Manufacturing Condition Output for Nonlinear Defect Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing quality management methods in manufacturing, particularly for die-cast products, face challenges with non-linear relationships between manufacturing conditions and quality, and are unable to effectively handle a large number of operation factors, leading to increased calculation complexity and inability to prevent defect occurrence.
Innovation Solution
A manufacturing condition output apparatus utilizing machine learning to generate a non-linear model through random forest algorithms, which analyzes manufacturing and inspection data to output change degree information for defect probabilities, enabling the determination of optimal manufacturing conditions to prevent defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If linear regression formula is used to control quality, then manufacturing conditions with linear relationships can be optimized, but it cannot handle non-linear relationships between manufacturing conditions and quality
Solution Approach 1:
The patent transforms the quality control approach by changing from linear regression parameters to tree-based decision parameters. The system uses multiple decision trees that can capture non-linear relationships by partitioning the feature space into regions, where each region has its own prediction model. This allows the system to adapt to complex non-linear relationships between manufacturing conditions and quality outcomes.
2Manufacturing precision
If automatic guidance on operating conditions combinations is provided, then quality improvement can be achieved, but calculation amount enormously increases as number of operation factors increases
Solution Approach 1:
The patent segments the complex multivariable analysis problem into multiple independent decision trees. Each tree focuses on a specific aspect of the quality prediction, breaking down the enormous calculation space into manageable segments. This segmentation allows the system to handle high-dimensional operation factor data efficiently without requiring exhaustive analysis of all possible condition combinations.
Solution Approach 2:
The system dynamically adapts to the data structure by building decision trees that automatically identify the most important variables and their interactions. Rather than statically analyzing all possible combinations, the dynamic tree-building process adapts to the actual relationships in the data, focusing computational resources on the most relevant factors and reducing overall calculation requirements.
3Reliability
If conventional quality management methods are used, then simple manufacturing processes can be managed, but they cannot effectively prevent defect occurrence in complex non-linear manufacturing scenarios
Solution Approach 1:
The patent implements a feedback mechanism where the random forest model continuously learns from manufacturing data and inspection results. The system takes predicted quality outcomes and actual inspection results as feedback, using this information to refine future predictions and recommendations. This feedback loop enables the system to improve defect prevention capability over time while adapting to the specific characteristics of complex manufacturing processes.
Data Source
AI summary
A manufacturing condition output apparatus of an embodiment is a manufacturing condition output apparatus which outputs a manufacturing condition of a product. The manufacturing condition output apparatus outputs change degree information which is information regarding degrees of change of values regarding defect probabilities for a plurality of variables relating to manufacturing of the product from model information of a model generated through machine learning on a basis of manufacturing data of the product and inspection result data of the product, as a manufacturing condition.