In the large-scale production of lithium batteries,the new requirements of”extreme manufacturing”,”minimal manufacturing”and”intelligent manufacturing”put forward by power battery enterprises also pose new challenges to AOI inspection of machine vision manufacturers.
“Intelligent manufacturing must have the same frequency of thinking as the factory,and reverse the possible solutions from the actual difficulties”.
In the face of the new trend of lithium battery industry development,Guo Jiayuan,General Manager of Huahan Weiye,also put forward his own response to the new challenges at the annual meeting of lithium battery.
In the view of Guo Jiayuan,General Manager of Huahan Weiye,industrial change comes from the needs of the times,and the development of manufacturing industry cannot be separated from the promotion of new technologies at all stages.In the TWh era,new technologies represented by AI began to promote the development of manufacturing industry in the direction of intelligence.
In industrial production,compared with other industrial sectors,the lithium battery industry has a higher degree of automation and a good data base,but it is greatly affected by the environment,light and material.Some key processes,such as the welding quality inspection of the battery top cover,the defect inspection of the seal pin weld bead,the appearance inspection of the soft package battery,and the defect inspection after the battery film,have been the pain points and difficulties of the industry inspection.In order to ensure the performance of the product,The yield of visual inspection must be close to 100%.
At the meeting,Guo Jiayuan,General Manager of Huahan Weiye,proposed a machine vision detection technology that combines traditional 2D/2.5D/3D imaging technology with AI deep learning,which solved the pain and difficulty of lithium battery industry.
In the detection of weld bead defects of sealing nails,the vibration correction algorithm is used to correct the image.Based on the image analysis technology of heterogeneous data fusion,the texture information of 2D image+the shape information of 3D image are fused for deep learning detection,which can avoid the machine vibration making the scanned image appear wavy,resulting in 3D visual detection NG.
In the same way,in the detection of top cover welding,the position of R corner is classified and determined by using the depth learning of heterogeneous fusion.The traditional algorithm+depth learning(secondary determination)can solve the problem of weld bead scale noise.The imaging of the starting point is very easy to cause the traditional algorithm to detect the error.It can detect the 0.1mm pinhole,the over-kill is less than 0.5%,and the missed detection is 0%.
In the case of defect detection of square shell core after membrane,the integrated application of 2D+2.5D+3D+AI and the detection scheme of multi-angle imaging+AI are adopted,so that the polar lug fold can always be clearly imaged in a certain light source at different angles,without overexposure or underexposure,to ensure the accurate identification of defects.
At the same time,in the case of the lithium battery industry he shared,it can also be seen that the technology of industrial vision in automatic production has gradually developed from the earliest 2D detection,to 3D detection,to today’s AI intelligent detection,from single scene,such as quality inspection,to the empowerment of the whole production line.
“Different downstream customers have different needs,but the same kind of scenario still has many commonalities to be refined”.
Guo Jiayuan,general manager of Huahan Weiye,said that the production processes of seal pin welding quality inspection,soft battery inspection,and battery pack blue film inspection are different,but the underlying technical logic has similarities.Based on the same technical logic,the algorithm can be integrated into different modules in a systematic manner,which is easier to summarize commonalities and experience,and thus improve efficiency.This solution has been verified in cooperation with a well-known integrator in the lithium battery industry.
It is reported that in the project implementation stage,it is not simply to add a set of AI visual detection system to the production line.Before the equipment enters the factory,the R&D personnel of Huahan Weiye will go to the factory for field research and find out each production link,so that the equipment can be seamlessly connected and started from the factory,and maintain continuous iteration of functions.Huahan provides an industry solution integrating software and hardware,which can get through data acquisition It is analyzed that the full-chain flexible customization precipitated by the algorithm has the ability of fast scene adaptation.
Based on the same underlying logic,we provide universal products and solutions for the same type of enterprises,and copy them to each product line.In two weeks,Sino-Weiye completed the data collection and analysis,model training,and delivered the defect detection and process analysis products in one month,with the effect exceeding the customer’s expectation by 20%.
Facts have proved that the exploration and innovation of Huahan Weiye in the key technology of machine vision is based on the self-developed algorithm platform.Whether it is the application of a single scene or the cross-link or the combination of multi-technology solutions,the traditional vision detection+AI detection solution in the field of industrial manufacturing can really solve the problems that have troubled the manufacturing industry for many years,which also makes Huahan Weiye go more and more smoothly on the road of AI+intelligent manufacturing industrialization.
Huahan Weiye,founded in 2015,has been deeply engaged in the machine vision industry for many years.Based on the continuous accumulation of technology research and development,it has formed certain technical barriers in optics,machinery,electrical,algorithms,software and other aspects,and formed the advantage of”optical,mechanical,computer and software”technology integration,providing strong support for the company to expand the market,and further enhancing the market competitiveness of the enterprise.
“With the penetration of key industries such as 3C and semiconductor lithium battery as the starting point,expand the solution capabilities of the”new display,automobile,photovoltaic and intelligent logistics and warehousing”industries,and further empower the entire industry.”
When talking about the direction of business layout,Guo Jiayuan said that with the higher requirements of intelligent manufacturing and precision machining for production processes and inspection standards,the 3D+AI vision system has also evolved towards a broader field of machine”vision”.These changes and requirements have opened up the application of 3D+AI vision and accelerated the wide implementation of 3D+AI machine vision in the manufacturing industry.
With the continuous deepening of AI technology in the manufacturing industry,factories will gradually realize the”three modernizations”of intelligence,automation and digitalization,and eventually lead to the emergence of”smart factories”.In the process of industrial intellectualization,as a supplier of machine vision software and hardware integration products and solutions,Huahan Weiye will also resonate with the downstream industry in the process of serving customers,actively carry out technical changes,improve the ability of automatic adaptation platform of hardware equipment,provide flexible customized detection solutions for customers in different fields and different scenarios in a short time,and broaden the boundaries of customer coverage,Realize truly fully compatible intelligent jobs.
In the global competition pattern,with continuous iterations and a vigorous visual revolution in progress,the low-cost and high-performance machine vision defect detection scheme will be an important part of intelligent manufacturing,and is developing at a high speed.”Big water and big fish”,both industry and investment circles,are giving AI+intelligent manufacturing positive feedback in this rapidly growing market,and actively layout the underlying AI system architecture capabilities,Enabling intelligent manufacturing and boosting the process of industrial intelligence will be the development goal of Chinese visual enterprises.
