Process Light Prediction for Melt Pool Deviation Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In powder-bed-based additive manufacturing, reliably detecting process deviations in melting processes is challenging due to the lack of a target value for location-dependent emission, making it difficult to recognize anomalies in radiation intensity and wavelength of process light.
Innovation Solution
A method using a machine learning model, trained with historical and synthetic power profiles and associated radiation intensities, to determine the radiation intensity and wavelength of process light, and subsequently controlling the melting process by adapting parameters like beam power and speed to reduce or eliminate process deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If statistical variance-based detection is used for process monitoring, then the system can operate without prior target values, but the detection limit is defined by statistical and systematic variance making it difficult to reliably recognize process deviations
Solution Approach 1:
The system performs preliminary actions by collecting and storing historical process data during normal operation. This data is then used to train machine learning models that predict expected process light characteristics. By preparing these predictive models in advance, the system establishes a baseline for comparison without needing pre-defined statistical target values, thereby improving detection precision while maintaining adaptability.
Solution Approach 2:
The patent introduces machine learning models as an intermediary between raw process light measurements and anomaly detection. These models act as mediators that translate historical data patterns into predictive baseline values. This intermediary layer enables the system to operate without direct statistical target values while achieving superior detection precision compared to traditional variance-based methods.
2Measurement precision
If machine learning models are used to predict radiation intensity and wavelength, then detection accuracy improves, but system complexity increases due to training requirements and computational demands
Solution Approach 1:
The system performs model training as a preliminary action during system setup or offline periods using historical process data. Once trained, the models are deployed for real-time prediction during manufacturing operations. This separation of training (preliminary) and prediction (operational) phases reduces the computational burden during actual production while maintaining high detection accuracy.
Solution Approach 2:
The patent uses synthetic process light data as a copy or approximation of real measurement data for training machine learning models. This synthetic data generation approach allows the system to train models without requiring extensive real historical data, thereby reducing the complexity of data collection and processing while still achieving accurate predictions during actual operation.
3Reliability
If closed-loop control is implemented by adapting beam parameters, then process deviations are reduced, but the control system complexity and response time requirements increase
Solution Approach 1:
The system implements feedback control by continuously comparing actual process light measurements against ML-predicted baseline values. When deviations are detected, the system adjusts beam parameters (power, speed, or path) to correct the process anomaly. This feedback mechanism improves process reliability by automatically correcting deviations while maintaining relatively simple control logic based on established thresholds and correction rules.
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
This approach improves the accuracy and reliability of detecting process deviations and enables closed-loop control of the melting process, enhancing the quality and consistency of the manufacturing process by using a machine learning model to predict and adjust for variations in radiation intensity and wavelength.
Implementation Method 1
process light (emission in visible and near infrared) is being increasingly ascertained for process monitoring
Data Source
AI summary
Various embodiments of the teachings herein include a method for determining a radiation intensity and/or a wavelength of a process light, wherein the melt pool underlying the process light can be generated by irradiating a metal material with an energy beam along a path, wherein the energy beam can be moved in accordance with a power profile along the path. The method may include:providing a power profile for a section of the path as an input variable for a machine learning model; training the model using historical and/or synthetic power profiles and associated historical or synthetic radiation intensities and/or wavelengths of the process light for the metal material; and determining the radiation intensity and/or the wavelength of the process light as an output variable of the model.