Press Release
Korea Deep Learning transfers AI that predicts missile aerodynamics with 98% accuracy to Seoul National University
Korea Deep LearningPublished

On August 27, 2021, Korea Deep Learning signed an AI flow analysis technology transfer agreement with the Hypersonic Flow Laboratory of the Department of Aerospace Engineering at Seoul National University's College of Engineering. It handed over an artificial neural network model that predicts missile aerodynamics in real time, so that flow analysis, which highly skilled researchers used to spend a long time on, can be produced instantly at about 98% accuracy.
Aerodynamic prediction that took time and skill
Flow analysis is the work of solving fluid dynamics problems numerically on a computer. Before AI was introduced, the flow analysis used to predict a missile's aerodynamics took a long time, and it was difficult enough to require highly skilled researchers. The burden on the national budget was large accordingly.
An AI model that produces results instantly even when the inputs change
The technology Korea Deep Learning transferred is an AI model for predicting missile aerodynamics, trained on flow analysis data. It uses an artificial neural network algorithm to predict aerodynamics in real time from inputs such as flow variables and shape variables. It cut the time analysis takes, and repeated validation raised its accuracy.
The results come down to three points. First, it secured flow analysis accuracy of about 98% for aerodynamic design. Second, because the neural network has learned the variables, it produces aerodynamic performance instantly even when the input variables change. Third, it obtains data in real time, cutting time while raising accuracy. Because it is a standardized program, less experienced researchers can also take part in analysis work, which reduces both time and labor costs.
The technology is also expected to help obtain aerodynamic data efficiently in the optimal design of guided missiles. Being able to secure aerodynamic data quickly makes it possible to apply a variety of optimization techniques in the design process.
"I'm glad our AI solution is helping advance national security, and it means even more that AI can now handle work that only highly skilled people could do before. Building on Korea Deep Learning's cutting-edge technology, we will go on to create a new standard for artificial intelligence."
— Kim Ji-hyun, CEO, Korea Deep Learning
From text recognition to engineering analysis
The technology transfer showed that Korea Deep Learning's deep learning technology can be used beyond text and image recognition, in engineering analysis as well. Korea Deep Learning has put forward 99.9% accuracy for DEEP OCR, its optical character recognition (OCR) technology built on its own deep learning engine. Its high-speed OCR system for Korea Post logistics recognizes postal waybills in under 0.1 seconds, and its meter recognition system for the Korea Centers for Disease Control and Prevention digitized radiation exposure records and built predictive statistics.
About Korea Deep Learning
Korea Deep Learning (KDL), founded in 2019, is an enterprise AI agent company. Built on KDL Frontier, its own vision-language model (VLM) specialized for documents, KDL makes AI agents that take on and complete the document work of businesses and public institutions. Its main products are DEEP OCR, DEEP Parser, DEEP Agent and DEEP Ops.







