CONNECT WITH US

AI visual inspection in electronics manufacturing

News highlights

Managing Production Lines With Stopwatches Was Time-Consuming and Much Less Efficient (Advantech)

With over 35 years of experience in electronics R&D as well as manufacturing, Might Electronics, launched a new plant in October 2021, which expanded its production capacity in Taiwan and also introduced the PowerArena AI vision inspection program, powered by the Advantech SKY-6400 industrial GPU server to their motor assembly line.

Managing Production Lines With Stopwatches Was Time-Consuming and Much Less Efficient

The company's previous motor assembly line relied on IE engineers conducting regular on-site visits with stopwatches to record the cycle time of each workstation. Due to limited manpower, IE engineers could only record about five cycles at each stop. Furthermore, stopwatch records contained only numbers without visual recordings, making it very difficult to discover the real reason behind excessive completion times, not to mention finding the right solutions.

In addition, the manual collection of cycle time data was an enormous burden for IE engineers. It took 1 to 2 working days just to collect data, which then had to be adjusted to confirm the cycle times of each station, leading to troublesome and time-consuming workloads.

Solution

Might Electronics introduced the PowerArena AI visual inspection program, which has cameras set up at each workstation to collect images and send them to the Advantech SKY-6400 server for AI analysis. Cycle times can then be calculated instantly by referring to production times and quantities for each workstation and operator.

If the cycle time of a workstation shows an abnormality, the system will actively alert the managers to pinpoint the cause, and by replaying images of the abnormal workstation, managers will understand the cause of the abnormality. Ray from Might Electronics stated that "With real-time production line data, Might Electronics can dynamically adjust production line resources, optimize on-site manpower allocation, and ensure that the production line is always in balance. This allows us to improve the overall production performance."

Gavin from PowerArena added that, when implementing the PowerArena AI visual inspection program, camera angles were specially adjusted to ensure that the faces of the workers are not recorded as it only focuses on hand movements. The company emphasizes communication to ensure the workers understand the introduction of AI is meant for quality improvement, rather than stricter labor supervision.

PowerArena AI system diagram

System Diagram (Advantech)

Three Major Features of AI Visual Inspections: Efficiency, Quality, and Comprehensive Production History

Firstly, the overall operation efficiency at Might Electronics was improved with the implementation of the AI system. In the past, IE engineers had to establish specific SOPs and countless working hours were spent on trivial tasks in order to collect data. Now, after setting up SOPs and completing disassemblies, engineers can obtain cycle time statistics from a computer. The process is not only faster but also includes larger sample sizes. Now, up to a whole day's data are available to assess each workstation. Furthermore, the time saved by using the AI solution can now be dedicated to planning production line optimizations and other improvements.

Second, product quality can be verified. Ray pointed out that for quality control, Might Electronics used to mainly rely on first-article inspection and production end sampling. Since first-article inspections can only be done independently by on-site personnel, uncertainties are inevitable. As the AI records the full processes of each product at each workstation, it will actively notify managers in real time when any abnormalities occur. The new method is similar to treating every product as the first product to be inspected, allowing for much greater control of product quality.

Third, a comprehensive production history is established. Thanks to the PowerArena AI visual inspection program, the complete visual records of each assembly operation can be kept. When the quality control staff finds a problem with a product or encounters a customer complaint, Might Electronics can trace it back to the records of the exact assembly operation within the system, confirming which part of the process caused the defect so that improvements can be conducted.

Advantech SKY-6400 Offers Both Performance and Scalability to Build a Solid Foundation for AI Applications

Might Electronics' success with implementing AI applications in their factories is thanks to PowerArena's AI algorithms and integration capabilities as well as the crucial hardware implementations. "At that time, the hardware envisioned by Might Electronics required good performance, long-term stable operations, and the hope of meeting AI computing demands with just one device. Advantech SKY-6400 was the perfect choice for our needs, and it was also offering the best price-performance ratio." Gavin from PowerArena explained that SKY-6400 is an industrial-grade server that can operate 24-7 with unquestionable operational stability. Furthermore, it is very scalable and can hold multiple additional graphics cards, making it the top choice for Might Electronics' future expansion plans for AI visual inspection applications.

For the future, Might Electronics is planning to use AI for SOP analysis. Such a system will require a large number of hardware resources. But with the scalability of Advantech SKY-6400, they will not be required to purchase additional main hardware when launching the system, as simply adding more graphics cards will offer the computing resources needed for SOP AI analysis.

Ray emphasized that in response to the trend of smart manufacturing development, Might Electronics has fully adopted the Wi-Fi 6 protocol in the new plant, laying a solid foundation for the introduction of AI applications. In the future, Might Electronics will also consider the possibility of AI applications from the perspective of personnel management and plant safety. The company aspires to accelerate the transition of its plants into intelligent factories, thereby enhancing Might Electronics' competitiveness in the global market.

To learn more about the complete case, please fill in your information in the form and we will send the complete case through your mail.