Injection Molding Condition Control Using Sensor-Based Defect Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In injection molding, determining the optimal molding conditions to prevent defects is challenging due to various environmental and equipment-related factors, requiring skilled operators and often resulting in defective products.
Innovation Solution
A device using machine learning models to predict molding defects by analyzing molding state data from sensors before the quality element of the molded article is inspected, allowing for adjustments to molding conditions to prevent defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If machine learning is used to automatically adjust molding conditions based on quality element data obtained after inspection, then the need for manual adjustment by operators is eliminated, but it is not possible to predict defects before the inspection process
Solution Approach 1:
The system performs preliminary action by obtaining molding state data during the molding process and using machine learning models to predict quality elements before the actual inspection occurs. This allows defect prediction and molding condition adjustment to happen in advance, enabling preventive rather than reactive quality control.
Solution Approach 2:
The system implements feedback by using the first machine learning model to predict quality elements from molding state data, comparing predicted values with target values, calculating adjustment amounts, and feeding this information back to adjust molding conditions for subsequent production, creating a closed-loop quality control system.
2Measurement precision
If molding conditions are adjusted based on quality element data obtained after inspection, then defective articles are detected, but production of defective articles cannot be prevented in real-time
Solution Approach 1:
The system performs preliminary prediction of quality elements using machine learning models before the inspection process, allowing molding conditions to be adjusted in advance to prevent defect production, thereby maintaining both quality accuracy and production efficiency.
Solution Approach 2:
The system replaces the traditional mechanical inspection process with a machine learning-based prediction system that analyzes molding state data to forecast quality elements, enabling real-time defect prevention without requiring physical inspection of each article.
3Manufacturing precision
If skilled operators manually determine molding condition changes considering various factors, then molding quality can be maintained, but unskilled operators cannot determine how and what to change
Solution Approach 1:
The system enables self-service by automatically analyzing molding state data, predicting quality elements, determining adjustment amounts, and suggesting molding condition changes without requiring operator expertise. The machine learning model performs the complex decision-making that previously required skilled operators.
Solution Approach 2:
The system replaces the human operator's decision-making process with a machine learning-based system that objectively analyzes molding state data and determines optimal condition adjustments, eliminating the dependency on operator skill level while maintaining molding quality.
4Measurement precision
If inspection process is used to obtain quality element data, then defect detection is possible, but defect prediction before inspection cannot be achieved
Solution Approach 1:
The system replaces the inspection process with a machine learning-based prediction system that forecasts quality elements by analyzing molding state data, enabling defect information to be obtained earlier in the production process without requiring physical inspection.
Solution Approach 2:
The system introduces an intermediary machine learning model that bridges the gap between molding state data and quality element outcomes, allowing defect prediction based on intermediate molding parameters before the final inspection point.
Data Source
AI summary
A device for assisting molding condition determination includes a molding state data adjustment amount obtaining unit and a molding condition element adjustment amount obtaining unit. The molding state data adjustment amount obtaining unit obtains, using a first learning model, a molding state data adjustment amount having a value equivalent to a difference between molding state data detected by a sensor and a molding state data target value. The molding condition element adjustment amount obtaining unit obtains, using a second learning model, an adjustment amount for a molding condition element corresponding to the molding state data adjustment amount.