The trend of "full stack" machine vision is gradually emerging, and AI empowers the intelligent manufacturing transformation of lithium battery production lines

The era of artificial intelligence has arrived, forging the intelligent capabilities of the entire lifecycle of the lithium battery industry has become a new means for enterprises to reduce costs, increase efficiency, and enhance competitiveness.

From the perspective of the production and manufacturing process of lithium batteries, defects such as burrs, scratches, exposed foil, bubbles, wrinkles, dark spots, bright spots, and material dropout generated in various stages such as battery materials, battery cells, and PACK packaging will seriously affect the performance and quality of lithium battery products. Battery companies are more demanding on the accuracy, speed, image transmission, and defect analysis of production line processes.

Against the backdrop of large-scale manufacturing in the lithium battery industry,Promoting the intelligent iteration of machine vision technology, taking battery quality control to a new level, has become an inevitable requirement for lithium battery enterprises to move towards large-scale intelligent manufacturing. It is also an inevitable choice for lithium battery production enterprises to solve quality and efficiency problems and achieve upgrading from ppm to ppb.

Based on this, machine vision enterprises with core competitiveness and the ability to respond to lithium battery intelligent transformation in a one-stop manner will have huge development opportunities.

Following the development trend of the lithium battery industry, through continuous technological innovation and industrial verification,HANSWELL has officially released the iSense AI intelligent visual inspection platform.

This platform is an end-to-end full process AI industrial quality inspection platform developed by Huahan Weiye based on deep learning systems, focusing on the industrial AI technology capability matrix. Based on the system platform of "1+N+X=∞", the iSense AI intelligent visual inspection platform can provide accurate, efficient, and intelligent inspection services for the entire process of lithium battery production.

By combining multiple modalities and multiple application scenarios, and iterating through the underlying algorithm logic of Huahan Weiye, over 10000 visual system deployment solutions can be formed in different industries, providing deep and customized services for machine vision detection in various enterprises.

By creating a full stack/one-stop intelligent machine vision solution for the production process of lithium batteries, Huahan Weiye iSense AI will also greatly improve the production efficiency and quality control capabilities of related enterprises, empowering the intelligent transformation of lithium batteries throughout their lifecycle.

The Evolution of Machine Vision Intelligence

Under the development requirements of the high-end and cutting-edge terminal industry, the demand for intelligent machine vision is becoming increasingly apparent.

As a trillion level track, the lithium battery industry is undergoing intelligent manufacturing transformation under large-scale delivery, and its strict quality requirements support the evolution logic of machine vision intelligence.

HANSWELL empowers the intelligent transformation of the lithium battery industry and has continuously iterated its underlying algorithm logic and deep learning network in machine vision.

this,HANSWELL has officially released the iSense AI intelligent visual inspection platform, which is in line with the quality requirements of the lithium battery industry, shifting from the PPM level to the PPB level. Based on its unique advantages in the field of visual algorithms, it solves multiple pain points in machine vision terminals for lithium battery production.

In the exploration of cutting-edge AI technology, the HANSWELL iSense AI intelligent visual detection platform constructs imaging technology from multi angle images, light and dark field sequence number images, multispectral photometric stereo images, RGB-D multimodal data, and phase offset images, adapting to different detection needs and accurately tracing the source of scratches, damages, wrinkles, and other problems.

In actual industrial validation scenarios,HANSWELL has accumulated high-precision algorithm models, which can quickly iterate machine vision detection models while enriching product functions.

For example, through intelligent defect area sampling, iSense AI can target defect area targets for ultra-high resolution segmentation algorithms, thereby improving the ability to extract defect features; Built in pre trained word library and universal model in OCR, achieving an accuracy rate of over 99.9%; In 3D anomaly detection, based on point cloud normal sample data, the detection and recognition of abnormal defects can be achieved, enabling precise analysis of subtle defects.

On this basis, the new generation iSense AI intelligent visual inspection platform of HANSWELL has built five core advantages, promoting the upgrading of industrial quality inspection.

①Continuous learning:Based on the characteristics of deep learning, learn new product forms based on new data, and reduce catastrophic forgetting.

②Rapid deployment and replacement of production lines:Support one click transfer learning to achieve rapid replacement and deployment of multiple models and types of workpieces; Visual scheme prototype verification, shortening the time cycle for visual project integration and deployment.

③Noisy learning:Improve learning robustness under high noise ratio; Automatically correct and eliminate abnormal data.

④Automatic defect sample generation:Generate defect samples to meet the needs of small sample learning. Enhance the richness of the sample and reduce the long tail effect of the data.

⑤Lower sample dependency:<200 samples are required, with a detection accuracy of>99%; Unsupervised learning, only requiring good products to meet the needs of anomaly detection; Super resolution image, non-uniform consistency sampling.

Building a full stack solution for machine vision

In response to the demand for large-scale manufacturing of lithium batteries, the role of machine vision inspection is increasingly prominent in forging stronger quality delivery capabilities throughout the entire lithium battery production chain.

From past process experience, it can be seen that due to the different process requirements and testing types in each process of battery manufacturing, visual inspection enterprises not only need to have technical knowledge in optical imaging, algorithm processing, and software architecture, but also need to have a deep understanding of different process stages.

HANSWELL has previously completed the full process visual inspection coverage of production, battery cells, modules/PACK< On the new generation of iSense AI intelligent visual inspection system, a systematic platform is built to further upgrade the entire process and form a full stack/one-stop visual inspection solution.

HANSWELL iSense AI intelligent visual inspection platform, based on deep learning, can achieve end-to-end full process AI industrial quality inspection by creating cross product form models.

Compared to traditional visual inspection, industrial problem detection and traceability are carried out in different processes,The HANSWELL iSense AI intelligent visual inspection system can provide one-stop visual inspection services for different battery production processes.

The iSense AI intelligent visual detection system creates a "1+N+X=∞" full scene adaptation capability, which includes one platform, N modes, and X application scenarios.In the multimodal field, HANSWELL can form different combinations of 2D+AI, 2D+3D+AI, 3D+AI, etc., adapting to different segmented scenarios, and forming an efficient and accurate intelligent detection service system.

Relying on the combination of "1+N+X" multimodal characteristics and underlying algorithms, the iSense AI intelligent visual inspection system has formed over 20000 visual system deployment solutions, providing deeply customized solutions for defect detection problems in different industries including lithium batteries, semiconductors, 3C, etc.

The production and manufacturing of lithium batteries is the touchstone of the Huahan Weiye iSense AI intelligent visual inspection platform. The quality demand under large-scale delivery puts forward extreme efficiency requirements for industrial quality inspection platforms.

But iSense AI's ambition goes far beyond that.Under the evolution of industrial intelligence transformation, the deep integration of the industrial chain driven by intelligence has become the empowerment of the ecological chain and value chain.

Ouyang Minggao, an academician of the CAS Member, said publicly that China has the world's largest power battery industry, but so far, it has faced a series of challenges, such as slowing growth, reducing profit margins, accelerating product iteration, and so on. The solution to this is to achieve full lifecycle intelligence of batteries based on large models.

Starting from the iSense AI system itself, with the goal of intelligent transformation, a closed-loop process has been established for the environment, products, and equipment. The iSense AI system builds a data circulation loop from the entire production process of lithium batteries, which not only greatly improves the reaction speed, defect inspection efficiency, and human efficiency of quality inspection adjustments, but also promotes high collaboration between upstream and downstream in the production process, becoming a key part of intelligent modification throughout the battery life cycle.

In the continuous upgrading competition of the industry, empowering the full stack/one-stop intelligent transformation of lithium battery production, HANSWELL will also unleash greater economic benefits.