New Business

Abnormality Sensing-Intelligent Monitoring

- Product Overview

The intelligent abnormality detection system senses and handles abnormal system conditions by comprehending processing contents inside the system and the inter-system interface contents in order to prevent run-out and production accidents caused by errors that occur during additional developments or O&M duties that are continuously required for each system process.

< Differentiated Structure Diagram of Abnormality Sensing To-Be System >

< Technical Development Method- Within Abnormality Sensing System Operation >

- Product Features & Merits

1) Event Monitoring
- As making movements in logistics or changes in equipment recipe is possible in event monitoring based on the production procedures, a tracking management should be enabled when such an event occurs
- For the purpose of event processing, a machine learning algorithm is applied based on the cluster analysis where events are classified based on the same characteristics

2) Real-time Log Analysis
- The size of a high-tech factory's log data is not suitable for real-time monitoring by humans, so it is necessary to introduce an automatic analysis system based on M/L, pattern, etc. to deliver key information through tools such as a system pop-up, in case of any trouble
- Log data are collected and classified based on area(FAB), product name, device, and the relationship between process and equipment, in order to analyze the patterns based on the elements suitable for the characteristics of the process
- For the matters to be performed in a specific process, an abnormality sensing model is developed according to the relevant pattern


3) Log Collection/Management
- The system log is divided into lot, system linkage, process flow, etc. for managing collection, optimization, and grouping of data, in order to improve data usability and re-usability and to apply them to various statistical analyses

4) DB Event Monitoring
- Error situations such as RDBMS DB-Lock are immediately notified, and if the error conditions persist, the system makes continuous notifications until they are finally resolved
- DB Event configures an early warning system based on immediate processing through streaming

5) Complex Process Execution Monitoring
- Executions of complex and joint processes requiring inter-system interface cannot cope with any error in most cases as they are performed with individual patches for each system and internal logics are operated, making it difficult to detect changes in other systems. Therefore, adopting a system which shuts off continuous progress of abnormalities by applying the progress execution monitoring based on cumulative learning of the information gained from each process is a differentiated function used for monitoring and controlling any abnormality for the entire process

- System Architecture

< Details of Main Development Linked with Big Data >

Technology & Specifications

1) Data handler
- Formal/Informal data, System log data & DB data collector

2) Data Analyzer
- Data analysis, Storage & refinement analyzer

3) Rule Manager
- Definition of rules needed for analysis, and learning model management tool using M/L method

4) Abnormality Sensing & Early Warning System
- Early warning system UI tool based on the abnormal sensing state and predicted M/L model value