Search

CN-122017438-A - Regression confidence band-based transformer electricity degree-temperature rise monitoring method

CN122017438ACN 122017438 ACN122017438 ACN 122017438ACN-122017438-A

Abstract

The invention discloses a transformer electricity degree-temperature rise monitoring method based on a regression confidence band, which relates to the technical field of electric variable measurement, and the method comprises the steps of calculating an electricity degree variation and a temperature rise value by collecting historical operation data of a transformer, carrying out linear regression analysis by taking the electricity degree variation as an independent variable and Wen Shengzhi as a dependent variable after pretreatment to obtain a linear regression equation and a regression standard error, and constructing a dynamic confidence band based on the regression standard error; and collecting current data in real time, and if the current temperature rise value exceeds the dynamic confidence band range, judging that the transformer is abnormal in operation and giving out early warning. According to the invention, the dynamic confidence band is used for replacing the traditional fixed threshold value, so that different load working conditions of the transformer can be self-adapted, the false alarm rate and the missing report rate are obviously reduced, and the accuracy and the reliability of temperature rise monitoring are improved.

Inventors

  • HAN HONGXIAN
  • WU LIPING
  • YU DONG
  • LIANG TAO
  • OU JIANJUN
  • WU HAI
  • SHEN FENG

Assignees

  • 宁波天仑电气股份有限公司

Dates

Publication Date
20260512
Application Date
20260416

Claims (8)

  1. 1. The transformer electricity degree-temperature rise monitoring method based on the regression confidence band is characterized by comprising the following steps of: S1, collecting historical operation data of a transformer, wherein the historical operation data at least comprises ambient temperature, equipment temperature and electric energy reading; S2, calculating the electric power variation and the temperature rise value of each sampling moment in a preset time window according to historical operation data, wherein the temperature rise value is the difference between the equipment temperature and the environment temperature; S3, preprocessing historical operation data, wherein the preprocessing comprises the steps of filtering invalid or abnormal data, and performing tail cutting processing on a temperature rise value corresponding to the same power variation so as to retain data with a middle preset proportion; S4, linear regression analysis is carried out by taking the electric power variable quantity in the preprocessed data as an independent variable and the temperature rise value as a dependent variable, so as to obtain a linear regression equation and a regression standard error; S5, calculating a dynamic confidence band based on the regression standard error, wherein the upper limit and the lower limit of the dynamic confidence band are respectively the regression standard error of a preset multiple added and subtracted from the predicted value of the linear regression equation; s6, collecting current operation data of the transformer in real time, and calculating a current electric temperature change amount and a current temperature rise value; and S7, substituting the current power variation into a linear regression equation to obtain a temperature rise predicted value, judging whether the current temperature rise value is in a dynamic confidence band, if so, returning to the step S6, and if not, judging that the transformer is abnormal in operation and sending out early warning information.
  2. 2. The regression confidence band-based transformer power-temperature rise monitoring method according to claim 1, wherein in the step S2, the preset time window is 1 hour, the power variation is the variation of the electric energy reading within 1 hour, and the power variation is rounded to an integer by rounding.
  3. 3. The method for monitoring the power-temperature rise of a transformer based on a regression confidence band according to claim 1, wherein in the step S3, the tail-cutting treatment is performed on the temperature rise value corresponding to the same power variation to retain the data of the middle preset proportion, specifically: And (3) sorting a plurality of temperature rise values with the same electric power variation according to the values, removing data of each preset proportion at two ends, and reserving data of the middle part as effective temperature rise data of the electric power variation.
  4. 4. The method for monitoring the power-temperature rise of a transformer based on a regression confidence band of claim 1, wherein in the step S3, the preprocessing further comprises: And filtering the historical operation data with the temperature rise value being greater than or equal to 50 ℃ to remove invalid samples under sensor faults or extreme abnormal working conditions.
  5. 5. The method for monitoring the power-temperature rise of a transformer based on a regression confidence band according to claim 1, wherein in the step S4, the linear regression analysis is performed when the amount of data after the pretreatment is greater than or equal to a preset amount, otherwise, the regression analysis is skipped and the linear regression equation and the dynamic confidence band which are successfully constructed last time are used.
  6. 6. The method for monitoring the power-temperature rise of a transformer based on a regression confidence band according to claim 1, wherein in the step S5, the preset multiple is 2.0, the corresponding dynamic confidence band is a linear regression prediction value ± 2 times regression standard error, and the dynamic confidence band corresponds to a 95% confidence level.
  7. 7. The regression confidence band-based transformer power-temperature rise monitoring method of claim 1, further comprising the steps of: And S8, storing regression parameters, regression standard errors and preset multiples of a dynamic confidence band of the linear regression equation into a database for the abnormal early warning call of an external monitoring system.
  8. 8. The method for monitoring the power-temperature rise of a transformer based on a regression confidence band according to claim 1, further comprising a standby early warning method: in the S1 step, the maximum value and the minimum value of the historical temperature rise corresponding to each integral electric power change are calculated and stored in advance; In the step S4, when the linear regression analysis cannot be executed or the dynamic confidence band is not available, the historical temperature rise extremum corresponding to the current electric power variation is used as an alarm boundary.

Description

Regression confidence band-based transformer electricity degree-temperature rise monitoring method Technical Field The invention relates to the technical field of electric variable measurement, in particular to a transformer electric degree-temperature rise monitoring method based on a regression confidence band. Background In the existing transformer running state monitoring technology, temperature monitoring is a key link for guaranteeing safe and stable running of equipment. Conventionally, a fixed threshold method is generally adopted in engineering practice to monitor the temperature rise of a transformer, namely, a fixed upper limit value of the temperature rise is preset, and when the temperature rise monitored in real time exceeds the threshold value, the system judges that the temperature rise is abnormal and gives an alarm. The method is simple to implement and small in calculated amount, and is widely applied to early-stage power system monitoring. In addition, part of monitoring systems can combine single variables such as load rate or current to construct simplified segmentation threshold values or static empirical formulas to attempt to roughly adjust alarm boundaries under different working conditions. In the prior art, a statistical method based on historical data is adopted, for example, the average value and standard deviation of temperature rise in a specific load interval are calculated, the average value plus or minus k times of standard deviation is used as an alarm boundary, and the method considers the distribution characteristics of historical operation data to a certain extent. However, the prior art has obvious limitations that the fixed threshold method cannot adapt to the dynamic change of the load of the transformer, is easy to misreport due to the natural low temperature rise under the low load working condition, and is also easy to misreport due to the normal high temperature rise under the high load working condition, the static empirical formula based on the segmentation threshold is improved to a certain extent, but the boundary setting still depends on manual experience, and is difficult to accurately attach to the actual operation characteristics of equipment, while the method based on the simple statistical interval utilizes the historical data, but the boundary is independently calculated by the discrete load interval, a continuous physical relation model is not established, and the boundary has jump and cannot characterize the inherent linear correlation between the electric variation and the temperature rise. In general, the prior art lacks a temperature rise abnormality judgment mechanism which can dynamically adapt to different load working conditions, establish continuous function relation and have statistical confidence meaning, so that the self-adaptation capability of a monitoring system is insufficient, and the false alarm rate are difficult to balance. Disclosure of Invention In order to dynamically adapt to different load working conditions and establish a continuous function relation and a temperature rise abnormality judgment mechanism with statistical confidence significance, the invention provides a transformer electricity-temperature rise monitoring method based on a regression confidence band, which comprises the following steps: S1, collecting historical operation data of a transformer, wherein the historical operation data at least comprises ambient temperature, equipment temperature and electric energy reading; S2, calculating the electric power variation and the temperature rise value of each sampling moment in a preset time window according to historical operation data, wherein the temperature rise value is the difference between the equipment temperature and the environment temperature; S3, preprocessing historical operation data, wherein the preprocessing comprises the steps of filtering invalid or abnormal data, and performing tail cutting processing on a temperature rise value corresponding to the same power variation so as to retain data with a middle preset proportion; S4, linear regression analysis is carried out by taking the electric power variable quantity in the preprocessed data as an independent variable and the temperature rise value as a dependent variable, so as to obtain a linear regression equation and a regression standard error; S5, calculating a dynamic confidence band based on the regression standard error, wherein the upper limit and the lower limit of the dynamic confidence band are respectively the regression standard error of a preset multiple added and subtracted from the predicted value of the linear regression equation; s6, collecting current operation data of the transformer in real time, and calculating a current electric temperature change amount and a current temperature rise value; and S7, substituting the current power variation into a linear regression equation to obtain a temperature rise predicted value, judgi