Machine Vision helps Smart Manufacturing of Lithium Batteries

The era of large-scale manufacturing of power batteries has come, and inspection is becoming stricter and more complex. Market requirements are becoming more refined, and it has become a common challenge for the industry to improve production efficiency and reduce product defect rates.
For example, extreme manufacturing requires the power battery defect rate to improve from ppm (parts per million) level to ppb (parts per billion) level and safety performance control from 6 sigmas to 9 sigmas.
2. Slow iteration of machine vision technology
Diversified application scenarios have accelerated the differentiation of power battery demand, the choice of materials and processes varies widely, new materials such as lithium manganese iron phosphate, silicon-based anode, high nickel ternary, etc. have been launched, and the application of new processes such as large cylindrical batteries, CTC and CTB has accelerated.
At the same time, new challenges of machine vision inspection arise. New materials and new processes are bound to bring new defects, machine vision learning ability directly affects the production time. At present, the speed of technology iteration of some machine vision companies is slow, with weak product learning ability, it takes up to a week or two months to achieve stable inspection. This does not meet the expectations of rapid growth in production capacity for new energy companies.
3. Weak data feedbacks
In the lithium machine vision inspection industry, the inspection data has not been better utilized since the beginning. One reason for this is that the industry has not formed a unified defect standard, the same battery factory using multiple machine vision inspection systems, which has a large difference in inspection and inconsistency; Secondly, the data is inadequate to support the use of capacity.
- Introduced lightweight semantic model and jointly developed G-Box intelligent platform with Nvidia. The platform combines embedded development efficiency, and GPU neural network acceleration features, and can achieve ultra 2GB/s processing bandwidth, which can well satisfy the requirements of battery industry inspection speed and accuracy and match extreme manufacturing.
- The algorithm is given a large margin so that the machine has a strong response capability when facing new processes; Meanwhile, LUSTER is equipped with industrial data pre-training technology and special structure small sample model technology, so the machine can learn the defects from one sample, which achieves fast process iteration and strong learning ability.
- The R&D of a quality arbiter for the lithium industry, with a multi-dimensional and multi-angle stable imaging system, combined with the company's intelligent manufacturing deep learning industrial brain platform, can form a stable source of data, laying a good foundation for deep learning applications, enterprise production yield improvement, and subsequent process improvement.
Product Matrix:
Currently, LUSTER's machine vision inspection solutions have covered the whole process of lithium battery production, and plan to focus on raw material diaphragm inspection, laser cutting and cutting and stacking integrated inspection of electrode section and welding inspection of post-process section.


If you are interested in our lithium battery vision inspection solutions, click on the link below to tell us the solution you are looking for by leaving a message.
https://en.lusterinc.com/about-luster/contact-us.html
