Accelerating the implementation of AI in Metal Processing Plants | Whuff News


Chin Fong Machinery Industry is a top five operator in the global stamping and forging machinery industry, with over 70 years of experience. It is also the largest professional mechanical punch manufacturer in Taiwan. It specializes in the manufacture of various types of machines, providing comprehensive solutions for stamping and forming for automobile sheet metal stamping, motor silicon steel sheet and computer chassis production lines.

As a pioneer in the Machine Industrial industry Chin Fong has, since 2015, created smart factory operations. Recently, the company began to incorporate AI solutions to implement lean production and develop smart equipment — and it was this need that led to the successful collaboration with ASUS IoT.

The challenge: Accelerate visual inspection of complex components

Industry 4.0 is a hot phrase, but it is not without substance. In manufacturing, this means combining the Internet of Things (IoT), digital factories, cloud services and communications to create ‘smart’ digital physical systems — introducing intelligence into production processes, and changing the business mindset of traditional manufacturing processes.

IoT technology enables machines to communicate with other machines, and with people, transforming traditional production methods into highly customized, intelligent and service-oriented business models. It also provides the ability to produce small quantities of products quickly, and respond to rapidly changing markets, to increase the competitiveness and profitability of enterprises.

Conventionally, defect detection on a press production line has demanded manual visual inspection. However, human inspectors lack efficiency and accuracy, especially for metal parts. Such components are particularly challenging due to reflectivity, meaning that parts often need to be flipped or rotated several times. It is important to understand the characteristics of optical surfaces and components to obtain accurate defect data.

In addition, the types of metal molds vary, further complicating human inspection. This is why automated optical inspection, or AOI, is increasingly being implemented in the field — and why Chin Fong Machine Industrial turned to ASUS IoT.

Solution: AI-powered visual inspection system from ASUS IoT

ASUS IoT AISVision is an easy-to-use software toolkit for AI-powered vision applications—and is ideal for monitoring metal stamping and plastic injection processes, and the assembly of electronic parts. The AI-powered training model means it can be quickly adapted to almost any visual inspection need, enabling accurate and efficient detection of a wide range of defects, including scratches, dents, dirt and more. It is also able to distinguish defects hidden in concentric circles and hairline metal parts.

Powered by these advanced capabilities, AISVision optimizes production processes and provides superior performance on anomaly detection, significantly shortens model training schedules from hours to minutes, and meets the need for rapid modeling and implementation in factories. Additionally, AISVision supports the Intel OpenVino framework for inference without additional GPU accelerators, which reduces hardware investment.

AISVision features fast, code-free modeling, with a training architecture that includes both supervised and unsupervised learning modes — so only a small number of samples are needed to build an accurate model.

ASUS IoT has developed an exclusive AI-powered visual inspection technology that leverages four functions. These include multi-object classification, accurate defect detection, fast object recognition and fast modeling anomaly detection, and provide high-speed and high-precision detection capabilities. Various data filtering functions can be performed to check the training status for different scenarios, and model validation reports are also compiled and exported in HTML format for further model training. A unique retraining mechanism ensures data privacy, and powerful API support is built right in, including C, C++ and C# — making program integration faster and easier.

The result: Drastically improved speed, efficiency and accuracy on the production line

Chin Fong Machine Industry has implemented various ASUS IoT AISVision solutions, installing the necessary cameras and lighting throughout its production line. Once part stamping is complete, image capture and inspection is performed immediately, with defective objects identified with new levels of speed and accuracy. Tracking accuracy has eliminated both human error and fatigue problems.

Compared to traditional AI-based projects, AISVision saves up to 80% of project development time, helping to achieve digital transformation for the smart factory of the future.

Partly as a result of its collaboration with ASUS IoT, Chin Fong Machines has transformed from an equipment maker to a systems integrator (SI), providing one-stop inspection services to help enterprise customers improve both profitability and competitive advantage. We continue to work together to use digital visualization technology to create the next generation ecosystem for the metal processing industry.

“We have focused on stamp industry applications in recent years, and we have implemented IoT and AI-based technologies as a framework to integrate stamp and forging operations and management issues. We look forward to working more with ASUS IoT to create high value for our customers through additional services and application software,” said Sheng-Ming Tseng, General Manager of Chin Fong Machine Industrial.

The ASUS IoT team believes that AI-based machine vision and edge computing are at the core of smart manufacturing. We provide clients with process optimization, cost reduction and improved process quality and efficiency, so that enterprises can increase revenue and create new business models. Through powerful AI technology, we offer customers intelligent solutions, accelerate the processing of big data, strengthen supply chains, and provide new benchmarks for AI-based manufacturing.



Source link