Machine Learning Device for Injection Mold Wear Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The timing for replacing molds in injection molding machines is difficult to determine due to varying molding conditions, leading to reduced dimensional and surface accuracy of molded articles, requiring experienced operators to assess mold wear.
Innovation Solution
A machine learning device and prediction system that acquires molding conditions and mold state information to generate a learned model, predicting mold wear and determining optimal replacement timing based on supervised learning and threshold values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If mold replacement timing is determined based on shot count, then replacement can be scheduled systematically, but the actual mold wear state cannot be accurately predicted due to varying molding conditions
Solution Approach 1:
The system changes from using a single parameter (shot count) to multiple parameters (resin temperature, injection pressure, injection speed, mold temperature, additive types and blending ratios) to accurately predict mold wear state, resolving the contradiction between prediction accuracy and system complexity by systematically incorporating relevant molding parameters
Solution Approach 2:
The system replaces operator experience-based mechanical assessment with an automated machine learning prediction system that processes molding condition data and outputs predicted mold wear states, eliminating the need for experienced operators while improving prediction reliability
2Measurement precision
If experienced operators assess mold wear, then accurate judgment can be made, but operator expertise is required and the process is not scalable
Solution Approach 1:
The system enables self-service by allowing the prediction system to automatically assess mold wear state based on molding condition data without requiring operator intervention or expertise, making the assessment process accessible to any operator while maintaining high accuracy
Solution Approach 2:
The system substitutes operator-based manual assessment with an automated machine learning model that processes molding condition parameters and predicts mold wear state, replacing the need for human expertise with an automated computational system
3Productivity
If mold is used beyond wear threshold, then productivity is maintained, but dimensional accuracy and surface accuracy of molded articles deteriorate
Solution Approach 1:
The system performs preliminary action by predicting mold wear state before actual wear occurs, allowing proactive scheduling of mold replacement at optimal timing that prevents quality deterioration while maximizing mold utilization and maintaining productivity
Solution Approach 2:
The system implements feedback by continuously monitoring molding condition parameters, predicting mold wear state, and using this information to determine optimal replacement timing, creating a closed-loop system that balances productivity and quality through data-driven decision making
Data Source
AI summary
Predicting a state of a mold after molding upon injection molding. A machine learning device includes: an input data acquiring unit that acquires input data including any molding condition including at least a type of resin, a type of additive, a blending ratio of the additive, and a temperature of the resin in molding any article molded by any injection molding machine, and state information indicating a wear amount of a mold before molding at the molding conditions; a label acquiring unit that acquires label data indicating state information of the mold after molding at the molding conditions included in the input data; and a learning unit that executes supervised learning using the input data acquired by the input data acquiring unit and the label data acquired by the label acquiring unit, and generates a learned model.