Machine Learning Control of Mass and Heat Flux in Additive Manufacturing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing additive manufacturing systems face challenges in efficiently controlling multiple non-linearly related operating parameters, leading to unpredictable results and prolonged reaction times to process changes due to the adage of adjusting 'one knob at a time'.
Innovation Solution
Implementing a machine learning model in an additive manufacturing system to simultaneously adjust two or more non-linearly related operating parameters based on mass and heat flux data from sensors, enabling parallel control of energy and powder delivery devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional one-by-one parameter adjustment method is used, then system complexity is reduced and ease of operation is improved, but manufacturing precision deteriorates and productivity decreases due to prolonged reaction times
Solution Approach 1:
The patent replaces traditional mechanical control systems with machine learning-based intelligent control. The machine learning model analyzes sensor data (mass flux, heat flux, melt pool characteristics) and automatically adjusts multiple operating parameters (laser power, powder feed rate, build speed) simultaneously, substituting manual one-by-one parameter tuning with automated intelligent control that achieves superior precision without increasing operational complexity
Solution Approach 2:
The system dynamically changes multiple operating parameters in parallel based on real-time sensor feedback and machine learning predictions. Instead of adjusting one parameter at a time, the system simultaneously optimizes laser power, powder feed rate, and build speed by analyzing their non-linear interactions through machine learning models, enabling rapid adaptation to process variations while maintaining precision
2Productivity
If multiple operating parameters are adjusted simultaneously using machine learning, then productivity is improved and manufacturing precision is enhanced, but device complexity increases
Solution Approach 1:
The machine learning control system serves multiple functions simultaneously: it predicts optimal parameter settings, analyzes sensor data from multiple sources (mass flux sensors, heat flux sensors, melt pool imaging), and coordinates adjustment of multiple operating parameters. This multi-functional approach consolidates what would otherwise require separate control systems into a single intelligent platform, managing complexity while enhancing productivity
Solution Approach 2:
The machine learning model acts as an intermediary between sensor data and actuator control. Rather than direct complex interactions between multiple sensors and multiple actuators, the machine learning model processes sensor inputs (mass flux, heat flux, melt pool characteristics) and generates coordinated control outputs, simplifying the overall system architecture while enabling sophisticated multi-parameter optimization
3Loss of time
If traditional control methods are used, then device complexity is minimized, but loss of time increases due to prolonged reaction times to process changes
Solution Approach 1:
The machine learning model is trained in advance on historical process data to learn optimal parameter relationships and predictions. During operation, the pre-trained model can rapidly predict the effects of parameter changes and suggest optimal adjustments before actual process deviations occur, enabling proactive rather than reactive control and significantly reducing response time to process changes
Solution Approach 2:
The system implements real-time feedback control by continuously monitoring sensor data (mass flux, heat flux, melt pool characteristics) and using the machine learning model to adjust operating parameters dynamically. This closed-loop feedback mechanism enables rapid detection and correction of process deviations, reducing loss of time compared to traditional open-loop or manual control methods
Data Source
AI summary
An additive manufacturing system may include an energy delivery device configured to deliver energy to a build surface of a component to form a melt pool in the build surface of the component; a powder delivery device configured to direct a powder stream toward the melt pool; a plurality of mass sensors, each mass sensor associated with a portion of the additive manufacturing system; a plurality of heat sensors; and one or more computing devices. The computing device(s) are configured to receive data from the plurality of mass sensors; determine an overall mass flux based on the data from the mass sensors; control the powder delivery device based on the overall mass flux; receive data from the plurality of heat sensors; determine an overall heat flux based on the data from the heat sensors; and control the energy delivery device based on the overall heat flux.