CN-122017715-A - Intelligent diagnosis method for electromagnetic transformer

CN122017715ACN 122017715 ACN122017715 ACN 122017715ACN-122017715-A

Abstract

The invention relates to the technical field of fault diagnosis of power equipment, in particular to an intelligent diagnosis method for an electromagnetic transformer, which comprises the steps of collecting operation signals of the electromagnetic transformer, wherein the operation signals comprise vibration signals, temperature rise signals and partial discharge signals; the method comprises the steps of inputting diagnosis characteristics into a preset fault diagnosis model, obtaining diagnosis results, generating corresponding early warning signals, carrying out early warning, and determining maintenance strategies of the electromagnetic transformer based on the early warning signals and historical data of the electromagnetic transformer. According to the invention, by collecting multisource operation signals such as vibration, temperature rise and partial discharge and extracting diagnosis characteristics, a fault diagnosis model based on a support vector machine is introduced, and the high-precision intelligent diagnosis of the degradation state and faults of the electromagnetic transformer is realized. The early warning signal can be generated, potential hidden trouble can be found in advance, the sudden fault and power failure risk are reduced, and the maintenance strategy is optimized by combining historical data, so that the transition from post-emergency repair to state maintenance and predictive maintenance is realized.

Inventors

CHEN YADI
LIU CHUANHUI
LI RUIDA
WU JIAFEI
FENG ERHAO
ZHANG XIAOBING
ZHAO YILIN
ZHANG XIANG
CHEN YAOQIANG
YIN SHI
Qi Haoxing
ZHANG WEN
You Aobo
LI GUOWEI
ZHANG RUILIANG
SUN KE
QIN YUELONG
YU NAN
ZHOU HAORAN
QIN JINGWEN
LIU DI
CHEN QINGFENG
CHEN SEN
DU BO
FENG HAIZHOU
ZHANG BAOSHAN
SUN XUEWU
GU WEI
Ye Wupeng
HE RENKE

Assignees

许继集团有限公司
许继集团有限公司科创分公司
河南源网荷储电气研究院有限公司

Dates

Publication Date: 20260512
Application Date: 20251128

Claims (10)

1. An intelligent diagnostic method for an electromagnetic transformer, comprising: collecting operation signals of the electromagnetic transformer, wherein the operation signals comprise vibration signals, temperature rise signals and partial discharge signals; Extracting the characteristics of the operation signals to obtain diagnosis characteristics; inputting the diagnosis characteristics into a preset fault diagnosis model to obtain a diagnosis result, wherein the preset fault diagnosis model is based on a support vector machine algorithm; Based on the diagnosis result, generating a corresponding early warning signal for early warning; And determining a maintenance strategy of the electromagnetic transformer based on the early warning signal and the historical data of the electromagnetic transformer.
2. The intelligent diagnostic method for an electromagnetic transformer according to claim 1, wherein the diagnostic features include vibration signal features, temperature rise signal features, and partial discharge signal features; the vibration signal characteristics comprise power spectral density, dominant frequency, harmonic frequency, root mean square value and kurtosis; The temperature rise signal characteristics comprise temperature change rate, temperature gradient, maximum temperature rise and average temperature rise; The partial discharge signal features include discharge amount, discharge frequency, discharge phase and partial discharge pulse width.
3. The intelligent diagnostic method for an electromagnetic transformer according to claim 1, wherein the fault diagnostic model comprises a kernel function, the kernel function being a linear kernel function, a polynomial kernel function, or a radial basis kernel function, the kernel function being used to map the diagnostic features to a high dimensional space.
4. The intelligent diagnostic method for an electromagnetic transformer according to claim 3, wherein the kernel function of the fault diagnostic model is expressed using the following formula: Wherein, the The output value of the kernel, i.e. the similarity of sample x to sample x' in high dimensional space, Representing the square of the euclidean distance between two samples, σ represents the width parameter of the kernel.
5. The intelligent diagnostic method for an electromagnetic transformer according to claim 1, wherein the preset fault diagnosis model is represented by the following formula: Where f (x) represents the diagnostic result, alpha i represents the Lagrangian multiplier, y i represents the label of the sample numbered i, Representing a kernel function, and b representing a bias term.
6. The intelligent diagnostic method for an electromagnetic transformer according to claim 5, wherein the fault diagnostic model is trained, the training comprising: Initializing the punishment parameters and the width parameters of the kernel function, wherein the punishment parameters represent punishment degrees of misclassification of the fault diagnosis model; constructing an optimization objective function; Solving the objective function by adopting a preset optimization algorithm to obtain an optimal value; determining a supporting sample range based on the optimal value; Calculating a bias term based on any one of the support samples within the support sample range; And obtaining a trained fault diagnosis model according to the optimal value and the bias term.
7. The intelligent diagnostic method for an electromagnetic transformer according to claim 6, wherein the optimized objective function is expressed using the following formula: wherein U represents an optimization objective function; It means that all possible alpha vectors are optimized to find an optimal alpha vector to minimize the value of the objective function, the alpha vector contains lagrangian multipliers for all samples, alpha i and alpha j are the components inside the vector alpha, represent lagrangian multipliers for each component, y i represents the label of the sample numbered i, y j represents the label of the sample numbered j, and K (x i ,x j ) represents the kernel function.
8. The intelligent diagnostic method for an electromagnetic transformer according to claim 6, wherein the bias term is calculated using the following formula: Where b represents the bias term, alpha i represents the Lagrangian multiplier, y i represents the label of the sample numbered i, The kernel function is represented by a function of the kernel, Representing the support samples, y s represents the labels of the support samples.
9. The intelligent diagnosis method for electromagnetic transformer according to claim 1, wherein the generating a corresponding early warning signal based on the diagnosis result, and performing early warning, comprises: When the diagnosis result is a typical fault, generating corresponding early warning signals, wherein the typical fault comprises turn-to-turn short circuit, insulation degradation and iron core local saturation, and the early warning signals comprise fault types, fault suspected positions, fault development trends and fault severity; And sending the early warning signal to an operation and maintenance personnel in real time.
10. The intelligent diagnostic method for an electromagnetic transformer according to claim 1, wherein the determining a maintenance strategy for the electromagnetic transformer based on the pre-warning signal and the historical data of the electromagnetic transformer comprises: based on the early warning signals and the historical data of the electromagnetic transformer, the health state and the residual life of the electromagnetic transformer are evaluated, and an evaluation result is obtained; and determining a maintenance strategy of the electromagnetic transformer according to the evaluation result.

Description

Intelligent diagnosis method for electromagnetic transformer Technical Field The invention relates to the technical field of fault diagnosis of power equipment, in particular to an intelligent diagnosis method for an electromagnetic transformer. Background The electromagnetic transformer is used as key measurement and protection equipment in a power system, is widely applied to current and voltage acquisition and transmission links, and the running state of the electromagnetic transformer is directly related to the safety and stability of the power system. Along with the continuous expansion of the power grid scale and the continuous improvement of the load level, the transformer operates in high voltage, high current and complex electromagnetic environments for a long time, and degradation and fault problems such as turn-to-turn short circuit, insulation aging, iron core local saturation, local discharge enhancement and the like are easy to occur. Once the transformer fails, measurement errors, abnormal power protection actions and even equipment damage and large-scale power failure accidents can be caused, so that the method has important significance for real-time monitoring of the running state and early diagnosis of the faults. In the prior art, electromagnetic transformer state monitoring mainly depends on periodic inspection of operation and maintenance personnel or simple threshold judgment of a single signal, and comprehensive analysis of multi-source operation signals such as vibration, temperature rise, partial discharge and the like is difficult to realize, meanwhile, the traditional diagnosis method often cannot accurately describe the mechanism characteristics of transformer material degradation, and has limited identification capability on early hidden troubles such as slight turn-to-turn short circuit, insulation partial degradation and the like. In addition, the prior art generally lacks an active early warning mechanism and an equipment maintenance strategy making method based on a diagnosis result, so that the predictability of faults is not strong, and the equipment overhaul replacement efficiency is low. Disclosure of Invention Object of the invention The invention aims to provide an intelligent diagnosis method for an electromagnetic transformer, which is used for realizing high-precision intelligent diagnosis of the degradation state and faults of the electromagnetic transformer by collecting multisource operation signals such as vibration, temperature rise, partial discharge and the like, extracting diagnosis characteristics, and introducing a fault diagnosis model based on a support vector machine. The early warning signal can be generated, potential hidden trouble can be found in advance, the sudden fault and power failure risk are reduced, and the maintenance strategy is optimized by combining historical data, so that the transition from post-emergency repair to state maintenance and predictive maintenance is realized. (II) technical scheme In order to solve the above problems, the present invention provides an intelligent diagnosis method for an electromagnetic transformer, comprising: collecting operation signals of the electromagnetic transformer, wherein the operation signals comprise vibration signals, temperature rise signals and partial discharge signals; Extracting the characteristics of the operation signals to obtain diagnosis characteristics; inputting the diagnosis characteristics into a preset fault diagnosis model to obtain a diagnosis result, wherein the preset fault diagnosis model is based on a support vector machine algorithm; Based on the diagnosis result, generating a corresponding early warning signal for early warning; And determining a maintenance strategy of the electromagnetic transformer based on the early warning signal and the historical data of the electromagnetic transformer. In another aspect of the present invention, preferably, the diagnostic features include vibration signal features, temperature rise signal features, and partial discharge signal features; the vibration signal characteristics comprise power spectral density, dominant frequency, harmonic frequency, root mean square value and kurtosis; The temperature rise signal characteristics comprise temperature change rate, temperature gradient, maximum temperature rise and average temperature rise; The partial discharge signal features include discharge amount, discharge frequency, discharge phase and partial discharge pulse width. In another aspect of the present invention, preferably, the fault diagnosis model includes a kernel function, the kernel function being a linear kernel function, a polynomial kernel function, or a radial basis kernel function, the kernel function being used to map the diagnosis feature to a high-dimensional space. In another aspect of the present invention, preferably, the kernel function of the fault diagnosis model is expressed by the following formula: Wherein, t