CN-122022759-A - Method and device for predicting tripping risk of substation equipment, computer equipment and medium

CN122022759ACN 122022759 ACN122022759 ACN 122022759ACN-122022759-A

Abstract

The application relates to the technical field of smart grids, in particular to a tripping risk prediction method, a tripping risk prediction device, computer equipment and a medium for power transformation equipment, which comprise the steps of obtaining time sequence data of the power transformation equipment, preprocessing the time sequence data, and obtaining a preprocessed time sequence data set; extracting time domain features through mean value, variance, maximum value, minimum value, skewness and kurtosis methods based on the preprocessed time sequence data set, extracting frequency features through a Fourier transform method, extracting time domain-frequency domain features through a short-time Fourier transform or wavelet transform method, screening the extracted features through principal component analysis or LASSO method to obtain a feature data set, inputting the feature data set into a pre-trained deep learning model to obtain a tripping risk prediction result, generating an emergency treatment scheme based on the tripping risk prediction result, and notifying operation and maintenance personnel of the emergency treatment scheme.

Inventors

ZHANG YING
WANG XUE
SHAO YING
LI MEI
LIN SHUGUANG
LI DONGBO
YANG YONG

Assignees

云南电网有限责任公司文山供电局

Dates

Publication Date: 20260512
Application Date: 20251218

Claims (10)

1. A method for predicting tripping risk of a power transformation device, the method comprising: Acquiring time sequence data of power transformation equipment, and preprocessing the time sequence data to obtain a preprocessed time sequence data set; extracting time domain features by means of mean, variance, maximum, minimum, skewness and kurtosis methods, extracting frequency features by means of Fourier transform methods and extracting time domain-frequency domain features by means of short-time Fourier transform or wavelet transform methods based on the preprocessed time sequence data set; screening the extracted features by a principal component analysis or LASSO method to obtain a feature data set; Inputting the characteristic data set into a pre-trained deep learning model to obtain a tripping risk prediction result; and generating an emergency treatment scheme based on the tripping risk prediction result, and notifying operation and maintenance personnel of the emergency treatment scheme.
2. The method for predicting tripping risk of power transformation equipment according to claim 1, wherein the step of obtaining the time sequence data of the power transformation equipment, preprocessing the time sequence data, and obtaining the preprocessed time sequence data comprises the following steps: denoising the time sequence data through Kalman filtering and wavelet transformation; And (3) performing linear interpolation on the denoised time sequence data, and supplementing the missing time sequence data to obtain a complete time sequence data set.
3. The power transformation equipment trip risk prediction method according to claim 2, wherein the expression of the kalman filter is: ; Wherein, the Is the kalman gain, H is the observation matrix, Is an observation value of the current, Is time sequence data; The wavelet transform has the expression: ; Wherein, the Is the original timing signal to be analyzed, Is a mother wavelet, a is a scale factor, a is equal to 0, b is a shift factor; is a normalization factor; the expression of the linear interpolation is: ; Wherein t is the timestamp corresponding to the missing data point to be filled, which satisfies the following condition ≤t≤ Is the interpolation interval. Refers to the time stamps corresponding to two known valid data points, 。
4. The method of claim 1, wherein the step of inputting the feature data set into a deep learning model to obtain a trip risk prediction result comprises: constructing the feature data set into a feature sequence arranged according to time steps; Inputting the feature sequence into the pre-trained deep learning model, wherein the deep learning model is a long-short-term memory network model LSTM or a gating cycle unit network model GRU; performing forward propagation calculation on the feature sequence by using the deep learning model, sequentially processing the input of each time step, and updating the internal hidden state of the feature sequence; And mapping to obtain the tripping risk prediction result based on the finally output hidden state of the deep learning model.
5. The method of claim 1, wherein the time series data of the power transformation device is an electrical parameter of the power transformation device collected by a current sensor, a voltage sensor, a temperature sensor and a vibration sensor.
6. The power transformation equipment trip risk prediction method according to claim 1, wherein the step of generating an emergency treatment scheme based on the trip risk prediction result and notifying an operation and maintenance person of the emergency treatment scheme comprises: Acquiring a fault type and a risk level in the tripping risk prediction result; Performing repair mode mapping based on the fault type and the risk level to obtain an emergency treatment scheme; And notifying the emergency treatment scheme to operation and maintenance personnel.
7. The power transformation device trip risk prediction method according to claim 1, further comprising: acquiring repair execution information from the operation and maintenance personnel; And optimizing the deep learning model through a weighted binary cross entropy loss function based on the restoration execution information.
8. A power transformation equipment trip risk prediction apparatus, the apparatus comprising: The data acquisition module is used for acquiring time sequence data of the power transformation equipment, preprocessing the time sequence data and obtaining a preprocessed time sequence data set; The risk prediction module is used for extracting time domain features through mean, variance, maximum value, minimum value, skewness and kurtosis methods, extracting frequency features through a Fourier transform method and extracting time domain-frequency domain features through a short-time Fourier transform or wavelet transform method based on the preprocessed time sequence data set; Based on the time domain, frequency domain and time-frequency domain characteristics, performing characteristic selection by using an LASSO method to obtain a characteristic data set; Inputting the characteristic data set into a deep learning model to obtain a tripping risk prediction result; And the emergency treatment scheme generating module is used for generating an emergency treatment scheme based on the tripping risk prediction result and notifying the emergency treatment scheme to operation and maintenance personnel.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the power transformation device trip risk prediction method of any one of claims 1 to 7 when the computer program is executed.
10. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when executed by a processor, implements the steps of the power transformation device trip risk prediction method of any one of claims 1 to 7.

Description

Method and device for predicting tripping risk of substation equipment, computer equipment and medium Technical Field The application relates to the technical field of smart grids, in particular to a tripping risk prediction method and device for power transformation equipment, computer equipment and medium. Background The transformer equipment is an important component part in the power system and is mainly used for receiving, converting and distributing electric energy, and the main task of the transformer equipment is to realize the transmission and adjustment of the electric energy through components such as a transformer, an electric switch, protective equipment and the like. The equipment is tripped, the tripping brings loss to power supply enterprises, and also brings great influence to the daily life of residents and the production of industrial enterprises, so that the customer experience of the power supply enterprises is seriously influenced, the dynamic and influence analysis and emergency decision after the equipment is tripped generally need the integration of multiple information sources, and deep model mechanism research and data mining are carried out. At present, the digital technology has a short board in the practical aspects of real-time sensing and accurate research and judgment of equipment tripping, and is mainly characterized in that: the data automatic tracking and intelligent linkage practical research based on the defect abnormality of the equipment in the station are still in a blank state, and the effect of improving the fault handling efficiency and the accuracy of the intelligent substation construction is further improved; The data of secondary equipment and auxiliary equipment in the station still have isomerism, so that the data are unified and efficiently fused, and the data utilization is urgently needed to be deeply dug; the co-processing technology application of device anomalies/faults stays at a shallower level, failing to achieve fault handling policy persona integration. Disclosure of Invention Based on the above, it is necessary to provide a tripping risk prediction method, a device, a computer device and a medium for transformer equipment, aiming at the technical problems that the existing tripping early warning method of the transformer substation in the prior art is single in feature extraction dimension, low in model prediction precision and poor in interpretability, and is difficult to realize accurate early warning and quick closed loop treatment due to insufficient feature capture caused by feature redundancy and lack of effective screening. In a first aspect of the present application, there is provided a method for predicting tripping risk of a power transformation device, the method comprising: Acquiring time sequence data of power transformation equipment, and preprocessing the time sequence data to obtain a preprocessed time sequence data set; extracting time domain features by means of mean, variance, maximum, minimum, skewness and kurtosis methods, extracting frequency features by means of Fourier transform methods and extracting time domain-frequency domain features by means of short-time Fourier transform or wavelet transform methods based on the preprocessed time sequence data set; screening the extracted features by a principal component analysis or LASSO method to obtain a feature data set; Inputting the characteristic data set into a pre-trained deep learning model to obtain a tripping risk prediction result; and generating an emergency treatment scheme based on the tripping risk prediction result, and notifying operation and maintenance personnel of the emergency treatment scheme. Preferably, the step of obtaining the time sequence data of the power transformation device, preprocessing the time sequence data, and obtaining the preprocessed time sequence data includes: denoising the time sequence data through Kalman filtering and wavelet transformation; And (3) performing linear interpolation on the denoised time sequence data, and supplementing the missing time sequence data to obtain a complete time sequence data set. Preferably, the expression of the kalman filter is: ; Wherein, the Is the kalman gain, H is the observation matrix,Is an observation value of the current,Is time sequence data; The wavelet transform has the expression: ; Wherein, the Is the original timing signal to be analyzed,Is a mother wavelet, a is a scale factor, a is equal to 0, b is a shift factor; is a normalization factor; the expression of the linear interpolation is: ; Wherein t is the timestamp corresponding to the missing data point to be filled, which satisfies the following condition ≤t≤Is the interpolation interval. Refers to the time stamps corresponding to two known valid data points,。 Preferably, the step of inputting the feature data set into a deep learning model to obtain a trip risk prediction result includes: constructing the feature data set into a fe