Probabilistic Model for Weld Quality Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for controlling the quality of welds, particularly in non-destructive testing, face challenges in robustly classifying welds as compliant, uncertain, or non-compliant, and are limited to specific materials, lacking a comprehensive and user-friendly approach for real-time evaluation across various materials.
Innovation Solution
A probabilistic statistical model using logistic regression is implemented for weld quality control, which includes a two-phase process: defining a qualification logic model and using it to classify welds based on mechanical resistance measurements, combined with smoothing techniques for anomaly detection in pyrometric and profilometric signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If alert thresholds are placed on weld characteristics for quality determination, then the inspection process is simplified, but the classification robustness deteriorates
Solution Approach 1:
The patent transforms the quality classification problem from using fixed alert thresholds to using a probabilistic statistical model with multiple parameters (mean, standard deviation, skewness, kurtosis) that dynamically assess weld quality. This allows the system to move away from simple threshold comparisons toward a more nuanced statistical evaluation that improves classification robustness while maintaining operational simplicity through automated model application.
2Reliability
If probabilistic statistical models are used for weld quality classification, then classification robustness is improved, but implementation complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-calculating and storing the probabilistic statistical model parameters (mean, standard deviation, skewness, kurtosis) from training data before actual weld inspection. During operation, the system only needs to compute these same statistical parameters for new welds and compare them against the pre-established model, significantly reducing implementation complexity while maintaining classification robustness.
Solution Approach 2:
The patent creates a statistical copy or representation of weld quality characteristics through the probabilistic model parameters. Instead of implementing complex physical measurement systems, the system uses statistical copies (mean, variance, skewness, kurtosis) that capture the essential quality features, simplifying the implementation while preserving classification accuracy.
3Ease of manufacture
If traditional quality control methods are used, then implementation is straightforward, but adaptability to different materials is limited
Solution Approach 1:
The patent implements a universal probabilistic statistical model that can be applied across different material types (metallic and non-metallic). The model uses general statistical parameters (mean, standard deviation, skewness, kurtosis) that are material-agnostic, allowing the same framework to adapt to various materials. The system maintains implementation simplicity by using the same statistical computation approach for all materials while achieving broad adaptability.
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
This approach enhances the robustness of weld quality classification, enabling real-time evaluation and user-friendly implementation across different materials, improving the accuracy and reliability of weld quality assessment.
Implementation Method 1
A second known means of collecting the temperature of the molten metal of the weld bead is the optical pyrometer. The optical pyrometer is a device which is able to capture the thermal radiation emitted by an element by means of a sensor and to provide a signal representative of the temperature of said element.
Implementation Method 2
A first known means comprises an infrared thermal camera, which provides an image representative of the temperature of the observed area
Data Source
Figure 1~3
Figure 4~5
Figure 6~7
AI summary
Method for inspecting the quality of a solder joint, characterized in that it implements a probabilistic statistical model that determines a rating for the quality of the solder joint.