Composite Layup Data Analytics for Inconsistency Root Cause Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional composite manufacturing processes are prone to errors and inconsistencies, leading to reduced yield, increased scrap, and performance/weight penalties due to structural knockdowns, especially when dealing with large composite components.
Innovation Solution
The implementation of data analytics using machine learning to optimize composite manufacturing processes by analyzing historical observations and applying them to artificial neural networks to classify and address localized inconsistencies in composite structures, thereby modifying process parameters to improve quality and consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional composite manufacturing processes are used, then manufacturing simplicity is maintained, but manufacturing precision and quality consistency deteriorate due to errors and inconsistencies
Solution Approach 1:
The system performs preliminary actions by collecting historical process data and training machine learning models before actual manufacturing. The ML models are pre-trained on historical observations of process parameters and defects, enabling them to predict and prevent quality issues before they occur during production. This preliminary preparation allows the system to maintain high manufacturing precision without adding operational complexity during the manufacturing process itself.
Solution Approach 2:
The system implements continuous feedback by monitoring process parameters in real-time and comparing them against the trained ML models. When deviations are detected, the system provides feedback to adjust process parameters, ensuring quality consistency. The feedback loop includes collecting new data from each manufacturing cycle and continuously improving the ML models, creating a self-optimizing system that maintains precision without requiring complex manual intervention.
2Productivity
If conventional manufacturing processes are used, then process simplicity is maintained, but productivity decreases due to reduced yield and increased scrap
Solution Approach 1:
The system enables self-service by allowing the manufacturing process to automatically monitor, analyze, and adjust itself using ML algorithms. The system independently identifies quality issues, predicts defects, and optimizes process parameters without requiring extensive external intervention. This self-service capability increases productivity by reducing scrap and rework while the complexity is encapsulated within the automated system rather than requiring complex human processes.
Solution Approach 2:
The system dynamically changes process parameters based on real-time data and ML predictions to optimize manufacturing outcomes. By automatically adjusting parameters such as temperature, pressure, and curing cycles based on historical patterns and current conditions, the system improves yield and reduces scrap. The parameter changes are managed algorithmically, maintaining simplicity in operation while achieving high productivity through data-driven optimization.
3Reliability
If conventional manufacturing processes are used, then operational simplicity is maintained, but reliability decreases due to performance penalties from structural knockdowns
Solution Approach 1:
The system replaces mechanical quality control methods with data analytics and machine learning algorithms. Instead of relying on physical inspections and manual quality checks, the system uses digital models and statistical analysis to predict and ensure structural integrity. This substitution improves reliability by providing more accurate and consistent quality assessment while the analytics complexity is managed through automated computational processes rather than manual procedures.
Solution Approach 2:
The system performs preliminary structural validation by analyzing historical data and predicting potential structural issues before manufacturing. The ML models are pre-trained on data including structural performance outcomes, enabling them to identify risk factors and optimize process parameters to ensure structural integrity. This preliminary action improves reliability by preventing structural knockdowns before they occur, while the analytics complexity is contained within the predictive modeling framework.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
A method (800) of manufacturing a composite structure (102) includes accessing (802) design data (410) for the composite structure (102) that is manufactured according to a process including forming a layup of plies of fibers using a machine tool. The method includes applying (804) the design data to an ANN classifier to classify a localized inconsistency of a type of inconsistency on the composite structure (102), the localized inconsistency spatially referenced to a location on the composite structure (102). The method includes performing (806) a root cause analysis to identify one or more of process parameters as a potential cause of the type of inconsistency, and modifying (808) one or more of the geometric model, the layup design, or values of the one or more of the process parameters to address the potential cause.