CN-121983944-A - Knowledge-driven digital twin-aware power distribution network state estimation method and system

CN121983944ACN 121983944 ACN121983944 ACN 121983944ACN-121983944-A

Abstract

The invention belongs to the technical field of power distribution network state estimation, in particular to a knowledge-driven digital twin-perception power distribution network state estimation method and system, which comprises the steps of firstly constructing a parameter correction generation type countermeasure network and a state estimation model, introducing a generated data correction function taking key parameters, random noise and multi-mode environment information of a power distribution network as input into a loss function of a generator and a discriminator of the parameter correction generation type countermeasure network, the multi-mode environment information of the power distribution network comprises load fluctuation, meteorological conditions and equipment health states, and then key parameters in the operation process of the power distribution network after parameter correction generation type countermeasure network expansion are used as input data to be input into a state estimation model, so that power distribution network operation state estimation is obtained. The method and the device can automatically optimize the generation and discrimination strategy when facing the running condition of the dynamic power grid, and improve the estimation precision of the whole state.

Inventors

YANG JIE
TANG JING
ZHOU YUJIE
YANG ZILI
ZHENG ZIJIAN
PENG WENYAN
LI YAXIN
XIA FANGZHOU
Shu dongsheng
XU JINGYOU
YE GAOXIANG
SHAO FEIFAN
ZHAO HONGSHENG
CHEN KE

Assignees

国网湖北省电力有限公司经济技术研究院

Dates

Publication Date: 20260505
Application Date: 20251128

Claims (10)

1. The knowledge-driven digital twin-perception power distribution network state estimation method is characterized by comprising the following steps of: The method comprises the following steps: The method comprises the steps of S1, constructing a parameter correction generation type countermeasure network, wherein the parameter correction generation type countermeasure network is used for expanding key parameter data quantity in the running process of a power distribution network, and introducing a generated data correction function taking key parameters, random noise and multi-mode environment information of the power distribution network as input into a loss function of a generator and a discriminator, wherein the multi-mode environment information of the power distribution network comprises load fluctuation, meteorological conditions and equipment health states; S2, a state estimation model is built, current key parameters of the power distribution network are input into the parameter correction generation type countermeasure network, output data of the parameter correction generation type countermeasure network and the current key parameters of the power distribution network are used as input data together to be input into the state estimation model, and a power distribution network running state estimation result is obtained.
2. The knowledge driven digital twin aware distribution network state estimation method of claim 1, wherein: In the step S2, the state estimation model obtains a preliminary state estimation result of the running state of the power distribution network through fuzzy classification based on input data, and then verifies the preliminary state estimation result through Kalman filtering to obtain a final running state estimation result of the power distribution network.
3. The knowledge driven digital twin aware distribution network state estimation method of claim 2, wherein: the fuzzy classification includes: firstly, defining a fuzzy set and a triangle membership function, wherein the fuzzy set comprises different power distribution network running states, and the triangle membership function has the expression: ; ; ; ; In the above-mentioned method, the step of, For inputting data At fuzzy aggregation Membership in (a) is determined; 、、 three vertexes of triangle respectively, formed by sliding window Is obtained by setting different quantiles; 、、 respectively at the current sliding window Lower quantile, median, upper quantile of the inner key parameter value; and (3) carrying out fuzzification processing on the input data, obtaining a membership matrix in the fuzzy set through a triangular membership function, and taking the membership matrix as a preliminary state estimation result of the running state of the power distribution network.
4. A knowledge-driven digital twin-aware distribution network state estimation method according to claim 2 or 3, characterized in that: The Kalman filtering comprises the steps of inputting a preliminary state estimation result output by fuzzy classification into a Kalman filtering model as an observation value, obtaining a predicted running state output by the Kalman filtering model, calculating Kalman gain, and correcting the predicted running state output by the Kalman filtering model based on the Kalman gain to obtain a final power distribution network running state estimation result.
5. The knowledge driven digital twin aware distribution network state estimation method of claim 4, wherein: The calculation formula of the power distribution network running state estimation result is as follows: ; In the above-mentioned method, the step of, Is that A power distribution network running state estimation result at moment; Is that The predicted running state output by the moment Kalman filtering model; Is that A Kalman gain at time; Is that The observation value of the moment, namely the preliminary state estimation result output by fuzzy classification; for observing matrix for describing predicted running state And observed value The linear mapping relation between the two components satisfies , Is that Innovative residual error of the moment Kalman filtering model; Is that Is a transpose of (2); The calculation formula of the Kalman gain is as follows: ; In the above-mentioned method, the step of, Is composed of Obtained by a lightweight scaling operation; Is that Prediction error covariance of the time-of-day Kalman filter model.
6. The knowledge-driven digital twin-aware distribution network state estimation method of claim 5, wherein: the observed noise covariance is obtained by a lightweight scaling operation according to the following formula : ; In the above-mentioned method, the step of, Is a scaling threshold; The coefficients are adjusted for the threshold.
7. The knowledge driven digital twin aware distribution network state estimation method of claim 1, wherein: The method further comprises the steps of: S4, uncertainty analysis is carried out on the obtained power distribution network running state estimation result, a confidence interval is calculated, accuracy of the power distribution network running state estimation result is verified based on the confidence interval, and a calculation formula of the confidence interval is as follows: ; ; In the above-mentioned method, the step of, Is a confidence interval; taking the running state estimation results of the power distribution network at different moments as a group of samples for sample mean values, and calculating the sample mean values of the group of samples; is the sample size; A normal distribution threshold for a standard at a given confidence level; taking the power distribution network running state estimation results at different moments as a group of samples for sample standard deviation, and calculating the sample standard deviation of the group of samples; Is the first And estimating the running state of the power distribution network.
8. The knowledge-driven digital twin-perception power distribution network state estimation system is characterized in that: The system comprises a data correction module and a state estimation module; The parameter correction generation type countermeasure network introduces a generated data correction function taking key parameters, random noise and multi-modal environment information of the power distribution network as input into a loss function of a generator and a discriminator, wherein the multi-modal environment information of the power distribution network comprises load fluctuation, meteorological conditions and equipment health states; The state estimation module is used for constructing a state estimation model, inputting current key parameters of the power distribution network into the parameter correction generation type countermeasure network, and inputting output data of the parameter correction generation type countermeasure network and the current key parameters of the power distribution network into the state estimation model together as input data to obtain a power distribution network running state estimation result.
9. Knowledge-driven digital twin-perception power distribution network state estimation equipment is characterized in that: The knowledge-driven digital twin-aware distribution network state estimation device comprises a memory and a processor, wherein the memory is used for storing computer program codes and transmitting the computer program codes to the processor, and the processor is used for executing the knowledge-driven digital twin-aware distribution network state estimation method according to the instructions in the computer program codes.
10. A computer readable storage medium having stored thereon a computer program which when executed by a processor implements the knowledge driven digital twin aware distribution network state estimation method of claim 1.

Description

Knowledge-driven digital twin-aware power distribution network state estimation method and system Technical Field The invention belongs to the technical field of power distribution network state estimation, and particularly relates to a knowledge-driven digital twin-aware power distribution network state estimation method, system, equipment and medium. Background With the popularization of renewable energy sources and the development of smart grid technologies, the complexity of a power distribution network is continuously increased, and the traditional state estimation method is gradually insufficient. These methods typically rely on static models that cannot accommodate dynamically changing operating conditions, resulting in inefficiency in failure detection and resource scheduling. Therefore, a method for improving the ability of dynamic monitoring and real-time state estimation of a power distribution network by considering environmental information is urgently needed. Disclosure of Invention The invention aims to solve the problems in the prior art and provide a knowledge-driven digital twin-aware power distribution network state estimation method, system, equipment and medium capable of adapting to dynamic change operation conditions of a power grid and improving real-time state estimation accuracy. In order to achieve the above object, the technical scheme of the present invention is as follows: in a first aspect, the present invention provides a knowledge-driven digital twin-aware power distribution network state estimation method, the method comprising: The method comprises the steps of S1, constructing a parameter correction generation type countermeasure network, wherein the parameter correction generation type countermeasure network is used for expanding key parameter data quantity in the running process of a power distribution network, and introducing a generated data correction function taking key parameters, random noise and multi-mode environment information of the power distribution network as input into a loss function of a generator and a discriminator, wherein the multi-mode environment information of the power distribution network comprises load fluctuation, meteorological conditions and equipment health states; S2, a state estimation model is built, current key parameters of the power distribution network are input into the parameter correction generation type countermeasure network, output data of the parameter correction generation type countermeasure network and the current key parameters of the power distribution network are used as input data together to be input into the state estimation model, and a power distribution network running state estimation result is obtained. In the step S2, the state estimation model obtains a preliminary state estimation result of the running state of the power distribution network through fuzzy classification based on input data, and then verifies the preliminary state estimation result through Kalman filtering to obtain a final running state estimation result of the power distribution network. The fuzzy classification includes: firstly, defining a fuzzy set and a triangle membership function, wherein the fuzzy set comprises different power distribution network running states, and the triangle membership function has the expression: ; ; ; ; In the above-mentioned method, the step of, For inputting dataAt fuzzy aggregationMembership in (a) is determined;、、 three vertexes of triangle respectively, formed by sliding window Is obtained by setting different quantiles;、、 respectively at the current sliding window Lower quantile, median, upper quantile of the inner key parameter value; and (3) carrying out fuzzification processing on the input data, obtaining a membership matrix in the fuzzy set through a triangular membership function, and taking the membership matrix as a preliminary state estimation result of the running state of the power distribution network. The Kalman filtering comprises the steps of inputting a preliminary state estimation result output by fuzzy classification into a Kalman filtering model as an observation value, obtaining a predicted running state output by the Kalman filtering model, calculating Kalman gain, and correcting the predicted running state output by the Kalman filtering model based on the Kalman gain to obtain a final power distribution network running state estimation result. The calculation formula of the power distribution network running state estimation result is as follows: ; In the above-mentioned method, the step of, Is thatA power distribution network running state estimation result at moment; Is that The predicted running state output by the moment Kalman filtering model; Is that A Kalman gain at time; Is that The observation value of the moment, namely the preliminary state estimation result output by fuzzy classification; for observing matrix for describing predicted running state And observed valueThe li