Over the past year, Baowu has orderly promoted the strategic plan of "Brain of the Steel Industry", which is strongly supporting the transformation of steel from "automation" and "informatization" to "intelligence" from point to point.
Under the guidance of Baowu Digital Intelligence Office and the Ministry of Science and Technology Innovation, after more than a year of exploration, innovation and practice, Baoxin Software has created a number of typical demonstration projects for the deep integration of artificial intelligence and steel, solved some of the "black box" and "uncertainty" problems in the manufacturing, service and governance process of the steel industry, and gave full play to the important enabling value of artificial intelligence technology.
The intelligence of decision-making depends on the accuracy and completeness of the information supporting decision-making
The steel production environment is complex and changeable, and intelligent visual recognition technology has the advantages of long-distance and non-contact, which can overcome the interference of high temperature, vibration, dust and other factors.
In order to improve the accuracy of the billet of the continuous casting secondary system, Baoxin Software developed a visual inspection model for Ma Steel, which was 15 seconds earlier than the current weighing method determination. The reduction of delay error improves the inconsistency between real logistics and information flow caused by the chaos of billet.
Visual inspection of large bales of open pouring
In order to promote on a large scale, Baoxin Software has developed a set of automatic development tools for visual models, so that engineers with zero algorithm foundation can independently develop visual recognition models in various steel fields.
With the cloud-edge collaboration mechanism of the industrial Internet platform, models can be quickly delivered to AI boxes deployed to the field. The tool enables the development of a model for a specific business scenario in a matter of days, significantly reducing the project implementation cycle compared to the traditional three- to four-month development cycle.
Visual pipeline application model
The core value of AI lies in optimizing decision-making
Compared with the human brain, AI technology is easier to make high-quality decisions when facing ultra-large-scale and ultra-complex scenarios.
Baoxin Software has created a headquarters and multi-base intelligent scheduling for Baosteel Packaging, covering 13 production lines across the country, which can intelligently arrange 3-month orders and nearly 3 billion cans in minutes.
Compared with manual order arrangement, it can reduce the frequency of tank production switching, improve the utilization rate of units, and become the first case in the can making industry to successfully apply artificial intelligence technology and bring economic benefits.
Baosteel packaging smart scheduling
The intelligent transformation and upgrading of steel manufacturing is no longer satisfied with the optimization of local processes and single processes, but should focus on the overall situation and carry out systematic optimization and integration
In the process of iron pre-mixing, sintering, coking and blast furnace production, batching is an important and complex work, manual single-process batching scheme mainly depends on experience, the factors that can be considered are very limited, the value of batching is relatively conservative, and there is great potential in reducing costs.
Based on the metallurgical principle and the actual situation on site, Baoxin Software carried out the development and application pilot of the integrated batching model before iron in Kunshan Iron and Steel Company, and according to the model output, focused on the global optimization work of the integration of the whole process and the whole process before the iron, and the model batching cost was reduced by 30 yuan / ton of iron compared with the manual batching cost. This technology has laid a solid foundation for the iron district to vigorously promote the implementation of "economic charge", bringing significant economic benefits.
Good decision-making also requires the ability to react quickly to emergencies on the spot
At present, China basically relies on manual experience to command and dispatch, but limited by many factors such as plant layout, process flow, logistics crossing, etc., once an abnormal link, manual command often makes the driving in the scheduling process passive yield, emergency stop, task delay and other phenomena frequently.
Baoxin Software has developed a driving scheduling model based on artificial intelligence technology for Baosteel Zhanjiang Iron and Steel, which intelligently schedules according to the real-time production status of the steel mill, and has entered the online operation stage.
The model has rapid response ability, which effectively guarantees the rhythm of steelmaking and ensures continuous casting while reducing the waiting time of heavy packaging by 3%, greatly reducing the energy consumption of molten steel.
Intelligent scheduling of ladle driving
Intelligence is also reflected in the prediction of future states based on the current state, combined with historical data, and then intervened in advance
Among them, the prediction of "black boxes" that are not easy to observe, as well as large time delays, highly nonlinear systems, is the most challenging. The differential pressure control of the blast furnace is a classic problem, and the current manual operation of the furnace can only be remedied by experience, and there is a large lag.
Baoxin Software tested the pressure difference prediction model based on deep learning technology in TISCO 6 blast furnace, and reached 80% prediction accuracy within 5kPa after one hour, which provided a valuable time difference for early intervention of blast furnace operation.
Blast furnace differential pressure prediction
With the gradual deepening of Baowu's "Steel Industry Brain" project, Baoxin Software will continue to actively explore the combination of data science and intelligent algorithms, pilot more application cases, and accelerate artificial intelligence to empower steel smart manufacturing.
Text: Zhang Xi
Image: Database, provided by Baoxin Software
Produced by: Baowurong Media Center