Lamb Wave Weld Penetration Depth Prediction Using Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for measuring weld penetration depth in thin structures are non-destructive but lack accuracy and efficiency, as they rely on complex signal processing of broadband Lamb waves, which are challenging to interpret due to their dispersive nature.
Innovation Solution
A non-contact, non-destructive method using a pulsed Nd:YAG laser to generate Lamb waves and an EMAT receiver, with signal processing via complex-valued Morlet wavelets and continuous wavelet transform to calculate transmission coefficients, feeding these into a neural network to predict weld penetration depths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional ultrasonic inspection methods using contact piezoelectric transducers are used, then weld penetration depth can be measured, but the method requires liquid couplants and is not suitable for automated real-time inspection
Solution Approach 1:
The patent replaces contact piezoelectric transducers with laser electromagnetic acoustic transducers (EMATs) that use electromagnetic fields to generate and detect ultrasonic waves without mechanical contact. This substitution eliminates the need for liquid couplants and enables automated real-time inspection while maintaining measurement capability
Solution Approach 2:
The patent introduces laser-generated Lamb waves as an intermediary medium to transmit ultrasonic energy through the weld specimen. These waves propagate through the material structure itself without requiring external couplants, enabling non-contact measurement while preserving the ability to measure weld penetration depth
2Measurement precision
If Time of flight diffraction (ToFD) technique is used to measure weld characteristics, then penetration depth can be evaluated, but the method becomes inaccurate when sample thickness approaches ultrasonic wavelength
Solution Approach 1:
The patent changes the ultrasonic wave parameters by using laser-generated Lamb waves with specific frequency ranges that are appropriate for thin structures. By adjusting the wave parameters to match the thickness of thin materials, the system maintains measurement accuracy where conventional ToFD fails
Solution Approach 2:
The patent employs dynamic signal processing techniques including wavelet transforms and neural networks that can adapt to varying wave propagation characteristics in thin structures. This dynamic approach allows accurate measurement across different thicknesses where static ToFD methods fail
3Productivity
If laser-generated broadband Lamb waves are used for inspection, then non-contact operation enabling real-time inspection is achieved, but signal processing becomes complicated due to dispersive nature
Solution Approach 1:
The patent segments the broadband Lamb wave signal into distinct frequency components using wavelet transforms. This segmentation separates the dispersive wave modes in the frequency domain, making it possible to process and analyze individual components separately, thereby reducing overall signal processing complexity
Solution Approach 2:
The patent introduces neural networks as an intermediary processing layer that automatically extracts features from the complex Lamb wave signals. This intermediary system handles the dispersive nature of the waves through pattern recognition, simplifying the processing pipeline while maintaining real-time inspection capability
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 fast, efficient, and accurate online measurement of weld penetration depths in thin structures, overcoming the limitations of conventional methods by simplifying signal processing and improving prediction accuracy.
Implementation Method 1
a high energy, very short duration pulse from the laser induces a rapid increase in the local temperature of the sample. The heated region expands thermoelastically and then slowly contracts when the laser pulse is momentarily shut off. The rapid expansion and slower contraction creates ultrasounds which propagate through the sample
Implementation Method 2
an electromagnetic acoustic transducer (EMAT), or a laser interferometer, to detect signals from the other side of the weld seam
Data Source
AI summary
A CWT-based method to calculate transmission coefficients of different Lamb waves using individual LEU signals. A neural network was trained to accurately predict WPDs based on the transmission coefficients of selected Lamb waves and the LEU signal energy. The method is capable of inspecting WPDs quickly along welds in thin structures.