Time-Series Classification for Early Substrate Process Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current substrate processing devices face challenges in efficiently determining abnormalities based on time series data, as existing methods lack effective classification and extraction techniques to identify deviations from normal operation before substantive issues arise.
Innovation Solution
A data processing method and device that acquire time series data, evaluate it using specific criteria, classify the data into predefined levels, and extract data points that deviate significantly from reference values, allowing for early detection of potential abnormalities and proactive control adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If time series data is analyzed to determine abnormalities in substrate processing devices, then abnormality detection capability is improved, but the complexity of data processing and analysis increases
Solution Approach 1:
The patent segments time series data into multiple classification groups based on evaluation values, dividing the complex analysis task into manageable categories. This segmentation allows systematic processing of abnormal patterns while reducing overall computational complexity.
Solution Approach 2:
The patent introduces evaluation values as intermediate parameters that transform raw time series data into classified categories. By changing the parameter representation from continuous time series to discrete classifications, the system simplifies subsequent analysis while maintaining abnormality detection capability.
2Reliability
If classification and extraction techniques are applied to time series data, then early abnormality detection is improved, but the processing time and computational resources increase
Solution Approach 1:
The patent performs classification of time series data into predetermined groups as a preliminary step before detailed abnormality analysis. This preliminary classification organizes data in advance, enabling faster retrieval and analysis of specific abnormal patterns when needed.
Solution Approach 2:
The patent extracts specific time series data that meet predetermined conditions from the overall dataset. By taking out only the relevant abnormal cases for further analysis, the system reduces processing time while maintaining early detection capability.
3Measurement precision
If multiple classifications are created for time series data, then the precision of abnormality identification is improved, but the complexity of classification management increases
Solution Approach 1:
The patent segments time series data into multiple classification groups based on evaluation values, dividing the complex analysis task into manageable categories. This segmentation allows systematic processing of abnormal patterns while reducing overall computational complexity.
Solution Approach 2:
The patent introduces evaluation values as intermediate parameters that transform raw time series data into classified categories. By changing the parameter representation from continuous time series to discrete classifications, the system simplifies subsequent analysis while maintaining abnormality detection capability.
Data Source
AI summary
A data processing method includes: acquiring time series data; acquiring evaluation values; performing classification; and performing extraction. In the acquiring of the time series data, a plurality of time series data acquired by a substrate processing device are acquired. In the acquiring of the evaluation values, the evaluation values of the plurality of time series data are acquired. In the performing of classification, each of the plurality of time series data is classified into one of a plurality of classifications based on the evaluation value. In the performing of extraction, the time series data corresponding to one of the plurality of classifications is extracted as extracted time series data.