Abstract : A distributed fiber optic sensing system based on a Mach-Zehnder interferometer can alarm and locate disturbance signals occurring in the sensing area with extremely high sensitivity. This paper first introduces the sensing and positioning principle of the system, and then, based on the idea of ​​pattern recognition, analyzes and discusses the feature identification and extraction method of the input signal of the signal processing module. Signal feature identification technology is of great significance for improving the positioning accuracy of this system, reducing the computational load, and expanding the application range. Keywords : fiber optic sensor, Mach-Zehnder interferometer, pattern recognition, information entropy 1. Introduction A fiber optic sensor is essentially a device. When light emitted from a light source propagates in an optical fiber, under the influence of external physical, chemical, biological, biomedical, or other similar factors, such as stress, temperature, electric field, and magnetic field, the characteristics of the optical waveguide in the fiber, such as amplitude, phase, wavelength, and polarization state, will change. This process is called light wave modulation. Specifically, this involves directing light emitted from a light source, whose optical properties remain unchanged, into an optical fiber using a fixed coupling method. The light then enters a modulation zone and returns from the same or a different optical fiber. The modulation zone can consist of the optical fiber itself or other media. The former is called a sensing-type sensor, and the latter a light-transmitting sensor. There are also sensors that couple light in the optical fiber to the measured quantity through an attenuation field; these are compromise sensors and are generally classified as sensing-type. (Full text download of signal identification analysis for distributed optical fiber sensing positioning systems is available.)