Related Experiment Video
Updated: Feb 14, 2026
11:27
Synthesis and Characterization of Functionalized Metal-organic Frameworks
Published on: September 5, 2014
49.3K
Engineering-Oriented Ultrasonic Decoding: An End-to-End Deep Learning Framework for Metal Grain Size Distribution
Le Dai1, Shiyuan Zhou1, Yuhan Cheng1
1School of Mechanical Engineering, Beijing Institute of Technology, No. 5 South Zhong Guan Cun Street, Haidian, Beijing 100081, China.
Sensors (Basel, Switzerland)
|February 13, 2026
Summary
This study introduces a deep learning model for predicting metallic grain size using ultrasonic data. The novel approach enhances accuracy and adaptability in material characterization.
Area of Science:
- Materials Science
- Non-Destructive Testing
- Artificial Intelligence
Background:
- Grain size significantly impacts metallic material properties and performance.
- Conventional ultrasonic methods for grain size analysis have limitations in adaptability and model assumptions.
- Accurate grain size characterization is crucial for quality control in metallic components.
Purpose of the Study:
- To develop a deep learning architecture for predicting grain size distribution in GH4099 using multimodal ultrasonic features.
- To improve the accuracy and adaptability of ultrasonic inspection for grain size characterization.
- To overcome the limitations of traditional ultrasonic methods.
Main Methods:
- A deep learning model utilizing multimodal ultrasonic features with spatial coding was proposed.
- A-scan signals were converted to time-frequency representations and processed by an encoder-decoder network.
- A thickness-encoding branch and elliptic spatial fusion strategy were incorporated for enhanced prediction.
- The model was trained and validated using C-scan measurements of GH4099.
Main Results:
- The proposed deep learning model achieved mean and standard deviation Mean Absolute Errors (MAEs) of 1.08 and 0.84 μm, respectively.
- The model demonstrated a low Kullback-Leibler (KL) divergence of 0.0031, indicating high prediction accuracy.
- Performance significantly outperformed traditional attenuation- and velocity-based ultrasonic methods.
- Transfer learning enabled rapid performance restoration under new experimental conditions.
Conclusions:
- The developed deep learning approach offers a practical and effective method for grain size characterization using ultrasonic inspection.
- The multimodal feature integration and spatial coding enhance prediction accuracy and adaptability.
- This work paves the way for advanced, AI-driven non-destructive evaluation in materials science.
Related Concept Videos
01:23Pore Size Distribution
495
In concrete, the pore size distribution significantly influences the material's properties. Capillary pores, markedly larger than gel pores, form a vast network within partially hydrated cement paste, reducing the concrete's strength and increasing its permeability. This heightened permeability leads to a greater risk of damage from environmental factors like freeze-thaw cycles and chemical attacks, with the extent of vulnerability also being tied to the water-to-cement ratio.
Adequate...
Adequate...
495
02:32Bonding in Metals
52.9K
Metallic bonds are formed between two metal atoms. A simplified model to describe metallic bonding has been developed by Paul Drüde called the “Electron Sea Model”.
52.9K
02:37Metallic Solids
20.9K
Metallic solids such as crystals of copper, aluminum, and iron are formed by metal atoms. The structure of metallic crystals is often described as a uniform distribution of atomic nuclei within a “sea” of delocalized electrons. The atoms within such a metallic solid are held together by a unique force known as metallic bonding that gives rise to many useful and varied bulk properties.
All metallic solids exhibit high thermal and electrical conductivity, metallic luster, and malleability....
All metallic solids exhibit high thermal and electrical conductivity, metallic luster, and malleability....
20.9K
02:58Properties of Transition Metals
30.1K
Transition metals are defined as those elements that have partially filled d orbitals. As shown in Figure 1, the d-block elements in groups 3–12 are transition elements. The f-block elements, also called inner transition metals (the lanthanides and actinides), also meet this criterion because the d orbital is partially occupied before the f orbitals.
30.1K
03:06Alkali Metals
25.0K
Group 1 elements are soft and shiny metallic solids. They are malleable, ductile, and good conductors of heat and electricity. The melting points of the alkali metals are unusually low for metals and decrease going down the group, while the density increases going down the group with the exception of potassium (Table 1).
Table 1: Properties of the alkali metals
Table 1: Properties of the alkali metals
25.0K
01:13Location and Orientation of the Heart
10.9K
The human heart, despite its modest size and weight, is an organ of remarkable strength and endurance. Roughly the size of a fist, the heart weighs between 250 and 350 grams and is nestled within the mediastinum, the medial cavity of the thorax. It extends obliquely for about 12 to 14 cm, resting on the superior surface of the diaphragm. The heart is positioned anterior to the vertebral column and posterior to the sternum, with two-thirds of its mass lying to the left of the midsternal line.
10.9K