metatron Solution

빅데이터를 위한 가장 효율적인 솔루션을 제공해 드립니다.

Use case
Manufacturing Equipment Yield Improvement - RCA (Root Cause Analytics)
prev next
Manufacturing Equipment Yield Improvement - RCA (Root Cause Analytics)

* RCA (Root Cause Analytics): This is a method that analyzes linkages using source data, trace data, measurement data, yield data, and production history data collected from equipment to determine which equipment or process parameters are major causes of low yield and to provide this information to on-site users.


Due to technology, expense, and performance limitations, Company A’s legacy RCA system could only analyze those parameters of major operations and recipes for which it had been previously developed. For this reason, on-site workers only had access to limited data and analytical results and because this process took a long time, the level of system utilization was very low.
The previous system depended on the experience and judgment of on-site expert engineers who applied and evaluated the various parameter settings and threshold values on production equipment. However, as processes became ever more complex, on-site workers found it difficult to stay familiar with and make judgments on the many parameters required for each machine and process and, in particular, it became nearly impossible to analyze long-term data.

System Components



- Execute parallel/distributed processing of manufacturing domain-specific algorithms to achieve processing speeds orders of magnitude faster than under legacy systems
- Conduct analyses covering entire periods and all parameters
- Provide on-site users with a self-discovery-oriented user-interface environment through intuitive analytical systems and a variety of analysis charts
- Enhance the experience and judgment of on-site engineers by providing fast, precise analyzed results in ways that had previously been unavailable to on-site workers
- Provide real-time data through system reports on major cause parameters not in line with the “golden rule”
* Golden Rule: Normal standard patterns analyzed and generated on processes and equipment based on ML/DNN