrations in the shops
• Overexposed images due to flashes coming from the ladle
• Camera view that is blocked by crane grippers
• Entering of other mechanisms into camera view that can be misidentified as control object

Overall, the system for video analytics consists of more than 100 cameras. To control condition and positioning of each individual camera, Mallenom Systems developed special service for the control of optical circuits. It automatically estimates the amount by which the camera view has shifted relative to its default position. If the deviation is small, this shift is taken into account by the algorithms by performing required level of adjustments. If the deviation is large, notification message is sent to the person responsible for setting up the cameras at the facility.

Monitoring subsystem includes:

• 113 video cameras
• 7 video analytics servers and 2 separate video servers

Overall, there are 461 video streams connected to 8 servers, with recording performed for 253 video streams, retransmission – for 268 streams, and video analytics – for 132 video streams.

The data generated by video analytics is transferred to the planning optimization system developed by Data-Center Automatika LLC.

The full-scale implementation of the system formed the basis for "GE-FEST: internal logistics and dispatching for steelmaking facility (NLMK)" digital service. The system won RusBase Digital Award 2021 in the "Production" category and Siemens "STAR 4.0 digital industrial innovation competition" in the "Reducing Life Cycle Costs" category.

Project results:

• Development of a comprehensive industrial system for visual data processing based on the utilization of the latest algorithms for machine vision and mathematical modeling
• Ability to perform both long-term and operational planning for the manufacturing
• Optimization of the facility infrastructure utilization
• Ability to flexibly reschedule operation procedures via online access

Economic efficiency of the project:

• Planned effect: 1.2 million USD per year
• Reducing the consumption of electricity, graphite electrodes and aluminum wire rods
• Saving of resources and materials is achieved by reducing the holdup time of the metal in the steel ladle by up to 7%
• Intermediate result: during the first 2 months of the pilot system operation, the average production holdup time decreased by 5%