Binary Classification of Vibration Weld Quality Using Sensor Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional process control methods for repeatable processes like vibration welding fail to consistently produce high-quality welds due to external factors such as material quality, leading to labor-intensive visual inspections and potential premature failures.
Innovation Solution
A system and method using a host machine and learning machine with sensors to predict the binary quality status of welds by extracting features from sensory signals and mapping them to a dimensional space, allowing for real-time classification into passing or failing states, thereby reducing the need for manual inspections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional process control methods monitor fixed variables against calibrated thresholds, then welding parameters can be maintained within limits, but weld quality consistency deteriorates due to external factors like material quality
Solution Approach 1:
The patent transforms fixed threshold monitoring into dynamic parameter monitoring by extracting multiple time-varying features (RMS amplitude, peak amplitude, frequency spectrum characteristics, time-domain waveforms) from sensory signals. These parameters are continuously updated and compared against learned patterns from training data, allowing the system to adapt to material quality variations and external factors while maintaining weld quality consistency.
Solution Approach 2:
The patent replaces manual visual inspection and physical picking tests with an automated sensory signal analysis system. Sensors capture vibration, acoustic, or force signals during welding, and a learning machine automatically processes these signals to predict weld quality, eliminating the need for labor-intensive manual inspection while improving reliability.
2Reliability
If visual inspection and manual picking are performed on each weld, then weld integrity can be verified, but production time and labor costs increase significantly
Solution Approach 1:
The system enables self-inspection by integrating sensory signals captured during the welding process itself into the quality assessment. The same sensors that monitor welding parameters also provide data for weld quality prediction, eliminating the need for separate inspection operations. The learning machine automatically analyzes the signals and provides real-time quality predictions without requiring manual intervention.
Solution Approach 2:
The patent implements continuous real-time quality monitoring during the welding process by continuously analyzing sensory signals. Instead of discrete inspection points, the system processes signals throughout the entire welding operation, providing ongoing quality assessment that maintains productivity while ensuring reliability through continuous verification.
3Measurement precision
If multiple features are extracted from sensory signals for classification, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent segments the complex feature extraction process into distinct, manageable components: time-domain feature extraction (RMS amplitude, peak amplitude), frequency-domain feature extraction (spectrum characteristics), and temporal pattern recognition. Each segment processes specific aspects of the sensory signals independently, then combines results for final classification, reducing overall computational complexity while maintaining high prediction accuracy.
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 significantly reduces false passes and manual inspections, improving production quality and efficiency by accurately classifying welds into good or bad states, minimizing Type II errors, and allowing for real-time monitoring of weld quality in vibration welding processes.
Implementation Method 1
vibration welding, which involves the controlled application of high frequency vibration energy
Implementation Method 2
Surface friction generates heat at a weld interface
Implementation Method 3
controlled application of high frequency vibration energy to a clamped work piece
Data Source
AI summary
A system includes host and learning machines. Each machine has a processor in electrical communication with at least one sensor. Instructions for predicting a binary quality status of an item of interest during a repeatable process are recorded in memory. The binary quality status includes passing and failing binary classes. The learning machine receives signals from the at least one sensor and identifies candidate features. Features are extracted from the candidate features, each more predictive of the binary quality status. The extracted features are mapped to a dimensional space having a number of dimensions proportional to the number of extracted features. The dimensional space includes most of the passing class and excludes at least 90 percent of the failing class. Received signals are compared to the boundaries of the recorded dimensional space to predict, in real time, the binary quality status of a subsequent item of interest.