Virtual Product Inspection With Self-Retraining Fault Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In industrial automation, existing data classification models struggle to adapt to new fault categories identified during production, especially in flexible production environments, requiring manual intervention and prolonged manufacturing processes due to the need for physical inspections.
Innovation Solution
A method and device for virtually inspecting products using AI classifier models that dynamically retrain based on new fault category information stored in an extension buffer, allowing for automatic retraining without human intervention and efficient inspection of work in progress items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical inspections are performed on all work in progress pieces, then inspection accuracy is improved, but manufacturing process time is prolonged and complexity increases
Solution Approach 1:
The patent creates a virtual copy of the physical inspection process using a data classification model that replicates inspection functionality. The model receives production information as input and generates inspection results without requiring physical inspection equipment or manual operations, thereby maintaining accuracy while eliminating time loss and process complexity
Solution Approach 2:
The patent replaces the mechanical physical inspection system with an information-based virtual inspection system. Instead of using physical inspection equipment and manual operations, the system uses data classification models that process production information digitally, substituting mechanical processes with information processing to achieve the same inspection objectives without the associated time and complexity costs
2Ease of manufacture
If classifier models are trained with historic data only, then initial model deployment is simple, but the model cannot adapt to new fault categories identified during production
Solution Approach 1:
The patent implements a feedback mechanism where inspection results are fed back into the data classification model. The system compares determined labels with actual inspection data, and when discrepancies or new fault categories are identified, the model automatically retrains using this feedback information. This closed-loop feedback enables the model to continuously adapt to new fault categories while maintaining deployment simplicity through automated processes
Solution Approach 2:
The patent transforms the classifier model from a static system trained once during deployment into a dynamic system that can adapt and evolve during production. The model dynamically adjusts its parameters and knowledge base by incorporating new inspection data and fault categories encountered during operation, allowing it to maintain both simplicity of initial deployment and high adaptability to changing production conditions
3Measurement precision
If new fault categories are identified during production, then product quality monitoring is improved, but the deployed classifier needs manual extension which reduces productivity
Solution Approach 1:
The patent enables the data classification model to perform self-updates when new fault categories are identified. The system automatically detects new fault patterns from inspection data, retrieves relevant production information, and retrains itself without requiring manual intervention. This self-service capability maintains high fault category detection precision while eliminating the productivity loss associated with manual model extension
Data Source
AI summary
A method of virtually inspecting a quality of a product in a production environment includes receiving production information associated with the product, wherein the information is indicative of an operation performed on the product in relation to a first process of the production environment. The method further includes: determining a label for the product using a first classifier model based on the received information associated with the product; comparing the determined label against inspection data associated with an inspection of the product; and storing the production information and the inspection data associated with the product in an extension buffer, based on the comparison of the determined label and the inspection data, for retraining the first classifier model. The above method allows for storing new fault category information that is then dynamically used to retrain the model. Accordingly, this allows for retraining of classifier models without human intervention.