Local Hot Spot Detection of Transformer Windings Based on Brillouin Optical Time Domain Peak Edge Analysis

The temperature distribution of transformer windings has always been the focus of power grid operators. Compared with traditional sensors, distributed fiber sensing has great advantages such as strong anti-electromagnetic interference ability and distributed measurement. Brillouin optical time domain analysis (BOTDA) has stable performance and is suitable for most scenarios. However, the spatial resolution of BOTDA is on the order of meters, making it difficult to distinguish local hot spots in winding operation. Starting from the mechanism of Brillouin scattering gain spectrum, this paper proposes a Brillouin optical time domain peak edge analysis (BOTDA-PEA) technique to improve the spatial resolution of BOTDA without increasing the complexity of the system hardware by describing the bimodal fitting principle and analyzing the edge of the abnormal peak intensity curve in the disturbed fiber, so as to meet the requirements of local hot spot detection in transformer windings. The basic principle of this technology is that when the length of the disturbed fiber is less than the spatial resolution, the hot spot is precisely located according to the descending edge of the subpeak scattering intensity curve, the traditional fitting results are modified according to the corresponding Brillouin frequency shift of the subpeak, and the actual hot spot temperature is demodulated. Firstly, three sets of simulation experiments were designed under laboratory conditions to study the performance of BOTDA-PEA in terms of different locations, different hot spot lengths and different hot spot temperatures along the fiber. When the length of the disturbed fiber segment was less than the spatial resolution of BOTDA, the influence of different variables on the Brillouin gain spectrum is obtained by using the control variable method. The experimental results verify that the position corresponding to the declining edge of the subpeak intensity curve in the bimodal region of Brillouin scattering spectrum is the exact position of the disturbed fiber segment, and its frequency shift represents the temperature information of the disturbed fiber segment. Compared with traditional single-peak fitting technology, BOTDA-PEA technology can stably improve the spatial resolution of BOTDA sensor system to less than half of the original spatial resolution, and up to 2 times of the sampling resolution. The local temperature test of transformer winding was simulated by setting heating strips, and the temperature distribution curves of local hot spots and their vicinity were obtained by using BOTDA-PEA technique to process the data. The experimental results show that the single-peak fitting technique used by traditional BOTDA cannot accurately identify the hot spot length less than the spatial resolution, and the temperature error is as high as 34%. However, BOTDA-PEA technique can successfully demodulate the temperature, and the relative error is controlled within 5%. It is concluded that BOTDA-PEA technology can realize the accurate location and length perception of winding abnormal hot spots and accurately demodulate the temperature of local hot spots in winding, which further verifies the practical feasibility and detection superiority of this method, and provides new thinking for improving the economy and feasibility of online monitoring of transformer winding temperature. At the same time, it provides a new idea for the early warning of local winding fault. © 2024 China Machine Press. All rights reserved.