Views: 0 Author: Site Editor Publish Time: 2023-06-13 Origin: Site
"The current era of artificial intelligence with large models as the core has come. We can learn from this concept and build a large battery model to reduce cost and increase efficiency for the whole life cycle of power batteries." On the afternoon of June 9, Ouyang Minggao, academician of the Chinese Academy of Sciences, said at the main forum of the 2023 World Power Battery Conference.
At present, China's power battery industry has entered a stage of high-quality development. While the industry is developing rapidly, it is also facing challenges such as slowing down the growth rate of installed capacity, reducing the gross profit margin of battery systems, and accelerating product iteration. In this regard, Ouyang Minggao pointed out that the battery life cycle intelligent technology is an important means and tools to solve related problems.
He said that the current era of artificial intelligence with large models as the core has come. Power batteries can learn from this concept to build a large battery model.
It is understood that the large model evolved from Transformer, the number of references can reach 100 million level. At present, the number of mainstream large models is between 10 billion and 100 billion, which can form intellectual performance and professional knowledge output.
"After collecting massive data, through pre-training, the formation of Tranformer and attention mechanism as the core of the 10 billion parameter large model." On this basis, the framework system has reasoning ability and can be used in different fields. Examples include ChatGPT for natural language and DriveGPT for smart mobility." "Therefore, we can also pre-train the neural network, and based on Tranformer, build our power battery pre-training large model, and apply it to power battery intelligent design, intelligent manufacturing, intelligent management, intelligent recycling and other links."
At the meeting, Ouyang Ming interpreted the intelligent technology roadmap of the whole life cycle of the battery in a high depth. He said: "In terms of intelligent design and intelligent batteries, we mainly use high-precision, multi-scale modeling technology and multidimensional sensing inside the battery; In intelligent manufacturing and intelligent equipment; Mainly rely on production line big data, advanced manufacturing technology, single intelligent and multi-machine collaboration; In terms of intelligent management and intelligent recycling, it is mainly based on large models and active regulation."
According to reports, in terms of intelligent battery design, China's power battery industry has gone through experimental trial and error, simulation drive stage, is moving towards the direction of intelligent automatic development. Intelligent automatic design includes two core technologies: high-precision modeling and efficient intelligent optimization algorithm. It can establish the exact structure-activity relationship between design parameters and core performance, and automatically find the optimal and fastest path for the design process. This technology can improve the efficiency of battery research and development by 1 to 2 orders of magnitude, and save 70% to 80% of research and development costs.
The intelligent battery manufacturing process can be realized through process digital twin technology, defect intelligent monitoring technology, and production line big data analysis technology. Ouyang Minggao introduced: "The process digital twin technology can promote the efficiency of process development, and is generally used in the manufacturing process of the front segment of the battery pole; Intelligent defect monitoring technology integrates the evolution mechanism of battery defects and artificial intelligence technology, which can make battery quality monitoring and management to a higher level, and is often used in mid-stage battery molding process. The production line big data analysis technology is used in the post-partition process, which can reduce costs and increase efficiency by fully mining the battery production line data for intelligent forecasting and decision-making."
Intelligent management is an indispensable part of intelligent technology in the whole life cycle of batteries. "We can put sensors in the battery to sense, evaluate and predict the temperature, potential, pressure and other conditions in the battery, and then manage the battery through a large model to further improve the safety, power and durability of the battery." Ouyang Minggao pointed out. Taking thermal runaway safety warning as an example, it was very difficult to achieve thermal runaway safety warning in the past, because the thermal runaway fire accident of the power battery was relatively rare, and it was difficult to form large-scale data. Today, it is possible to generate a large database based on a small amount of data through artificial intelligence digital twin technology to achieve thermal runaway prediction and thermal reaction regulation.
Battery recycling also requires smart technology. Intelligent battery recycling includes intelligent disassembly, life extension and repair, reorganization and step utilization, monomer disassembly and material recycling. "We can do non-destructive repairs with smart technology, and we can also make predictions about battery life." Ouyang Minggao said.
Ouyang Minggao said that with the arrival of the era of artificial intelligence 2.0, the large model will greatly improve productivity, intelligent to the eve of rapid development, but the power battery industry still faces some challenges in the process of intelligent development of the whole life cycle, such as data scarcity, how to integrate with the new electrochemical system development. "Electrochemical systems continue to iterate and upgrade, and it remains to be studied how the large battery model can be quickly applied to new systems such as all-solid-state batteries." Ouyang Minggao said.
