On-line Inspection Of Automobile Assembly Based on Deep Learning

Ling Yun Industrial

Industrial, automotive component manufactuer.

The cost of manufacturing testing is high with human capital. Manufacturers can only inspect partially (not all) of the products manufactured. The level of automation and equipment intelligence has been low.

Using thousand Mega high-speed wired transmission network, data is transmitted to the big data computing and analysis platform. At the same time, using 4 industrial cameras to take pictures of the finished product from the output machine at the end of the production line.

The maximum data flow of the camera is 29.5M/s. The data flow of the four cameras is 118M/s, the amount of image data generated per day is about 3T, transmitted on a gigabit wired network.

After the analysis and comparison on the online quality real-time judgment system, the comparison and analysis results are obtained.

• [Management Effectiveness - Operation]
  

  Reduce production scrap rates. Shorten production cycle. Improve the production management efficiency of enterprises.

• The efficiency has been improved, and the overall efficiency of the equipment has been increased from 65% to 70%.

• The quality has been improved, and the qualified product rate has increased from 95% to 98%.

• The inspection method has been changed from the original random inspection to full inspection, and comprehensive control of product quality - reduced testing personne by 2 person-days per line (10 person-years in total).