SEOUL, South Korea
Hankook Tire & Technology Ltd. claims to be pushing the envelope again on tire manufacturing, disclosing that it's using artificial intelligence (AI) and Internet of Things (IoT) technology to improve the accuracy of "anomaly detection" in the production process.
Hankook claims its AI/IoT-based facility abnormality prediction system, or Condition Monitoring System Plus (CMS+), will allow it to accelerate the development of "smart" factories.
The Seoul-based tire maker is using CMS+ in its factories in South Korea and is in the process of applying it to its factories in China, Hungary and the U.S.
This is the third form of AI-based technology that Hankook has disclosed in recent months as it focuses on developing innovative R&D and digital technology capabilities and to realize digital transformation.
The others involve automating the final inspection process for tires by applying AI- and digital-sensor technology to enhance efficiency and consistency and using AI in an approach called Virtual Compound Design System (VCD), which provides what the company calls "predictive models for tire compound properties" that can cut development process time in half.
Manufacturing abnormalities can bring production lines to a halt, Hankook said, and bringing production back on line can take hours, raising costs.
Some symptoms of these abnormalities include changes in the outputs, abnormal temperature rise, noise and vibration, etc., so identifying minor symptoms in real time and performing maintenance in advance can play a key role in preventing major facility failures, Hankook said.
Abnormality prediction systems typically use a vibration sensor attached to key components of the equipment, Hankook said, which then requires experts to analyze the information and determine if there are any symptoms.
Hankook's CMS+, on the other hand, uses a three-step AI algorithm that enables data analysis with precise prediction that the company claims is three to four times faster than existing systems.
By incorporating IoT capabilities, Hankook claims CMS+ collects and analyzes data every second and then sorts and stores automatically data suspected of abnormalities.
Hankook developed this original AI algorithm in cooperation with the Korea Advanced Institute of Science and Technology (KAIST), a respected science and technology university in South Korea.
CMS+ also collects different types of data together, including sensor data, temperature and operational information, to predict abnormal conditions of the facility in advance. In the event of anomalies, the system notifies the plant manager in real time to allow appropriate action be taken faster.
The company said it also is working on incorporating augmented reality (AR) to help identify data flow that is difficult to distinguish in the field.
As for the final inspection process, Hankook said it is developing the technology as proprietary, but said it believes it could turn the process into a stand-alone business that it would consider licensing to third parties at some point.
Hankook noted final-stage inspection involves three types of examination to detect possible defects: internal inspection using a type of shearography; X-ray inspection of a tire's internal structure; and external visual inspection.
In particular, Hankook said it believes AI-based technology can be brought to bear in shearography — which it refers to as Interferometer Tire Testing (ITT) — and X-ray inspections.
Currently, experienced technicians evaluate the ITT images visually, Hankook said. Using AI technology would allow the computer to find defects in nonconforming patterns systematically, thus accelerating the process.
Shearography, or interferometric measurement, is a non-destructive, contactless inspection process that compares two laser-based images of a tire — one at rest and one under vacuum — to determine if there are any anomalies in a tire's internal construction.
Hankook said it has collaborated with AI experts in the Department of Industrial and Systems Engineering at KAIST to develop and implement this system based on machine-learning technology.
"Hankook has been positioning itself as a digital leader," Hyunshick Cho, vice chairman and president of Hankook Technology Group, said.
"The development of the automatic inspection system is yet another feature aligned to such innovation, making it possible for us to secure a leading position in digital transformation in this fast-changing business environment."
Hankook sees the new technology as a way to maximize consistency and efficiency of final inspection while also reducing the decision-making time at this stage and thus improve overall plant efficiency.
"We will continue to pursue innovation and advance towards a global top-tier company," Mr. Cho said.
For now, the AI-based system is in place at one plant, the Geumsan, South Korea, car and truck tire factory, but the company plans to begin installing it at other plants worldwide after the Geumsan testing is finished in October, Hankook said.
The third application of AI, in VCD, is designed to give compounding recommendations to the developers, which helps them with the first-round verification process, Bon Hee Ku, senior vice president, chief technology officer and head of research and development, said.
"The information is then passed back to the VCD after the actual compounding, which is also then processed by the AI system as part of the big data. So, as data continues to grow, artificial intelligence will become smarter, leading ways to new possibilities to support the entire research development process," he said.
VCD predicts compound characteristics and determines the "most optimal combination of materials through artificial intelligence analysis," Hankook said.
Typical compound development can take anywhere from six months to three years using a traditional approach, Hankook said, but using AI in the process can help cut that time in half.
Work on the VCD project dates to 2015 and also involves working with KAIST. Cooperating with the university researchers yielded a significant improvement on data analysis accuracy, Mr. Bon said, which the company indicated is now over 95%.
"This year, we are doing test runs with various compound developers and will improve the AI system based on feedback. Starting next year, we will gradually apply the AI system into our actual compound development process. We plan to complete our Virtual Compound Design technology by 2023, and it will be an essential part of the tire development process," Mr. Bon said.
"Our top priority is always to develop and deliver the safest products for our customers. The value of VCD is that it adds efficiency to our current work and we will continue to use the latest technologies available to enhance future research projects," Mr. Bon said. "At the end of the day, we believe that incorporating both future and tradition is still the smartest way to go."
Jim Johnson, Tire Business reporter, contributed to this article.