Resistance Welding Spatter Prediction via Classification Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current resistance welding processes struggle to predict and prevent welding spatters effectively, leading to poor welding quality and increased maintenance due to the inability to detect spatter occurrence before it happens, requiring additional cleaning and frequent tool maintenance.
Innovation Solution
A method using classification models trained with data on electrode tip state, welding voltage, and current profiles to predict spatter occurrence, allowing for early adjustment of welding parameters such as voltage, current time, and electrode force to prevent spatter, utilizing algorithms like Logistic Regression, Support Vector Machine, and Random Forest, and processing techniques like Haar wavelet transform.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If resistance welding processes are carried out with high intensity heat and power, then welding speed and productivity are improved, but welding spatters occur leading to poorer welding quality and contamination
Solution Approach 1:
The system performs preliminary actions by training classification models on historical welding data before actual welding operations. The models learn patterns that precede spatter occurrence, enabling prediction and prevention before spatters actually occur during high-intensity welding processes
Solution Approach 2:
The system implements feedback by continuously monitoring welding parameters (current, voltage, time) and using classification models to predict spatter risk. When spatter is predicted, the system can adjust welding parameters in real-time to prevent spatter while maintaining productivity
2Device complexity
If welding spatters are allowed to accumulate on workpiece surfaces, then welding process simplicity is maintained, but additional cleaning steps are required increasing time and cost
Solution Approach 1:
The classification models perform preliminary prediction of spatter occurrence before it happens during welding. By predicting spatter risk in advance, the system can adjust parameters to prevent spatter accumulation, eliminating the need for subsequent cleaning operations
Solution Approach 2:
The system converts the potentially harmful effect of spatter into a beneficial predictive signal. By analyzing patterns that precede spatter, the system uses the threat of spatter to trigger preventive actions that actually eliminate spatter and the need for cleaning
3Productivity
If electrode caps are not maintained frequently, then maintenance time is reduced, but spatter accumulation requires frequent milling or dressing steps
Solution Approach 1:
The system performs preliminary prediction of spatter that would accumulate on electrode caps. By predicting spatter risk before it occurs, the system can adjust welding parameters to prevent spatter accumulation, reducing the frequency of electrode cap maintenance operations
Solution Approach 2:
The classification model continuously monitors welding conditions and automatically adjusts parameters to prevent spatter accumulation on electrode caps. This self-regulating system reduces the need for external maintenance interventions
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate and early prediction of welding spatters, reducing the need for additional measurement tools and allowing for proactive countermeasures to maintain welding quality and reduce maintenance, improving the overall efficiency of the resistance welding process.
Implementation Method 1
A method using classification models trained with data on electrode tip state, welding voltage, and current profiles to predict spatter occurrence
Implementation Method 2
resistance welding process with the features of claim 1... welding current, the current time, and the electrode force... extreme intensity of heat and power, which are applied at or near the weld joint
Implementation Method 3
It is known from DE 102013216966 A1 to monitor welding processes for an occurrence of welding spatter. For this purpose a resistance measurement is carried out
Data Source
Figure 1
Figure 2
AI summary
The present application relates to a method of predicting welding spatters during a resistance welding process, wherein a classification model, eg. Logistic Regression, Support Vector Machine or Random Forest, is used to get a likelihood of an upcoming welding spatter, wherein the classification model can first be trained based on parameters of the resistance welding process. A computing unit and a computer program achieving the method steps are also defined.