Machine Learning Defect Detection in Metal Powder Bed Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Additive manufacturing systems face challenges in detecting and characterizing defects, such as lack of fusion, porosity, and inclusions, which can lead to premature failure of finished parts due to the difficulty in interpreting in-process data and predicting defect types.
Innovation Solution
Employing a machine learning algorithm that uses sensors to collect and process data on thermal emission density and spectral peaks to identify defects, trained with known defective parts to distinguish between different types of defects, allowing for real-time detection and correction during the manufacturing process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sensor data collection methods are used during additive manufacturing, then the manufacturing process can be monitored, but the defect detection precision and ability to distinguish defect types remains insufficient
Solution Approach 1:
The patent transforms raw sensor data into multiple derived parameters including thermal emission density (TED), thermal emission Planck (TEP), and their sigma values that capture different aspects of the melting process. These parameter transformations enable the machine learning algorithm to distinguish between different defect types by analyzing variations in thermal radiation patterns, temperature distributions, and energy absorption characteristics during the additive manufacturing process.
2Reliability
If machine learning algorithms are implemented to detect defects, then defect identification capability is improved, but the system complexity increases
Solution Approach 1:
The patent introduces a machine learning algorithm as an intermediary component that bridges the gap between raw sensor data and defect identification. The algorithm processes thermal emission data, calculates multiple parameters, and applies trained models to classify defects, thereby automating the complex analysis tasks while maintaining system modularity and enabling reliable defect detection without requiring complex manual intervention systems.
3Productivity
If real-time defect detection is implemented, then manufacturing efficiency is improved through informed decision-making, but the data processing requirements and computational load increase
Solution Approach 1:
The patent implements a two-stage approach where a machine learning algorithm is trained offline on labeled defect data to learn defect patterns and characteristics. During real-time manufacturing, the pre-trained algorithm quickly processes thermal emission parameters to detect and classify defects, significantly reducing online computational requirements while maintaining high detection accuracy and enabling rapid decision-making for manufacturing efficiency.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables effective detection and identification of defects, improving manufacturing efficiency by allowing for informed decision-making to repair or stop the build process, reducing waste and enhancing part quality.
Implementation Method 1
a power source arranged to emit a beam of energy at the build plane and fuse the metallic powder
Implementation Method 2
the sintering and/or melting of powdered or granular raw material, layer by layer using one or more high power energy sources such as a laser
Implementation Method 3
one or more sensors arranged to detect electromagnetic energy emitted from the build plane when the beam of energy fuses the metallic powder
Implementation Method 4
The processor converts the received data into one or more parameters that indicate one or more conditions at the build plane
Data Source
AI summary
An additive manufacturing system comprises an apparatus arranged to distribute layer of metallic powder across a build plane and a power source arranged to emit a beam of energy at the build plane and fuse the metallic powder into a portion of a part. The system includes a processor configured to steer the beam of energy across the build plane and receive data generated by one or more sensors that detect electromagnetic energy emitted from the build plane when the beam of energy fuses the metallic powder. The received data is converted into one or more parameters that indicate one or more conditions at the build plane while the beam of energy fuses the metallic powder. The one or more parameters are used as input into a machine learning algorithm to detect one or more defects in the fused metallic powder.