2D AI AOI Equipment for High - precision Detection of Board - Level Silk Screen/OCR Characters

In the electronics/PCBA industry, the quality inspection of board-level silk screen/OCR characters is crucial. WeLinkirt's 2D AI AOI equipment brings new breakthroughs to the quality control of this process with advanced technology.

User scenario: In the PCBA production line of a leading electronics manufacturing company, after the board-level silk-screening process, the silk-screen content and OCR characters on the products need to be inspected. The inspection objects include various logos, texts, and numbers on the circuit boards. These silk-screen and characters are important bases for product information transmission and function recognition.

Pain points: In the general trend of combining industrial visual inspection and intelligent manufacturing, traditional inspection methods face many difficulties. Manual inspection is inefficient. The inspection time for each circuit board is long, resulting in a slow production line beat and high labor costs. At the same time, the manual missed-detection rate is 3%, and the false-alarm rate is about 20%, which is difficult to meet the quality control requirements of large-scale production. In addition, when changing the production line model, it takes 30 minutes to readjust the inspection standards and processes, seriously affecting production efficiency.

Technical Principle

WeLinkirt's 2D AI AOI equipment uses high-resolution 2D imaging technology. Its hardware is equipped with high-precision industrial cameras, which can capture clear images of the silk-screen and characters on the circuit boards. The deep-learning secondary image-judgment algorithm is the core of the device. It constructs a visual basic model through a large number of positive-sample learning (1-20 good samples). During inspection, the image is first analyzed preliminarily, and then the model is used for secondary image judgment to accurately identify the defects of the silk-screen and characters. The semantic false-alarm filtering function is based on the understanding of image semantics, filtering out false alarms caused by factors such as the environment and illumination. The effective combination of this algorithm and imaging principle is because the deep-learning algorithm can continuously learn and optimize to adapt to different silk-screen and character features, while the high-resolution imaging provides an accurate data foundation for the algorithm.

• High - resolution 2D imaging provides clear image data for subsequent analysis.
• The deep-learning secondary image-judgment algorithm accurately identifies defect patterns by learning positive samples.
• Semantic false-alarm filtering reduces false judgments of non-real defects based on semantic understanding.
• The collaborative work of the algorithm and hardware improves the accuracy and efficiency of inspection.

WeLinkirt's Solution and Product

The solution centered around the 2D AI AOI equipment provides an efficient quality control method for board-level silk-screen/OCR character inspection. The device has the ability of high-speed online full inspection and can detect defects in silk-screen and characters, such as missing printing, wrong printing, and blurring, with micron-level accuracy. In the implementation, the DaoAI AI AOI software system is used for programming. Only 5 minutes are needed to complete zero-code automatic programming for a good product, greatly shortening the programming time. At the same time, with the APDT positive-sample/few-sample learning function, only 1-20 good samples are required to complete model training, quickly adapting to different product model changes.

WeLinkirt's 2D AI AOI equipment brings an efficient and accurate solution to the board-level silk-screen inspection in the electronics/PCBA industry.

Quantitative results: After using the 2D AI AOI equipment, the detection rate has increased to 99.2%, and the missed-detection rate has decreased to <0.8%, effectively ensuring product quality. The false-alarm rate has been reduced by -75%, reducing unnecessary reinspection work. The production-line model-change time has been shortened from 30 minutes to 5min, greatly improving production efficiency.