Machine Vision + AI | Lingyunguang’s innovative solution overcomes the problem of lithium battery burr detection

In recent years, frequent fire accidents in the fields of electric vehicles and energy storage have triggered large-scale recalls, and the responsible parties have affected battery companies. At the same time, with the entire industry under heavy pressure of overcapacity and price wars intensifying, battery companies are increasingly demanding high-quality control and safety in the production process.

Pictures come from the Internet

Small problems, big risks

According to the IEEE1725 battery cell safety performance audit and testing standards, burrs are metal protrusions that extend upward or downward from the foil to the outside of the pole piece, taking the foil (i.e., the uncoated area) as the baseline. ; Or it is a coated area, a metal protrusion extending upward or downward from the coating to the direction outside the pole piece; measuring the burr size is to measure the height of the burr in the vertical direction.

The content that needs to be rewritten is: glitch diagram (note: the diagram comes from the Internet)

Too large burrs can easily

pierce the diaphragm

, causing the positive and negative electrodes to contact and cause a short circuit . If a short circuit occurs when the battery is fully charged, a large amount of heat will be immediately released inside, causing an explosion, seriously endangering the safety of consumers' lives and property. Although the burrs are small, the hidden dangers are huge

There are many demands and detection becomes difficult

In actual production, in order to make batteries of different sizes, the pole pieces after coating and rolling need to go through processes such as slitting and punching. Taking the slitting process as an example, the physical and mechanical properties of the pole piece itself, the amount of overlap of the paired tools, and the wear state of the cutting edge will all affect the generation and size of burrs. From the perspective of testing requirements, if we want to comprehensively control product quality, we must not only

detect whether there are burrs, but also make different judgments

based on the size of the burrs. From the perspective of detection methods, pole piece burr detection mostly uses visual microscopes for offline detection, but even using a large depth-of-field microscope still has problems with insufficient depth of field and insufficient magnification. At the same time, in order to complete the measurement function, multiple systems need to operate at the same time, which makes the operation cumbersome and increases the cost.

The burr detection process of battery companies has strict requirements on the existing detection solutions and cannot achieve the goal of efficient manufacturing

Innovative solution "Machine Vision Artificial Intelligence" to solve difficult problems

As one of the earliest domestic companies to enter the field of machine vision, Lingyunguang has always adhered to the R&D concept of independent innovation, and has successively overcome many problems such as computational imaging and precision imaging, constantly transforming cutting-edge R&D results into productivity and also into Implemented solutions to solve customers’ actual problems. Faced with the problem of burr detection, Ling Yunguang launched a lithium battery burr online full inspection system and used "machine vision AI" solutions to solve industry problems.

Innovative optical solution: When pursuing speed and accuracy, we must take into account the balance between the two

In order to complete defect detection without affecting the overall production capacity, online full inspection is undoubtedly the best solution. To stably detect micron-level burr defects in high-speed production and high-vibration environments, factors such as space and investment must be comprehensively considered, which is very difficult.

High-speed imaging with high-speed cameras as the core can closely link the spatial and temporal information during high-speed motion changes for image recording, so it can effectively detect transient or high-speed motion situations. It can continuously record and quickly transmit signals to image processing equipment for corresponding operations. It has been widely used in industry, military, transportation, medicine and other fields.

Ling Yunguang has migrated and applied this technology to the field of pole piece burr detection. At the same time, combined with

's self-developed optical system, it realizes adaptive focusing

, improves the system's depth of field, and achieves high-resolution, large field of view, and large depth of field high-speed imaging, achieving 150m/min speed Stable and clear imaging, complete inspection of pole piece edge burrs,

detection resolution 2μm.

Adaptive focus animation demonstration

In addition, we can also complete high-precision full inspection offline with a resolution of 1μm

during the process quality inspection process such as quality analysis and process traceability of the pole pieces.

Deep Learning: Improving Detection Accuracy

F.Brain is a deep learning platform independently developed by Lingyunguang for industrial quality inspection scenarios, and has been widely used in the field of lithium battery testing. In the burr detection scenario, defects are first screened and classified by traditional algorithms, and the detected defects are segmented and then sent to the classification algorithm. For high-speed and low-cost scenarios, F.Brain has developed a self-developed multi-scale lightweight model, which reduces inference time by 35% compared to general models on the market. The detection speed can reach 150m/min, and the false alarm rate is less than one in 100,000, meeting the strict requirements of customers.

Software interoperability: seamless connection

For battery companies, detecting and handling defects is only the first step. Full production line management and full process management are the development direction to further improve product quality and ensure product safety.

The burr online full inspection system also has functions such as labeling linkage, alarm and anti-fooling, and can be linked with the company's own MES to interact with data centers and terminal storage to meet The retention period of enterprise defect information is required to achieve data traceability and traceability.

In the face of new changes and new demands in the lithium battery industry in the future, Lingyunguang will maintain its technical foundation, deeply explore customer needs, and use innovative products and solutions to help customers demand quality in every detail and in every link. Strict management to achieve the goal of zero complaints!

The above is the detailed content of Machine Vision + AI | Lingyunguang’s innovative solution overcomes the problem of lithium battery burr detection. For more information, please follow other related articles on the PHP Chinese website!