Splice fault position detection of single mode fiber using feed-forward neural networks

Abstract

AbstractThe purpose of this work is to detect the splice fault that decrease the signal level of single mode optical fiber for a length of 10 Km by using an optical source of 1550 nm wavelength, the principle of work depends on the comparison between fiber undefected and fiber that contains defect which are its signal were taken by optical time domain reflectometer (OTDR), this comparison was operated by artificial neural networks, it is found that the splice fault location inside the optical fiber can be detected with high speed using feed-forward neural (FFN) networks, the proposed Mean Square Error (MSE) is (10e-7) to check the performance of (FFN), the results that obtained by using MATLAB programming were with high accuracy .