A forum has been set up to discuss how to achieve productivity by applying AI in research and development and manufacturing.

Experts have argued that the entire process from development to customer service must be connected as one process in order to provide digital twin technology that combines AI and other technologies to optimize processes in manufacturing sites.

On the 9th, the Korea Industrial Technology Association (Koita) held the 'Industrial Technology Association-Hankyung AI Economic Research Institute Joint Forum' on its YouTube channel and shared industrial AI application cases. LG Innotek Managing Director Kim Min-gyu presented LG Innotek's response cases in the materials and parts sectors due to uncertain changes.

In modern industrial settings, unlike in the past when businesses mass-produced by predicting demand, small-quantity production of a wide variety of products is required. In an uncertain environment, component life cycles are shortening and profit margins are shrinking.

Accordingly, the development of ultra-precision parts has become more customer-specific, increasing complexity and ensuring flexibility has become essential.
Managing Director Kim Min-gyu argued that “Crying curves, which generate profits through manufacturing alone, are not competitive, and ultimately, they must move to a service form that increases the added value of manufacturing, such as the Smile Curve.”

As a measure for this, the implementation of a digital twin was chosen, which can shorten the time by more than 30% compared to existing actions. This means that it should not stop at the general virtualization level but should be expanded to a cognitive digital twin.

In other words, it refers to the concept of developing customer services by integrating generative AI, etc. from implementing physical factories into the virtual world.

■ Manufacturing as a Service, ‘MaaS on Digital Twin’

LG Innotek proposes the concept of ‘Manufacturing as a Service (MaaS)’, which designs a platform and connects it to a manufacturing service.

Key features introduced include: △virtual prototyping service, △optimal design service, △optimal production service, and △on-time manufacturing service.

LG Innotek is continuing discussions to digitize the analog environment and move it to the intelligent stage.

Kim said that in order to implement a true cognitive digital twin, it must be supported by a 'physics-based AI' that reflects reality by combining causal and correlational data, a 'generative AI' that converses with humans and accumulates data to develop judgment capabilities, and a transition to digital processes.
'JEDFe (Joined Evolving Development Framework)' is a platform that connects LG Innotek's entire process from development to production and customer service into a single digital system.

Patents with AI There are applications such as 'InnoTAAPs' for judgment, 'InnoDesign' for design automation, 'InnoSimPL' for simulation automation, and 'InnoLAB' for real-time data acquisition, and recently, 'InnoMLOps' has been added as a data-based AI automatic learning tool.

The 'InnoSimPL' function systematically manages virtual verification and process execution processes and data, and can secure development quality early. All simulations can be provided with more than 90% performance within a day.

In addition, in an environment that integrates AI/ML development and operation, it is possible to utilize automated and intelligent AI solutions by building pipeline processes such as data collection, analysis, learning, and inference. For example, LG Innotek applied AI to optimize the design of customers' semiconductor drawings to identify vulnerabilities and block defects, thereby shortening the design time.

LG Innotek is providing the concept of digital process connection as a service to enhance customer competitiveness and suggest new values, and is currently in the initial stage with partners.

Finally, Executive Director Kim added, “AI absolutely needs a step-by-step strategy. In addition to solving time and accuracy issues, we need to accumulate experience in platforms and other areas so that AI can be applied to areas that collect and judge data as a cognitive digital twin.”