[Application Case] Appearance defect detection of battery pack after blue film

Project background

In the TWh era, with the large-scale expansion of power battery production capacity, battery companies have increasingly high requirements for battery product safety, performance, and quality control. Among them, the requirements for improving the production efficiency of power batteries and extreme manufacturing have put forward higher requirements and challenges for machine vision in terms of detection accuracy, detection speed, image transmission, defect analysis, and other aspects.

In the process flow of power batteries, blue film serves as an insulation material that can separate the battery cells from each other, blocking the impact of various faults on other cells caused by a single battery cell, and preventing "all damage and loss"; Secondly, it can prevent surface scratches and leakage of the battery during subsequent transportation and assembly processes; In addition, it can also play a waterproof and dustproof role, thereby better protecting the battery.

Project difficulties

Due to the fact that blue film is a highly reflective material, its optical properties make many defect features unclear, resulting in extremely high kill and miss rates in traditional algorithms. At the same time, the thickness of the blue film is relatively small, only between 0.015mm-0.20mm, with a wide variety of defect types, and some samples of defect types are difficult to collect.

The traditional 2D vision based detection scheme in the past can no longer meet the increasingly stringent detection requirements of power batteries, and the limitations of traditional algorithms for defect recognition also hinder the digital pace of the lithium battery industry.

Based on years of process accumulation in the lithium battery industry, Huahan Weiye has launched standardized solutions for complex testing scenarios in the industry, effectively helping battery manufacturers achieve quality control and yield improvement.

Project highlights

HANSWELL integrates 2D+2.5D+3D+AI to achieve image level feature fusion and meet the online accurate detection of subtle and low contrast defects.

In the detection process, some defects require accurate delineation of the defect contour to distinguish their defect characteristics. To address this difficulty, HANSWELL uses 3D+AI detection technology to automatically resample defects such as pits and wrinkles, obtaining depth information of the defects. This can improve the attention of small defects and achieve accurate detection of fine defects.

For the detection of defects such as small bubbles and pockmarks, Huahan Weiye uses 2.5D+AI detection technology and a time-sharing flash detection scheme, achieving a zero missed detection rate and a misjudgment rate of less than 1%.

Project Summary

As an important component of intelligent manufacturing, machine vision can ensure and improve the production efficiency and quality safety of lithium batteries to the greatest extent, enabling lithium battery manufacturers to accelerate their adaptation to the rapidly expanding market.

The 2D+2.5D+3D+AI testing solution configured by HANSWELL leverages product advantages such as ultra fast scanning speed, ultra-high precision, coverage of a wider field of view, and high cost-effectiveness. It covers multiple process steps from battery cells to modules, battery cell top cover welding inspection, battery cell appearance inspection, module welding, and PACK assembly, creating a visual overall solution for power batteries for customers in the lithium battery industry, helping lithium battery enterprises achieve precise quality control, project implementation, and empowering flexible intelligent manufacturing of lithium batteries.