Virtual-Real Data ML for Additive Manufacturing Sensor Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manufacturing industry faces challenges in predicting sensor values accurately for complex additive manufacturing processes, such as 3D printing, due to numerous print parameters and evolving physics at different scales, which can be costly and time-consuming using physical simulations.
Innovation Solution
A method that combines virtual and real data to train a machine-learning software model, allowing it to predict sensor behavior without modeling exact sensor workings or extensive calibration, using computer-based simulations to generate virtual data and real-world sensor readings for supervised training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical simulations are used to predict sensor values for additive manufacturing processes, then prediction accuracy can be improved, but computational cost and time consumption increase significantly
Solution Approach 1:
The patent pre-generates virtual sensor data through simulations during the training phase before actual manufacturing operations. This preliminary action creates a ready-to-use training dataset that eliminates the need for real-time physical simulations during production, thus reducing time consumption while maintaining prediction accuracy.
Solution Approach 2:
The patent creates virtual copies of sensor data through simulations rather than relying on actual physical sensor measurements during training. These synthetic sensor readings replicate real sensor behavior and are used to train machine learning models, avoiding the need for extensive physical experimentation and real-time simulation during manufacturing.
2Measurement precision
If physical simulations are used to model complex additive manufacturing processes, then prediction accuracy can be improved, but computational resources and cost increase
Solution Approach 1:
The patent performs computationally intensive simulation work beforehand to generate virtual training data. By completing simulations during the offline training phase rather than during production, the system reduces real-time computational costs while maintaining high prediction accuracy for quality metrics.
Solution Approach 2:
The patent uses cost-effective machine learning models trained on virtual data as substitutes for expensive, resource-intensive physical simulations during actual manufacturing. Once trained, the ML models provide accurate predictions at minimal computational cost compared to running full physical simulations.
3Measurement precision
If extensive calibration efforts are made to model sensor workings accurately, then measurement precision can be improved, but time and complexity of the process increase
Solution Approach 1:
The patent uses virtual sensor data that copies real sensor behavior patterns without requiring physical calibration of actual sensors. The simulated sensor readings capture the essential characteristics and noise profiles of real sensors, enabling accurate model training without complex calibration procedures.
Solution Approach 2:
The machine learning model automatically learns sensor behavior patterns from the virtual training data without requiring manual calibration interventions. The system self-adjusts to capture sensor characteristics through the training process, eliminating the need for extensive manual calibration efforts.
Data Source
AI summary
A method includes simulating a process, with computer-based software, to produce virtual data about the process; identifying process parameters for a real-world version of the process; providing a real-world sensor to sense a parameter associated with the real-world version of the process; receiving sensor readings from the real-world sensor while the real-world version is being performed; and training a machine-learning software model to predict a behavior of the real-world sensor based on the virtual data about the process, the process parameters, and the sensor readings.