Identification of Reflection Peaks in Multiparameter Constrained Based on Frequency Domain Reflection Method

In the current Frequency Domain Reflection method (FDR), the reflection peak identification mainly relies on the subjective experience judgment of human, and lacks the criteria for clear determination. This paper provides a multiparameter constrained reflection peak identification method based on based on FDR. Firstly, the Locally Weighted Regression (LOESS) is used to calibrate the location spectrum and extract the effective features. Secondly, local mean values, full width at half maximum (FWHM) and extreme values are used comprehensively, as the basis for identifying the reflection peaks. Finally, to prove the validity of the proposed method, it is tested on a 2000 m in-service cable. The experimental findings show that the method can identify five reflection peaks in the cable and locates them. It provides an effective solution for identifying reflection peaks in a cable based on the FDR.