Temperature History Prediction Model for Additive Welding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing temperature prediction methods for deposited bodies in additive manufacturing are inefficient, requiring extensive calculations and struggling to achieve high accuracy and speed, especially as the size of the deposited body increases.
Innovation Solution
A learning device and temperature history prediction device that utilize machine learning to generate a prediction model for temperature history during the building of a deposited body. This model predicts temperature distributions for unit elements based on initial and subsequent temperature distributions, allowing for high-speed and accurate predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If finite element method is used for temperature prediction in additive manufacturing, then calculation accuracy is maintained, but calculation time increases significantly
Solution Approach 1:
The patent pre-divides the deposited body into unit elements before actual temperature prediction. This preliminary segmentation creates a reusable framework that can be quickly applied during manufacturing without performing full FEM calculations each time, thus reducing calculation time while maintaining accuracy through the pre-established element structure.
Solution Approach 2:
The patent creates a virtual model (copy) of the deposited body with pre-defined unit elements that mirrors the physical structure. This virtual model allows temperature predictions to be performed on the copy rather than requiring full physical simulations, significantly reducing calculation time while preserving measurement accuracy through the faithful representation of thermal behavior.
2Volume of stationary object
If the size of the deposited body increases, then manufacturing capability is improved, but calculation complexity increases
Solution Approach 1:
The patent divides the deposited body into multiple unit elements, transforming a complex large-scale calculation problem into multiple simpler, manageable sub-problems. Each unit element can be processed independently, reducing overall calculation complexity while enabling the analysis of larger deposited bodies by distributing the computational load across segments.
Solution Approach 2:
The patent applies different treatment to different regions by focusing on local unit elements rather than treating the entire deposited body uniformly. This allows complex thermal behavior to be captured in critical regions while simplifying less critical areas, reducing overall calculation complexity while maintaining accuracy where it matters most for large-scale deposits.
Data Source
AI summary
A learning device includes: a temperature distribution acquisition unit configured to obtain a first temperature distribution representing temperatures of a plurality of unit elements at a specific time of a deposited body and a second temperature distribution representing temperatures of the plurality of unit elements at a time when a predetermined time elapses from the specific time; and a learning unit configured to generate a prediction model by performing machine learning on a relation between the first temperature distribution and the second temperature distribution obtained by the temperature distribution acquisition unit, in association with the predetermined time.