Manufacturing Quality Inspection Using Retrospective Data Labeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for recording and analyzing operational data from manufacturing devices are limited, as they only allow for continuous, uninterrupted data recording, making it difficult to prepare data for machine learning models and retrospectively assign quality or status information, and do not enable adjustments to trigger values or status data.
Innovation Solution
A computer-implemented method that obtains operational and status data from manufacturing devices, labels subsets of this data, and uses it to train machine learning models for quality inspection, allowing for the selection and visualization of individual process sections and enabling the output of quality indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If operational data is recorded continuously without interruption, then data completeness is improved, but data usability for machine learning models deteriorates due to inability to retrospectively label and segment data
Solution Approach 1:
The system introduces a preliminary action by storing trigger values and status data in advance during the manufacturing process. These trigger values are stored before they are needed for labeling, allowing the system to retrospectively assign labels to operational data segments without interrupting continuous data collection. This resolves the contradiction by enabling data segmentation and labeling capability while maintaining continuous recording.
Solution Approach 2:
The patent introduces an intermediary mechanism - a database that stores trigger values and status data separately from the operational data stream. This intermediary storage layer allows the system to bridge between continuous data collection and the need for segmented, labeled training data. The trigger values act as mediators that connect operational data to quality outcomes, enabling retrospective labeling while preserving data completeness.
2Reliability
If trigger values and status data are fixed during manufacturing, then system reliability is improved, but system adaptability deteriorates due to inability to make adjustments
Solution Approach 1:
The system implements dynamics by allowing trigger values and status data to be modified after the manufacturing process completes. The database structure enables these parameters to transition from fixed during manufacturing to adjustable afterwards. This resolves the contradiction by providing reliability during the actual manufacturing process while enabling adaptability for optimization and retraining of machine learning models later.
3Ease of operation
If operational data is stored on individual manufacturing devices, then data accessibility is improved, but system complexity deteriorates due to manual work required for each device
Solution Approach 1:
The patent implements universality by creating a centralized database system that serves multiple manufacturing devices through a single unified interface. The same database structure and trigger value storage mechanism work across all devices, eliminating the need for device-specific implementations. This resolves the contradiction by providing easy data access through a universal system while reducing overall complexity compared to individual device configurations.
4Measurement precision
If manual work is used to search and visualize operational data, then measurement precision is improved, but productivity deteriorates due to tedious manual processes
Solution Approach 1:
The system implements self-service by automatically performing data segmentation, labeling, and preparation for machine learning models using the stored trigger values and status data. Instead of requiring manual searching and visualization, the system autonomously processes operational data through the trained machine learning model to generate quality indicators. This resolves the contradiction by maintaining precise data analysis through automated processes while dramatically improving productivity.
Data Source
AI summary
A computer-implemented method for quality inspection of a component of a manufacturing device includes obtaining operational data relating to operation of the manufacturing device. The operational data includes a time series of one or more physical properties of the manufacturing device. Status data relating to a component of the manufacturing device is obtained. The status data includes events relating to and/or characteristic properties relevant for utilization of the component within the manufacturing device. The computer-implemented method includes labelling one or more subsets of the operational data by associating one or more of the events and/or characteristic properties to the one or more subsets and providing the one or more subsets as labelled training data for training a machine learning model. The machine learning model serves for outputting a quality indicator based on the labelled training data input. The trained machine learning model is provided for quality inspection.