Feature extraction method based on wavelet transform for the pressure tactile reflection spectrum of long chirped fiber grating

Abstract

In this study, a feature extraction method for long chirped fiber grating (LCFBG) pressure tactile sensing spectral signal is proposed and verified. Firstly, the pressure sensing theory and spectral characteristics of LCFBG are analyzed, including spectral pattern, peak, FWHM and other characteristic parameters. The spectral characteristics of LCFBG were analyzed by VI transfer matrix method. Based on the simulation data, different wavelet basis functions are compared and analyzed for the effect of spectral information feature extraction. Bior4.4 wavelet basis function was used to decompose the denoised spectrum at four levels, and the error of spectral zero and extreme point was the minimum, up to 0.4 pm. To verify the extraction effect of LCFBG pressure tactile sensing spectral features, the point contact, surface contact and deformation sensing systems of LCFBG are established, and the experimental results are analyzed by using wavelet transform feature extraction algorithm. The results show that the sensitivity coefficients of the sensing system for point contact force loading and surface contact force loading are 1.04 dBm/N and 1.87 dBm·mm/N respectively. The linearity of the reflection spectrum intensity and the pressure contact force is above 99.5 %. In the deformation sensing experiment, the reflection spectrum has a good linear relationship with the strain change, and the central wavelength interval of the peak and valley determines the width of the deformation region. The relevant research results can provide reference and direction for the feature extraction and recognition of LCFBG tactile sensing spectral signals.