研究目的
To develop a bio-inspired training mechanism for support vector regression for shape sensing in structures mounted with Fiber Bragg Gratings, aiming for high accuracy, low computational requirements, and enhanced prediction times.
研究成果
The proposed bio-inspired training method for support vector regression demonstrated high accuracy, low computational requirements, and enhanced prediction times for shape sensing in structures mounted with Fiber Bragg Gratings. It showed significant advantages over conventional training mechanisms, making it suitable for embedded sensing solutions where power and computational complexities are limited.
研究不足
The experimental setup was scaled down for controlled environments, which may not fully represent real-life applications. The method's accuracy might decrease with increased setup sizes or greater non-linearities.
1：Experimental Design and Method Selection:
A bio-inspired training mechanism for support vector regression was designed for shape sensing using Fiber Bragg Gratings. The methodology involved experimental validation on a simply supported beam and an aircraft wing model under different loading conditions.
2：Sample Selection and Data Sources:
A simply supported aluminum beam and an aircraft wing model were used as samples. Data was collected from the wavelength shifts of Fiber Bragg Gratings mounted on these structures.
3：List of Experimental Equipment and Materials:
Equipment included Fiber Bragg Gratings, an Optical Spectrum Analyzer (Yokogawa AQ6317C), a broadband source (Denselight DL-BX9-CS5254A), and a National Instruments PXIe-4844 Optical Sensor Interrogator.
4：Experimental Procedures and Operational Workflow:
The beam and wing were loaded at different positions, and the deflections were monitored using a travelling microscope and photogrammetry. The wavelength shifts from the FBGs were recorded and used to train the SVR model.
5：Data Analysis Methods:
The deflection values were interpreted from the wavelength shifts using the specially modified Support Vector Regression. The method's accuracy and computational efficiency were compared with conventional techniques.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
