CN-121980463-A - Transformer oil flow abnormality detection method, device and equipment

CN121980463ACN 121980463 ACN121980463 ACN 121980463ACN-121980463-A

Abstract

The application discloses a transformer oil flow anomaly detection method, device and equipment, which comprise the steps of collecting multi-mode data in the running process of a transformer, determining a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field in a mode of reconstructing a fluid field, carrying out feature fusion on the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and a vibration spectrum through a preset time-space correlation model to obtain fusion feature vectors, carrying out anomaly analysis processing on the fusion feature vectors by adopting a preset VAE (variable value) based on reconstruction contrast to determine anomaly scores, and carrying out anomaly detection analysis based on dynamic threshold and feedback optimization according to the anomaly scores and load power to obtain anomaly detection results. The method can solve the technical problems that the prior art relies on fixed threshold and manual detection, and is difficult to capture the multi-physical field coupling characteristic and the time-space correlation characteristic of the abnormal oil flow, so that the abnormal oil flow detection efficiency and the abnormal oil flow detection accuracy of the transformer in an actual scene are low, and the application requirements are difficult to meet.

Inventors

TANG QI
GAO XUE
FAN XINMING
WANG ZHIJIAO
Liang Nianbai
HUANG JING
HU ZHIPENG

Assignees

广东电网有限责任公司佛山供电局

Dates

Publication Date: 20260505
Application Date: 20260127

Claims (10)

1. The method for detecting the abnormal oil flow of the transformer is characterized by comprising the following steps of: Acquiring multi-mode data in the operation process of the transformer, and determining a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field by reconstructing a fluid field, wherein the multi-mode data comprises oil flow velocity, oil temperature distribution and vibration spectrum; Performing feature fusion on the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration frequency spectrum through a preset space-time correlation model to obtain fusion feature vectors; Performing anomaly analysis processing on the fusion feature vector based on reconstruction contrast by adopting a preset VAE, and determining an anomaly score; and carrying out abnormality detection analysis based on dynamic threshold and feedback optimization according to the abnormality score and the load power to obtain an abnormality detection result, wherein the abnormality detection result comprises an abnormality positioning and alarming signal.
2. The method for detecting abnormal transformer oil flow according to claim 1, wherein the steps of collecting multi-modal data during operation of the transformer and determining a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field by reconstructing a fluid field include: acquiring oil flow speed, oil temperature distribution and vibration spectrum of the transformer in the operation process by an electromagnetic flow, a distributed optical fiber temperature sensor and an MEMS vibration sensor respectively to obtain multi-mode data; Respectively carrying out standardization treatment on the oil flow velocity, the oil temperature distribution and the vibration spectrum; reconstructing a fluid field from the normalized oil flow velocity and the oil temperature profile based on a temperature expansion coefficient; and performing simplified difference calculation based on a central difference method on the fluid field to obtain a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field.
3. The method for detecting transformer oil flow anomaly according to claim 1, wherein the feature fusion is performed on the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration spectrum by a preset space-time correlation model to obtain a fusion feature vector, comprising: modeling the oil duct topological structure of the transformer according to the graph convolution coding operation to obtain an oil duct topological graph model; based on the oil duct topological graph model, constructing a preset time-space correlation model through graph convolution calculation and LSTM time sequence calculation; And inputting the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration spectrum into the preset space-time correlation model for feature fusion to obtain a fusion feature vector.
4. The transformer oil flow anomaly detection method according to claim 1, wherein the performing anomaly analysis processing based on reconstruction contrast on the fusion feature vector by using a preset VAE, determining an anomaly score, comprises: Performing feature coding processing based on potential variables on the fusion feature vectors to obtain coded potential variables; calculating a reconstruction error according to the fusion feature vector and the reconstruction feature vector based on a preset VAE; And carrying out exception analysis processing according to the encoding potential variable, the reconstruction error and the preset KL divergence, and determining an exception score, KL divergence loss and updating the potential variable.
5. The method for detecting abnormal transformer oil flow according to claim 4, wherein the performing an abnormality detection analysis based on a dynamic threshold and feedback optimization according to the abnormality score and the load power to obtain an abnormality detection result comprises: Constructing a threshold updating formula according to the KL divergence loss, the updated potential variable and the load power, and carrying out feedback optimization calculation to obtain a dynamic threshold; And if the abnormal score exceeds the dynamic threshold, triggering to generate an alarm signal, and realizing abnormal positioning according to the flow velocity gradient to obtain an abnormal detection result.
6. An abnormal transformer oil flow detection device, comprising: The reconstruction computing unit is used for acquiring multi-mode data in the operation process of the transformer and determining a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field in a mode of reconstructing a fluid field, wherein the multi-mode data comprises oil flow velocity, oil temperature distribution and vibration frequency spectrum; the feature fusion unit is used for carrying out feature fusion on the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration frequency spectrum through a preset space-time correlation model to obtain fusion feature vectors; the anomaly analysis unit is used for carrying out anomaly analysis processing on the fusion feature vector based on reconstruction contrast by adopting a preset VAE and determining an anomaly score; And the abnormality detection unit is used for carrying out abnormality detection analysis based on dynamic threshold and feedback optimization according to the abnormality score and the load power to obtain an abnormality detection result, wherein the abnormality detection result comprises an abnormality positioning signal and an alarm signal.
7. The transformer oil flow anomaly detection device of claim 6, wherein the reconstruction calculation unit is specifically configured to: acquiring oil flow speed, oil temperature distribution and vibration spectrum of the transformer in the operation process by an electromagnetic flow, a distributed optical fiber temperature sensor and an MEMS vibration sensor respectively to obtain multi-mode data; Respectively carrying out standardization treatment on the oil flow velocity, the oil temperature distribution and the vibration spectrum; reconstructing a fluid field from the normalized oil flow velocity and the oil temperature profile based on a temperature expansion coefficient; and performing simplified difference calculation based on a central difference method on the fluid field to obtain a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field.
8. The transformer oil flow anomaly detection device of claim 6, wherein the feature fusion unit is specifically configured to: modeling the oil duct topological structure of the transformer according to the graph convolution coding operation to obtain an oil duct topological graph model; based on the oil duct topological graph model, constructing a preset time-space correlation model through graph convolution calculation and LSTM time sequence calculation; And inputting the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration spectrum into the preset space-time correlation model for feature fusion to obtain a fusion feature vector.
9. The transformer oil flow anomaly detection device of claim 6, wherein the anomaly analysis unit is specifically configured to: Performing feature coding processing based on potential variables on the fusion feature vectors to obtain coded potential variables; calculating a reconstruction error according to the fusion feature vector and the reconstruction feature vector based on a preset VAE; And carrying out exception analysis processing according to the encoding potential variable, the reconstruction error and the preset KL divergence, and determining an exception score, KL divergence loss and updating the potential variable.
10. An equipment for detecting abnormal oil flow of transformer, characterized in that the equipment comprises a processor and a memory; the memory is used for storing program codes and transmitting the program codes to the processor; the processor is configured to execute the transformer oil flow anomaly detection method of any one of claims 1-5 according to instructions in the program code.

Description

Transformer oil flow abnormality detection method, device and equipment Technical Field The application relates to the field of transformer anomaly monitoring, in particular to a method, a device and equipment for detecting transformer oil flow anomaly. Background As a core device of a power system, a transformer oil flow abnormality detection technology mainly relies on a fixed threshold method and a single-mode sensor analysis, such as an electromagnetic flowmeter or a temperature sensor, but the abnormality detection technology based on the transformer oil flow abnormality detection technology still has obvious defects. The adoption of a preset fixed threshold value can not cope with dynamic working conditions such as load fluctuation of the transformer, seasonal temperature change and the like, and high-frequency false alarm or missing alarm can be caused. While the prior art is mostly based on a single sensor, the characteristic of multi-physical field coupling of abnormal oil flow is difficult to capture. In addition, the existing majority of algorithms ignore space-time correlation, have strong manual dependence, and are difficult to adapt to the complex working condition of the novel transformer particularly due to the fact that threshold calibration and fault diagnosis depend on expert experience. These technical defects lead to the difficulty in meeting the detection requirements of practical application scenes in the existing transformer oil flow abnormality detection technology. Disclosure of Invention The application provides a transformer oil flow abnormality detection method, device and equipment, which are used for solving the technical problems that the prior art relies on fixed threshold and manual detection, and the multi-physical field coupling characteristic and time-space correlation characteristic of the oil flow abnormality are difficult to capture, so that the transformer oil flow abnormality detection efficiency and accuracy of an actual scene are low, and the application requirements are difficult to meet. In view of the above, the first aspect of the present application provides a method for detecting abnormal transformer oil flow, including: Acquiring multi-mode data in the operation process of the transformer, and determining a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field by reconstructing a fluid field, wherein the multi-mode data comprises oil flow velocity, oil temperature distribution and vibration spectrum; Performing feature fusion on the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration frequency spectrum through a preset space-time correlation model to obtain fusion feature vectors; Performing anomaly analysis processing on the fusion feature vector based on reconstruction contrast by adopting a preset VAE, and determining an anomaly score; and carrying out abnormality detection analysis based on dynamic threshold and feedback optimization according to the abnormality score and the load power to obtain an abnormality detection result, wherein the abnormality detection result comprises an abnormality positioning and alarming signal. Preferably, the collecting multi-mode data during the operation of the transformer and determining the flow velocity gradient matrix, the pressure fluctuation matrix and the temperature field by reconstructing the fluid field includes: acquiring oil flow speed, oil temperature distribution and vibration spectrum of the transformer in the operation process by an electromagnetic flow, a distributed optical fiber temperature sensor and an MEMS vibration sensor respectively to obtain multi-mode data; Respectively carrying out standardization treatment on the oil flow velocity, the oil temperature distribution and the vibration spectrum; reconstructing a fluid field from the normalized oil flow velocity and the oil temperature profile based on a temperature expansion coefficient; and performing simplified difference calculation based on a central difference method on the fluid field to obtain a flow velocity gradient matrix, a pressure fluctuation matrix and a temperature field. Preferably, the feature fusion is performed on the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration spectrum through a preset space-time correlation model to obtain a fusion feature vector, which includes: modeling the oil duct topological structure of the transformer according to the graph convolution coding operation to obtain an oil duct topological graph model; based on the oil duct topological graph model, constructing a preset time-space correlation model through graph convolution calculation and LSTM time sequence calculation; And inputting the flow velocity gradient matrix, the pressure fluctuation matrix, the temperature field and the vibration spectrum into the preset space-time correlation model for feature fusion t