Founded in 2007, UDMTEK is a company that develops smart factory operating systems based on AI and digital twin technologies.

We analyze everything from internal control logic to sensor data, supporting root cause analysis, predictive maintenance, and quality improvement.

Traditional diagnostics typically rely only on external observation values such as IoT sensor data. In contrast, UDMTEK’s solution not only uses these external values but also interprets the internal control logic in real-time, analyzing how sensor data behaves in direct relation to control program execution. Additionally, UDMTEK has a unique technology that allows fast and detailed collection of even difficult-to-access external signals from a wide range of equipment. This combination of deep control logic analysis and comprehensive data acquisition provides a much more accurate and insightful diagnosis than traditional methods.

A virtual model that digitally reproduces the control, operations, and sensor flows of a real factory to simulate and improve processes. UDMTEK defines it as creating new value through digital representation, transformation, or modeling of a target phenomenon. The digital twin evolves with expert feedback and real-time data.

MLP refers to the technology that interprets machine control programs and links data flow with program execution to perform analysis.

Applicable to nearly all manufacturing sectors including automotive, secondary batteries, electronics, display, shipbuilding, and plant industries.

UDMTEK can collect and standardize data automatically from various controllers like PLCs and PCs. Even with outdated models lacking standard communication protocols, we can collect meaningful information through custom development. If protocols are unavailable, we install our edge-type DAS (Data Acquisition System) to gather data from key sensors.

Simply install the Edge system and link it with existing control programs and data. It can be applied without disrupting or stopping the production line.

All factory-related data, including control signals, sensor data, operator inputs, and camera footage.

Collected data is sorted chronologically and compared with AI-learned normal patterns to detect anomalies.