ML Welding Condition Prediction for Additive Bead Shape Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The adjustment of welding conditions in additive manufacturing is complex due to numerous condition combinations and the variability of bead shapes, which are influenced by the relationship with surrounding beads, making it difficult to determine appropriate welding conditions.
Innovation Solution
A machine learning device and method that utilize a learned model to determine welding conditions by considering the block pattern formed by weld beads, adjusting the conditions to ensure the difference between derived shape data and design data does not exceed a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual extraction of welding conditions is performed to achieve manufacturing accuracy, then the manufacturing precision is improved, but the complexity and time required increases significantly
Solution Approach 1:
The system performs self-learning by automatically extracting welding conditions and bead shape information from manufacturing data without requiring manual specification. The learning device autonomously processes the complex task of determining appropriate welding conditions by training on historical data, thereby eliminating the need for manual extraction while maintaining high manufacturing precision.
Solution Approach 2:
The patent replaces manual mechanical extraction processes with an automated machine learning system. The learning device uses computational algorithms to process manufacturing data and extract welding conditions, substituting the manual operator's work with an automated information processing system that handles the complexity without human intervention.
2Manufacturing precision
If comprehensive welding condition combinations are considered to improve bead shape control, then the manufacturing precision is improved, but the time and computational resources required increase
Solution Approach 1:
The system performs preliminary learning by training the machine learning model on historical manufacturing data before actual production. This pre-computational phase allows the system to establish relationships between welding conditions and bead shapes in advance, so that during actual manufacturing, the system can quickly determine appropriate conditions without performing exhaustive analysis in real-time.
Solution Approach 2:
The learning device autonomously analyzes comprehensive welding condition combinations and bead shape outcomes from historical data, performing the time-consuming analysis work automatically. Once trained, the system can rapidly determine suitable welding conditions for new cases without repeating the exhaustive analysis, significantly reducing the time required for condition determination.
3Manufacturing precision
If bead arrangement patterns are included in the database to account for surrounding bead influence, then the manufacturing precision is improved, but the database complexity and creation effort increase
Solution Approach 1:
The patent merges multiple data elements into a unified learning dataset. Instead of creating separate databases for welding conditions and bead arrangement patterns, the system combines both types of information into a single comprehensive dataset that the machine learning model processes together. This integration reduces the complexity of managing multiple separate databases while capturing the interrelationships between welding conditions and surrounding bead configurations.
Solution Approach 2:
The learning device automatically extracts and processes bead arrangement pattern information from manufacturing data without requiring manual database creation. The system autonomously identifies relevant spatial relationships and incorporates them into the learning model, eliminating the manual effort of creating and maintaining complex databases while improving bead shape consistency through comprehensive pattern recognition.
Data Source
AI summary
A machine learning device that performs machine learning of a welding condition for manufacturing an additively-manufactured object by welding a filler metal and depositing weld beads, the machine learning device includes: at least one hardware processor configured to perform a learning process for generating a learned model using a welding condition of a weld bead and a block pattern formed by the weld bead as input data and shape data of the weld bead as output data.