Probabilistic Quality Prediction for Injection Molding Inspection Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current quality prediction systems for molding products in injection molding machines rely on point estimation, which can lead to confusion in determining the necessity for inspection and may not effectively reduce the inspection burden, as they lack a comprehensive index for determining the quality and variability of the products.
Innovation Solution
A quality prediction device and method that employs a probabilistic prediction model, such as Bayesian linear regression, to generate quality indexes like expected values, variances, and probability density distributions based on training data related to molding machine conditions and product quality, aiding in determining the necessity for inspection by using a confusion matrix to adjust determination parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If point estimation is used for quality prediction, then the prediction process is simple, but the accuracy and reliability of quality assessment deteriorates
Solution Approach 1:
The patent transforms the quality prediction approach from point estimation to probabilistic prediction by changing the output parameters from single values to distributions characterized by expected values and variances. This allows the system to provide not only predicted quality values but also confidence intervals, thereby improving assessment reliability without significantly increasing system complexity
Solution Approach 2:
The patent adds a new dimension to quality prediction by introducing probability distributions instead of single point estimates. By calculating expected values and variances, the system provides a two-dimensional view (value + uncertainty) that enhances decision-making for inspection necessity determination
2Productivity
If comprehensive quality indexes are introduced to reduce inspection burden, then inspection efficiency improves, but the complexity of quality determination increases
Solution Approach 1:
The patent applies partial action by introducing a threshold-based determination mechanism. Instead of requiring comprehensive analysis of all quality parameters, the system uses the probabilistic prediction model to identify cases where inspection is clearly unnecessary (high confidence in quality) or clearly required (low confidence or predicted defects), thereby reducing inspection burden while maintaining reliability
Solution Approach 2:
The probabilistic prediction model acts as an intermediary between manufacturing data and inspection decisions. By introducing expected values and variances as intermediate parameters, the system bridges the gap between raw production data and quality determination, simplifying the decision process while improving accuracy
3Reliability
If probabilistic prediction model is used to calculate quality indexes, then the reliability of quality assessment improves, but the calculation complexity increases
Solution Approach 1:
The patent implements feedback mechanisms by using the calculated variances and confidence intervals to continuously improve the prediction model. The system learns from past predictions and actual inspection results, adjusting the probabilistic model parameters to enhance reliability over time while managing calculation complexity through iterative optimization
Data Source
AI summary
A quality prediction device includes a probabilistic prediction model generation unit that generates a probabilistic prediction model on the basis of a plurality of training data in which a log relating to a molding operating condition or to a state of a molding machine and a quality value of a molding product corresponding to the log are associated with each other, and a quality prediction unit that calculates a quality index of the molding product from the log using the probabilistic prediction model.