share_log

用AI“爆改”工厂 实探宝钢股份汽车板产线的AI云端化操作车间|高质量发展调研行

Using AI to “explode” the factory to explore the AI cloud-based operation workshop of Baosteel's automobile board production line | High quality development research project

cls.cn ·  May 24 08:35

① This year is the first year Baosteel Co., Ltd. comprehensively promoted the AI strategy. The goal is to achieve 100 scenario applications, covering AI production line control, AI visual recognition, and AI intelligent decision-making. ② Baosteel has established an AI computing power center and formed a digital intelligence team. Currently, data resources from various manufacturing bases are being concentrated to verify the effects of large models on the electrogalvanized automobile sheet production line.

“Science and Technology Innovation Board Daily”, May 24 (Reporter Zhang Yangyang) In a steel production workshop at Novo University, unmanned driving equipment shuttles back and forth while holding a cold-rolled steel coil at high altitude. This is a scene that happened in Baosteel's “AI-driven workshop”.

ST247viW0r.png

Picture/ “Science and Technology Innovation Board Daily” reporter

Recently, an interview event on the theme of “High Quality Development Research Tour” organized by the Propaganda Department of the CPC went to Shanghai. At the first stop of the interview and investigation, the “Science and Technology Innovation Board Daily” reporter visited Baosteel Co., Ltd. in Baoshan District.

The “AI-driven workshop” in front of us is the C008 hot-dip galvanizing smart workshop at the Baoshan base cold rolling mill. It's not like the bright lights, the furnace is red, and the heat is in full swing. There are very few people working here, only the loud sound when lifting steel coils. It also reminds me that this is a factory that is still in operation 24 hours a day.

ZY731491uV.png

Picture/ “Science and Technology Innovation Board Daily” reporter

In the industry, this “black light factory” has long been famous. It is directed by a “six in one” control room not far apart. In this small room, there were only 4 operators. Facing dozens of screens, the real-time production data and analysis of the factory workshop flickered on the screen. The actual production situation at Baosteel's Zhanjiang production base, which is thousands of miles away, was also clear at a glance.

Cj6WQuQ04z.png

Picture/ “Science and Technology Innovation Board Daily” reporter

“We have made rough statistics. Previously, during real-time monitoring, operators had to adjust the operation every three minutes or so. After using AI as the main driver, it was enough to intervene once in half an hour, reducing the workload on personnel by more than 90%.” Liu Decheng, deputy director of the cold rolling mill of Baosteel Co., Ltd., in charge of intelligent manufacturing, told the “Science and Technology Innovation Board Daily” reporter.

Baoshan base cold rolling plant C008 hot galvanizing intelligent workshop. This workshop is used to produce automobile panels. A series of black light factories, including this workshop, is one of the first batch of intelligent manufacturing pilot demonstration zones named by the Ministry of Industry and Information Technology. It is also one of the first batch of intelligent chemical plants in Shanghai.

Liu Decheng said that Baosteel's cold rolling department has experienced nearly 10 years of digital intelligence construction. By promoting extreme automation such as robots and autonomous vehicles, the labor efficiency of the automobile board production line has increased by more than 30%.

In the past three years, Baosteel Co., Ltd. has focused on promoting big data and artificial intelligence. The goal is not only to reduce physical load, but also to reduce heavy mental load, especially in the core production control process of steel manufacturing.

However, AI has entered the core of traditional steel manufacturing, and it is still a very challenging issue for steel manufacturers. In the process industry, high reliability and high stability are basic requirements. However, the unexplainability of AI itself poses a risk of uncertainty. The contradiction between the two is stark.

Liu Decheng said that through nearly three years of exploration and iteration, Baosteel Co., Ltd. has found a method and path, and successfully implemented AI (cloud-based operation to achieve cross-interface integration) in the automotive board production line.

Take an annealing furnace running at high speed with a decommissioning line as an example. This is a complex system with 121 furnace rollers and 380 radiation tubes. The AI master created by Baosteel Co., Ltd. processes more than 800 data and parameters to predict changes in process parameters for the next 30 minutes. The AI master will directly send adjustment instructions to the PLC control system to complete real-time control at the millisecond level.

Through the application of AI, not only has the production capacity level been improved, but the closing rate of the unit's temperature mismatch, the frequency of strip deviation, and the duration of a single strip run have all been significantly improved. Even, in a specific case of tension correction, the AI detected the anomaly 54 seconds earlier than the operator and responded, and the AI's adjustment actions were consistent with mature historical experience expectations.

In terms of large models, Baosteel Co., Ltd. has been tracking and studying the development of large models for the past two years. This year, it has entered the actual combat stage of large models. The goal is to further improve the ability to inspect the surface quality of automobile panels through large models.

According to Liu Decheng, Baosteel has established an AI computing power center, formed a digital intelligence team, and is currently concentrating data resources from various manufacturing bases to verify the effects of large models on the electrogalvanized automobile board production line. In the future, it will be promoted through the three stages of “basic model, industry model, and professional big model” to further improve the ability to inspect the surface quality of automobile boards.

“This year is the first year Baosteel Co., Ltd. comprehensively promoted the AI strategy. The goal is to achieve 100 scenario applications similar to AI, including AI production line control, AI visual recognition, and AI intelligent decision-making.” Liu Decheng said.

The translation is provided by third-party software.


The above content is for informational or educational purposes only and does not constitute any investment advice related to Futu. Although we strive to ensure the truthfulness, accuracy, and originality of all such content, we cannot guarantee it.
    Write a comment