Application Cases

Experiment Name: Data-Driven Research on Magnetic Barkhausen Noise Characterization Method for Surface Stress in Structural Steel

Research Direction: Material Testing

Experiment Objective:
Magnetic Barkhausen Noise (MBN) technology can be used for the quantitative evaluation of surface stress in ferromagnetic materials. Current MBN-based stress assessment techniques face challenges such as difficulty in selecting characteristic parameters, complexity of quantitative prediction models, and low fitting accuracy to calibration datasets. This paper proposes a data-driven nonlinear mapping algorithm to fit the relationship between MBN noise and stress. Time-frequency features based on wavelet packet transform coefficients are studied as replacements for statistical features, reducing the computational load of sample data. Wavelet packet transform coefficients of MBN noise in the time-frequency domain are used as feature vectors. A data dimensionality reduction algorithm based on singular value decomposition is employed to reduce the dimensionality of the feature vectors. The dimensionality-reduced feature vectors are then input into a BP neural network for model training to establish a prediction model. The results indicate that using the singular value decomposition-based data dimensionality reduction algorithm can reduce model complexity. Training the BP neural network with the dimensionality-reduced wavelet packet transform coefficient feature vectors enables high-precision prediction of surface stress in ferromagnetic materials. The characterization method established in this paper effectively addresses the problem of stress distribution imaging in ferromagnetic components and has broad application prospects in damage warning  areas such as preventing stress corrosion and improving fatigue strength.

Testing Equipment: The system consists of hardware and software components. The hardware part includes an excitation module, an MBN sensor, a signal conditioning module, and an AD acquisition module. The software part comprises a host computer designed with LabVIEW, featuring data acquisition, analysis, and processing functions. The excitation module is composed of an AFG3102C arbitrary waveform function signal generator and an ATA-4014 high-voltage power amplifier.

Figure 1: Schematic Diagram of the MBN Detection System

Experimental Procedure:
The MBN sensor comprises three parts: a U-shaped magnetic yoke, an excitation coil, and a detection coil. The U-shaped magnetic yoke is made of manganese-zinc ferrite with high magnetic permeability and low electrical conductivity. The magnetic pole faces are square planes, with a center-to-center distance of 19 mm between the two magnetic poles. Both the excitation coil and the detection coil use enameled wire, wound on the U-shaped magnetic yoke and a magnetic rod, respectively. The detection coil is positioned at the center directly below the U-shaped magnetic yoke. When a sinusoidal signal passes through the excitation coil, an approximately uniform alternating magnetic field is generated between the two poles of the U-shaped yoke, forming a magnetic circuit within the U-shaped yoke and the test block, thereby generating weak MBN signals at the millivolt level. The MBN signal picked up by the detection coil is transmitted via cable to the signal conditioning module for filtering, amplification, and other processing, making it suitable for the AD acquisition module to send to the host computer for signal display, data storage, and other functions. The figure below shows a comparison of MBN signal waveforms before and after filtering.

Figure 2: Comparison of MBN Signal Waveforms Before and After Filtering

Experimental Results:

1. By converting the MBN signal time-domain waveform into the time-frequency domain using wavelet packet transform, the time-frequency redundant features of the MBN signal are extracted, which can qualitatively reflect different stress levels.

2. A wavelet packet transform coefficient matrix was constructed using the feature vectors of wavelet packet transform coefficients from different frequency bands. Through singular value decomposition, the dimensionality of the MBN feature vectors was reduced, decreasing the complexity of the BP neural network model.

3. A cantilever beam gradient stress calibration experiment was conducted. MBN signals corresponding to the gradient stress on the upper surface of the cantilever beam were collected. A training sample set for the BP neural network was established based on the stress magnitudes at different measurement points.

4. Through an MBN signal acquisition experiment on the gradient stress of the lower surface of a structural steel test block under three-point bending conditions, a test set for the BP neural network was established, and the prediction accuracy of the data-driven model was tested.

5. Using wavelet packet coefficients as time-frequency redundant features, after dimensionality reduction via singular value decomposition, a BP neural network stress prediction model was established. This model exhibits low complexity and fast response speed, with prediction errors for most samples below 10%, achieving data-driven quantitative assessment of surface stress in ferromagnetic components. In practical engineering applications, combined with technologies such as automated robotic arm scanning, sensor adaptive  contour optimization, and signal processing, high-precision imaging of surface stress distribution in ferromagnetic components can be achieved.

   
   

Figure: ATA-4014C High-Voltage Power Amplifier Specifications and Parameters

The experimental materials in this document are compiled and released by Xi'an Aigtek Electronics. Aigtek has become a large-scale instrument and equipment supplier with a wide range of products in the industry, offering demo units for free trial. Xi'an Aigtek Electronics is a high-tech enterprise specializing in the research, development, production, and sales of electronic testing instruments such as power amplifiers, high-voltage amplifiers, power signal sources, preamplifiers for weak signals, high-precision voltage sources, and high-precision current sources.