디엘정보기술

Based on over 20 years of experience in building smart factories,
DL Information Technology Co. Ltd. is creating manufacturing big data analysis
and AI application cases in various industrial fields

AI/Big Data Construction Case _
AI-based Semiconductor Cleanroom Management Predictive Maintenance Function Application Case

DL Information Technology Co., Ltd. provides the AI-based semiconductor cleanroom management predictive maintenance function

Raw data Analysis environment/Technology Data collection/Loading Visualization

Raw Data

1Real-time air conditioning system data
2Collection of real-time temperature/humidity/
particle sensor data
3Collection of real-time differential pressure
sensor data

Analysis environment / technology

Time Series Forecasting Analysis

Analyzing big data with AI, predicting and alerting cleanroom environment that affects product completion

Saving equipment and cleanroom environment information in DB

Entering equipment and cleanroom environment information into learned AI and performing predictive maintenance

Applying collected temperature/humidity, differential pressure, particle (3㎛, 5㎛, 10㎛) time series data
to SARIMAX machine learning algorithm to predict
short-term (20 minutes) and long-term (1 hour) future

Data Collection / Loading

Collected data: Information on operation of real-time clean room environment control equipment; Real-time temperature/humidity, differential pressure, particle sensor data

Collection method: Collecting through RS-485 Modbus; Collected data is loaded into MS-SQL DB in the form
of time series

Visualization

Real-time prediction of temperature/humidity/
differential pressure/particle environment
Monitoring operation of cleanroom control system

Paint Point

Remote collection
not possible
Check cleanroom
every 3 hours
Late problem
recognition time
High defect rate

Reduced work efficiency due to manual collection of cleanroom environment data: Workers check the cleanroom status every three hours and manually write down the result because remote collection
of cleanroom environment data is not possible; Three hours from time occurrence of cleanroom environment problems to problem recognition; Products manufactured during cleanroom environment problems have a high defect rate, which lowers productivity

Introduction effect

Automatic collection of cleanroom
environmental data

Reduction of simple and repetitive tasks for workers through automatic collection
of cleanroom environmental data

Reduction of time required for detecting environmental abnormalities by monitoring real-time cleanroom condition
Process improvement through AI application

AI predicts future temperature/humidity, differential pressure,
and particles every minute and detects environmental abnormalities in advance, enabling preemptive response
before abnormalities occur

Using tracking and improvements to the cause of particle occurrence as a basis for process improvement

Contributing to corporate ESG management by minimizing unnecessary control system operation through monitoring of temperature/humidity and clean room control system operation

Client company

REFERENCE

Raw Data

Analysis environment / technology

Data Collection / Loading

Visualization