Standardized AI Molding Condition Correction Across Multiple Molds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing molding condition setting for injection and extrusion molding requires significant trial and error based on operator experience, leading to inefficient production and the intentional production of defective products, especially when applying AI models like reinforcement learning, which necessitates large training datasets and is specific to each mold or product.
Innovation Solution
A standard model learner that standardizes and normalizes measurement and inspection data to determine molding condition modifications, applicable across various molds, using a molding condition modification device, method, and computer program, which includes a learner trained on the relationship between measurement data, inspection results, and modification amounts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a learner (AI model) is built for modifying molding conditions using reinforcement learning, then molding condition modification can be automated, but thousands of training shots are required and one learner is needed for each mold or product
Solution Approach 1:
The patent creates a universal learner that can handle multiple molds and products through standardization. The learner is trained with standardized molding condition data that can be applied across different molds and products, eliminating the need to create separate learners for each mold or product while maintaining automated modification capabilities
2Extent of automation
If reinforcement learning is used for molding condition modification, then automated optimization is achieved, but large amounts of training data including defective product data are required
Solution Approach 1:
The patent performs preliminary standardization of molding conditions and creation of standardized training datasets before actual training. By pre-processing and standardizing the data structure, the system reduces the total volume of training data needed while maintaining the effectiveness of the reinforcement learning process
Solution Approach 2:
The patent transforms molding conditions into standardized parameters that can be universally applied. This parameter standardization allows the system to generalize from fewer training examples, reducing the quantity of training data required while maintaining optimization effectiveness
3Reliability
If defective product data is used for training the learner, then comprehensive training coverage is achieved, but intentional production of defective products adversely affects factory production plans
Solution Approach 1:
The patent creates standardized representations and models of molding conditions and defects without physically producing defective products. By using virtual modeling and standardized data representations, the system achieves comprehensive training coverage while avoiding the need to intentionally manufacture defective products that would disrupt production schedules
4Ease of operation
If operator experience is used for molding condition setting, then flexibility in handling various conditions is maintained, but significant trial and error is required leading to inefficient production
Solution Approach 1:
The patent implements a feedback mechanism where the learner continuously receives information about molding results and automatically adjusts conditions. This closed-loop system maintains the flexibility to handle various conditions while eliminating the trial-and-error process, significantly improving production efficiency
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This molding condition correcting device for correcting a molding condition of a molding machine comprises: an acquiring unit for acquiring measurement data obtained by measuring a state of the molding machine, and inspection result data obtained by inspecting a condition of a molded product molded by the molding machine; and a learning device which learns a relationship between the measurement data and the inspection result data, and a correction amount of the molding condition, and which determines the correction amount on the basis of the acquired measurement data and inspection result data. At least one of the measurement data, the inspection result data and the correction amount handled by the learning device is standardized or normalized.