Identification of Loading Location and Amplitude in Conductive Composite Materials via Deep Learning Method

Area of Science:

• Materials Science
• Electrical Engineering
• Artificial Intelligence

Background:

• Current electrical self-sensing methods for composite structural health monitoring (SHM) have limitations.
• These include complex electrode arrangements and difficulties in precisely locating and quantifying external loads.

Purpose of the Study:

• To develop a deep learning-based self-sensing method for accurate load localization and amplitude identification.
• To address the limitations of existing methods in composite SHM.

Main Methods:

• Preparation of conductive filler-filled resin composites with 3D conductive networks.
• Installation of four electrodes on the composite plate edges to collect boundary electrical resistance responses.
• Development of a residual learning-based Convolutional Neural Network (CNN) model for load analysis.

Main Results:

• The trained CNN model accurately determined both the location and amplitude of applied mechanical loads.
• Achieved a localization error of 0.91 mm and an amplitude error of 0.13 N.
• Demonstrated superior performance compared to previous studies.

Conclusions:

• The proposed deep learning method offers a highly accurate and efficient approach for predicting load location and amplitude.
• This technique has significant potential for widespread application in composite structural health monitoring.