Time-Series Causal Factor Analysis for Actionable Process Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing time-series data in manufacturing processes can identify factors linked to defects but fail to provide actionable countermeasures, especially when the influencing factors cannot be directly controlled, and lack the ability to estimate the effectiveness of control operations in real-time.
Innovation Solution
A factor analysis support system that calculates and restores the contribution degree of time-series variables to an objective variable, allowing for the identification of root factors causing quality deterioration and enabling the presentation of implementable countermeasures by analyzing the transition of contribution degrees through a time-series causal model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI and XAI technology are used to extract factors contributing to defect degree, then the ability to identify linked factors is improved, but the ability to determine actionable countermeasures deteriorates
Solution Approach 1:
The patent introduces a time-series causal model as an intermediary between the AI factor identification and countermeasure determination. This model includes a contribution degree calculation unit that quantifies the relationship between factors and defect degree over time, and a contribution degree restoration unit that simulates the effect of control operations. This intermediary structure enables the translation of identified factors into actionable countermeasures by showing the causal relationships and expected outcomes of control actions.
2Loss of information
If the system identifies factors linked to defects in time-series data, then the analysis capability is improved, but the ability to determine control operations deteriorates
Solution Approach 1:
The patent implements a feedback mechanism through the contribution degree restoration unit. This unit calculates the restored contribution degree when a control operation is applied, providing feedback information about the expected effect of the control operation. The system compares the restored contribution degree with the original contribution degree to determine the effectiveness of the control operation, enabling operators to make informed decisions about which control operations to implement.
3Measurement precision
If the system determines which data is linked to defects, then the diagnostic capability is improved, but the ability to implement countermeasures at the right timing deteriorates
Solution Approach 1:
The patent applies preliminary action by calculating and storing the time-series causal model and contribution degrees in advance, before actual defects occur. The system pre-establishes the relationships between factors and defect degree, and pre-calculates the expected effects of various control operations. When a defect is detected, the system can immediately retrieve and present appropriate countermeasures with their expected effectiveness, eliminating the need for real-time analysis and enabling rapid response.
Data Source
AI summary
By using a time-series causal model stored in a time-series causal model storage unit and time-series data of an analysis target, a contribution degree restoration rate is calculated by evaluating how much a contribution degree of each time-series variable at each time is required to be restored to the contribution degree of another time-series variable at another time. Further, the contribution degree of each time-series variable at each time is restored, based on the calculated contribution degree restoration rate, to the contribution degree of the other time-series variable at the other time, and the transition of the contribution degree by a root factor with respect to an objective variable is calculated. Additionally, the contribution degree by the root factor with respect to the calculated objective variable is output.