CN-121637365-B - Power transformer fault early warning method and system

CN121637365BCN 121637365 BCN121637365 BCN 121637365BCN-121637365-B

Abstract

The invention relates to the technical field of power equipment monitoring and fault diagnosis, and discloses a power transformer fault early warning method and system. The method comprises the steps of collecting load current, ambient temperature and top-layer oil temperature sequences of a transformer, calculating initial actual measurement temperature rise based on top-layer oil temperature and ambient temperature, determining thermal response lag time through cross correlation analysis, carrying out translation correction on the load current according to the lag time, constructing a time-aligned excitation input set and an actual measurement temperature rise sequence based on a public effective time window, substituting the excitation input set into a thermal circuit model based on a thermal balance differential equation to calculate theoretical reference temperature rise, calculating residual errors by combining the actual measurement temperature rise sequences, carrying out moving average filtering, constructing a fault feature vector, inputting the fault feature vector into a fault diagnosis classification model to determine fault types and severity, and generating a grading early warning instruction according to the fault type and severity. The method can eliminate thermal hysteresis interference and realize accurate quantitative early warning of latent thermal faults of the transformer under dynamic working conditions.

Inventors

ZHANG WEIGANG
YANG YAJIE
LI YANG
ZHENG LONG
HAO LIJING

Assignees

邢台市华兴电器设备有限公司

Dates

Publication Date: 20260505
Application Date: 20260204

Claims (8)

1. The utility model provides a power transformer fault early warning method which is characterized in that the method comprises the following steps: Collecting real-time running state data of the transformer, wherein the real-time running state data comprises a load current sequence, an environment temperature sequence and a top oil temperature sequence; performing difference operation based on the top layer oil temperature sequence and the environment temperature sequence to obtain an initial actually measured temperature rise sequence, performing cross-correlation analysis on the load current sequence and the initial actually measured temperature rise sequence, and determining thermal response lag time; Performing time sequence translation on the load current sequence according to the thermal response lag time, determining a common effective time window of the translated load current sequence, the environment temperature sequence and the initial measured temperature rise sequence, and performing synchronous data interception based on the common effective time window to construct an excitation input set after time alignment and a measured temperature rise sequence after time alignment; substituting the excitation input set after time alignment into a preset transformer thermal circuit model based on a thermal balance differential equation, and calculating to obtain a theoretical reference temperature rise sequence; Performing time domain differential operation on the actually measured temperature rise sequence after time alignment and the theoretical reference temperature rise sequence to obtain a temperature rise residual sequence, performing moving average filtering treatment on the temperature rise residual sequence, and constructing a fault feature vector based on the processed temperature rise residual sequence; Inputting the fault feature vector into a preset fault diagnosis classification model, and determining a potential fault type label and a fault severity level; If the fault severity level exceeds a preset safety threshold, generating a corresponding grading early warning instruction according to the potential fault type label; Substituting the excitation input set after time alignment into a preset transformer thermal path model based on a thermal balance differential equation, and calculating to obtain a theoretical reference temperature rise sequence, wherein the method comprises the following steps of: According to the load current data in the excitation input set after time alignment, winding loss power and iron core heating power of the transformer are calculated, and a heat source input vector is constructed; acquiring a heat capacity coefficient matrix and a heat resistance network parameter of a transformer, and constructing a first-order heat conduction differential equation set; Substituting the heat source input vector and the environmental temperature data in the excitation input set after time alignment into the first-order heat conduction differential equation set for iterative solution, calculating to obtain a node temperature difference value changing along with time, and generating the theoretical reference temperature rise sequence.
2. The power transformer fault pre-warning method according to claim 1, wherein the collecting real-time operation state data of the transformer, the real-time operation state data including a load current sequence, an ambient temperature sequence and a top-layer oil temperature sequence, includes: Continuously and synchronously acquiring the top layer oil temperature, the ambient temperature and the load current of the transformer according to a preset sampling frequency through a temperature sensor and a current transformer; arranging the collected continuous numerical points according to the collection time stamp sequence to generate an original data stream containing time dimension; and carrying out pretreatment of filling the missing value and removing the abnormal extremum on the original data stream, and extracting according to columns to respectively obtain the load current sequence, the environment temperature sequence and the top layer oil temperature sequence.
3. The power transformer fault pre-warning method according to claim 1, wherein the performing a difference operation based on the top-layer oil temperature sequence and the environmental temperature sequence to obtain an initial measured temperature rise sequence, and performing a cross-correlation analysis on the load current sequence and the initial measured temperature rise sequence to determine a thermal response lag time includes: Subtracting the value of the corresponding moment in the environment temperature sequence from the value of each moment in the top layer oil temperature sequence to generate the initial actually-measured temperature rise sequence; Setting a time sliding window, and calculating a cross correlation coefficient function of the load current sequence and the initial measured temperature rise sequence under different time offsets; and searching a maximum peak value point of the cross-correlation coefficient function, acquiring a time offset corresponding to the maximum peak value point, and determining the time offset as the thermal response lag time.
4. The power transformer fault early warning method according to claim 1, wherein the performing time sequence translation on the load current sequence according to the thermal response lag time, determining a common effective time window of the translated load current sequence, the environment temperature sequence and the initial measured temperature rise sequence, performing synchronous data interception based on the common effective time window, and constructing an excitation input set after time alignment and a measured temperature rise sequence after time alignment, includes: translating the load current sequence back along the time axis for the thermal response lag time to obtain a translated load current sequence; Determining an overlapping time period between an effective starting time point of the translated load current sequence and an effective ending time point of the initial measured temperature rise sequence as the public effective time window; intercepting the translated load current sequence and the environment temperature sequence respectively according to the start-stop time stamps of the public effective time window, and combining to obtain the excitation input set after time alignment; And synchronously intercepting the initial measured temperature rise sequence according to the start-stop time stamp of the public effective time window to obtain the time-aligned measured temperature rise sequence.
5. The power transformer fault pre-warning method according to claim 1, wherein the performing time domain difference operation on the time-aligned measured temperature rise sequence and the theoretical reference temperature rise sequence to obtain a temperature rise residual sequence, performing moving average filtering processing on the temperature rise residual sequence, and constructing a fault feature vector based on the processed temperature rise residual sequence includes: calculating a point-by-point difference value between the actually measured temperature rise sequence after time alignment and the theoretical reference temperature rise sequence to obtain an original residual sequence; Adopting a sliding average filter with a preset window width to carry out smooth denoising on the original residual sequence to obtain the processed temperature rise residual sequence; Extracting mean deviation, fluctuation variance, peak density and amplitude energy of the processed temperature rise residual sequence in a statistical period; And carrying out normalized combination on the mean deviation, the fluctuation variance, the peak value density and the amplitude energy to construct the fault feature vector.
6. The power transformer fault pre-warning method according to claim 1, wherein the inputting the fault feature vector into a preset fault diagnosis classification model to determine a latent fault type label and a fault severity level comprises: Inputting the fault feature vector into the preset fault diagnosis classification model, and calculating the geometric interval distance from the fault feature vector to each fault class hyperplane; Determining classification confidence according to the geometric interval distance, and determining the category with the highest confidence as the potential fault type label; And respectively carrying out normalization processing on the classification confidence coefficient and the fault feature vector, carrying out weighted summation on the normalized classification confidence coefficient and the normalized fault feature vector, and calculating to obtain the fault severity level.
7. The power transformer fault pre-warning method according to claim 1, wherein if the fault severity level exceeds a preset safety threshold, generating a corresponding hierarchical pre-warning instruction according to the latent fault type tag, comprises: Mapping the fault severity level to a preset risk classification table; If the mapping result belongs to the mild risk interval, generating a yellow early warning instruction for suggesting enhanced monitoring; and if the mapping result belongs to the severe risk interval, generating a red early warning instruction for suggesting shutdown maintenance.
8. A power transformer fault early warning system, comprising: the data acquisition module is used for acquiring real-time running state data of the transformer, wherein the real-time running state data comprises a load current sequence, an environment temperature sequence and a top oil temperature sequence; the time sequence analysis module is used for carrying out difference operation on the top layer oil temperature sequence and the environment temperature sequence to obtain an initial actual measurement temperature rise sequence, carrying out cross-correlation analysis on the load current sequence and the initial actual measurement temperature rise sequence, and determining thermal response lag time; The data alignment module is used for carrying out time sequence translation on the load current sequence according to the thermal response lag time, determining a public effective time window of the translated load current sequence, the environment temperature sequence and the initial measured temperature rise sequence, carrying out synchronous data interception based on the public effective time window, and constructing an excitation input set after time alignment and a measured temperature rise sequence after time alignment; The model calculation module is used for substituting the excitation input set after time alignment into a preset transformer thermal circuit model based on a thermal balance differential equation, and calculating to obtain a theoretical reference temperature rise sequence; The feature extraction module is used for carrying out time domain differential operation on the actually measured temperature rise sequence after time alignment and the theoretical reference temperature rise sequence to obtain a temperature rise residual sequence, carrying out moving average filtering treatment on the temperature rise residual sequence, and constructing a fault feature vector based on the treated temperature rise residual sequence; The diagnosis decision module is used for inputting the fault characteristic vector into a preset fault diagnosis classification model and determining a potential fault type label and a fault severity level; The early warning response module is used for generating a corresponding grading early warning instruction according to the potential fault type label if the fault severity level exceeds a preset safety threshold; Substituting the excitation input set after time alignment into a preset transformer thermal path model based on a thermal balance differential equation, and calculating to obtain a theoretical reference temperature rise sequence, wherein the method comprises the following steps of: According to the load current data in the excitation input set after time alignment, winding loss power and iron core heating power of the transformer are calculated, and a heat source input vector is constructed; acquiring a heat capacity coefficient matrix and a heat resistance network parameter of a transformer, and constructing a first-order heat conduction differential equation set; Substituting the heat source input vector and the environmental temperature data in the excitation input set after time alignment into the first-order heat conduction differential equation set for iterative solution, calculating to obtain a node temperature difference value changing along with time, and generating the theoretical reference temperature rise sequence.

Description

Power transformer fault early warning method and system Technical Field The invention relates to the technical field of power equipment monitoring and fault diagnosis, in particular to a power transformer fault early warning method and system. Background At present, the power transformer is used as a core hub for power grid energy transmission and conversion, and the operation reliability of the power transformer is directly related to the safety and stability of the whole power system. With the advancement of smart grid construction, the maintenance strategy of the power equipment is gradually changed from traditional 'post maintenance' and 'regular maintenance' to 'predictive maintenance' based on state awareness, which makes the fault prediction and health management (Prognostics AND HEALTH MANAGEMENT, PHM) technology for the key components of the transformer a hotspot and focus of current industry research. In one prior art, the operating condition of a transformer is typically assessed using a single threshold monitoring method based on oil level thermometer or an off-line diagnostic method based on analysis of dissolved gases in oil (DGA). The method mainly comprises the steps of setting a fixed temperature upper limit value or a gas content threshold value, triggering an alarm when monitoring data exceeds a preset standard, or judging whether the equipment is abnormal based on a static index at a single time point. Since the transformer is a physical system with great thermal inertia, the internal temperature rise of the transformer has obvious thermal response hysteresis relative to the change of load current, and is extremely susceptible to the sudden change of ambient temperature. The traditional static threshold or simple model often ignores the dynamic heat conduction hysteresis characteristic and environmental coupling effect, so that false alarm is easy to generate (normal heat delay is regarded as a fault) when the load severely fluctuates or the environment suddenly changes, and false alarm is generated due to being covered by the noise of an environment substrate when the early weak fault occurs, so that the requirements of high-precision fault prediction and health management on instantaneity and robustness cannot be met. In the prior art, the problem that normal thermal fluctuation and early latency faults are difficult to accurately distinguish under dynamic complex working conditions exists. Disclosure of Invention The invention provides a power transformer fault early warning method and system, which are used for solving the problem that in the prior art, normal heat fluctuation and early latent faults are difficult to accurately distinguish under dynamic complex working conditions. In order to solve the above technical problems, the present invention provides a power transformer fault early warning method, including: Collecting real-time running state data of the transformer, wherein the real-time running state data comprises a load current sequence, an environment temperature sequence and a top oil temperature sequence; performing difference operation based on the top layer oil temperature sequence and the environment temperature sequence to obtain an initial actually measured temperature rise sequence, performing cross-correlation analysis on the load current sequence and the initial actually measured temperature rise sequence, and determining thermal response lag time; Performing time sequence translation on the load current sequence according to the thermal response lag time, determining a common effective time window of the translated load current sequence, the environment temperature sequence and the initial measured temperature rise sequence, and performing synchronous data interception based on the common effective time window to construct an excitation input set after time alignment and a measured temperature rise sequence after time alignment; substituting the excitation input set after time alignment into a preset transformer thermal circuit model based on a thermal balance differential equation, and calculating to obtain a theoretical reference temperature rise sequence; Performing time domain differential operation on the actually measured temperature rise sequence after time alignment and the theoretical reference temperature rise sequence to obtain a temperature rise residual sequence, performing moving average filtering treatment on the temperature rise residual sequence, and constructing a fault feature vector based on the processed temperature rise residual sequence; Inputting the fault feature vector into a preset fault diagnosis classification model, and determining a potential fault type label and a fault severity level; and if the fault severity level exceeds a preset safety threshold, generating a corresponding grading early warning instruction according to the potential fault type label. In a second aspect, the present invention provides a power transformer fau