Welding Abnormality Estimation Using Multimodal Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current welding quality inspection methods are limited in accuracy due to the inability to effectively detect internal defects and the interference of arc light, and the disturbance of welding sound and voltage waveform data does not reliably indicate defects.
Innovation Solution
A welding system employing multiple sensors and a machine-learning based estimation unit to analyze data from camera images, welding sound, and voltage/current waveforms to predict welding abnormalities, using a trained model to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only camera images are used for welding inspection, then the inspection system is simple, but the accuracy is limited because internal defects cannot be detected and arc light interferes with the weld zone
Solution Approach 1:
The patent combines multiple sensor types (camera, microphone, current sensor, voltage sensor) into an integrated inspection system that simultaneously captures visual, acoustic, and electrical data during welding, enabling comprehensive defect detection including internal defects that single sensors cannot detect
Solution Approach 2:
The inspection system is segmented into multiple independent sensor components, each responsible for detecting specific aspects of welding quality (visual appearance, sound anomalies, current characteristics, voltage characteristics), allowing the system to overcome the limitations of any single sensor type
2Measurement precision
If welding sound and voltage waveform monitoring are used, then additional defect detection capability is provided, but false alarms increase because disturbances are not necessarily associated with actual defects
Solution Approach 1:
The system merges data from multiple sensor types (camera, microphone, current sensor, voltage sensor) and uses machine learning to analyze the combined information, enabling the system to distinguish between actual defects and normal disturbances by recognizing patterns across multiple data sources simultaneously
Solution Approach 2:
The system uses machine learning models trained on labeled data to provide intelligent feedback on welding quality, continuously improving its ability to distinguish between normal variations and actual defects through pattern recognition and predictive analysis
3Measurement precision
If multiple sensors and machine learning are used, then prediction accuracy is improved, but the complexity of data processing and model training increases
Solution Approach 1:
The system performs preliminary actions by pre-training machine learning models with large datasets before actual welding inspection, enabling the model to learn normal patterns and defect patterns in advance, so that during actual inspection the processing is faster and more accurate
Solution Approach 2:
The machine learning system performs self-service by automatically learning from data and improving its own prediction capabilities through continuous training, reducing the need for manual programming and complex configuration of detection algorithms
Data Source
AI summary
This welding system comprises: a welding device, various different types of a plurality of sensors which detect an event according to welding performed by a welding device; and an estimation unit which uses a trained model that is pre-generated by machine-learning by taking, as input data, a plurality of pieces of data for learning obtained by detecting events according to welding by means of the same types of sensors as the plurality of sensors, and, as training data, labels representing whether the welding is normal or abnormal, thereby estimating an abnormality of the welding performed by the welding device from a plurality of pieces of detection data generated by the plurality of sensors.