CN-121984012-A - Power transmission and distribution line state evaluation system and method thereof

CN121984012ACN 121984012 ACN121984012 ACN 121984012ACN-121984012-A

Abstract

The invention relates to the technical field of power system automation and intelligent power grids, in particular to a power transmission and distribution line state evaluation system and a method thereof, comprising the steps of model construction, namely constructing a graph model of a power grid and defining topological cognitive entropy indexes; the method comprises the steps of initial evaluation, calculation of an entropy value, calculation of a topology cognition entropy, active detection and reconstruction, wherein the initial evaluation comprises the steps of extracting features by utilizing a graph neural network to obtain initial probability distribution, the calculation of the entropy value comprises the steps of calculating the topology cognition entropy to represent the deviation degree between a digital model and a physical power grid, the active detection and reconstruction comprises the steps of injecting a perturbation excitation signal in response to the entropy value exceeding a threshold value, reconstructing the model by utilizing Bayesian causal reasoning based on physical response data, and outputting a correction result.

Inventors

LIU YUHAN
LI JIAN
Guan Jiahang
XING YE
Lv jiahang
LUO XIAOXUE
LIU BIN
LI ZHONG

Assignees

国网冀北电力有限公司廊坊供电公司

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (8)

1. A power transmission and distribution line state evaluation method, comprising: Constructing a graph model of a power grid comprising a physical power grid topological structure and source load nodes, and defining a topology cognition entropy index for quantifying topology identification uncertainty; acquiring running state data of a power transmission and distribution line in real time, wherein the running state data comprises a node voltage sequence and line tide data; in response to the operational status data update, performing a status assessment procedure comprising: step 1, extracting non-Euclidean space characteristics through a graph neural network model based on the running state data, and acquiring initial probability distribution of a line state; step 2, combining the initial probability distribution with a graph model of the power grid, and calculating topology cognition entropy at the current moment, wherein the topology cognition entropy is used for representing the deviation degree of the digital model and the physical power grid; Step 3, responding to the topology cognition entropy being smaller than a preset safety threshold, judging that the reliability of the model meets the requirement, and outputting a state evaluation result based on the initial probability distribution; Step 4, generating a micro-disturbance excitation signal and injecting the micro-disturbance excitation signal into a physical power grid in response to the topology cognitive entropy being greater than or equal to the preset safety threshold, and observing physical response data of the physical power grid to the micro-disturbance excitation signal; And step 5, reconstructing a graph model of the power grid by utilizing Bayesian causal reasoning based on the physical response data, and outputting a corrected line state evaluation result.
2. The power transmission and distribution line state evaluation method according to claim 1, wherein the graph neural network model comprises: A graph roll stacking layer for processing node features; A gating loop unit for handling time series dependencies; The step 1 comprises the following steps: mapping the running state data into graph structure data; Inputting the graph structure data into a preset graph neural network model, and outputting an implicit layer feature vector through neighborhood aggregation and state updating; and generating the initial probability distribution through full-connection layer mapping based on the implicit layer feature vector.
3. The power transmission and distribution line state evaluation method according to claim 1, wherein the calculating the topology cognitive entropy at the current time in the step 2 includes: based on the initial probability distribution, extracting multiple hypothesized states of the topological structure and corresponding probability values thereof; Calculating the information entropy value of the multiple hypothesis state based on the probability value by using an information entropy calculation formula; Calculating the distance between the current operating point and a preset tripping threshold point in an electric state space to be used as a physical failure boundary distance; And carrying out normalization processing on the information entropy value and the physical failure boundary distance, and carrying out weighted summation according to preset weights to generate the topology cognitive entropy.
4. The power transmission and distribution line state evaluation method according to claim 1, wherein the generating the micro-disturbance excitation signal in the step 4 includes: Acquiring operation constraint parameters of current adjustable equipment; Constructing an optimization function aiming at maximizing physical response difference under different topological assumptions; solving a voltage set point offset based on the optimization function within a range that satisfies the operating constraint parameter; And converting the voltage set point offset into a device control instruction as the perturbation excitation signal.
5. The power transmission and distribution line state evaluation method according to claim 4, wherein the observing physical response data of the physical power grid to the perturbation excitation signal comprises: in a preset time window after the micro-disturbance excitation signal is injected, collecting the voltage fluctuation characteristic and the current abrupt change characteristic of the key node at high frequency; and carrying out space-time alignment on the voltage fluctuation characteristic and the current abrupt change characteristic to form the physical response data.
6. The power transmission and distribution line state evaluation method according to claim 1, wherein the reconstructing the power grid graph model by bayesian causal reasoning in the step 5 includes: constructing a causal graph model based on perturbation excitation and physical response; Calculating posterior probability of the physical response data under different topological assumptions; Selecting a topological structure with the maximum posterior probability as a real physical topology; and correcting the connection relation data of a graph model of the power grid based on the real physical topology.
7. The power transmission and distribution line state evaluation method according to claim 1, further comprising: And after the corrected line state evaluation result is output, feeding back the corrected topological structure to the graph neural network model so as to update the weight parameters of the graph neural network model.
8. A power transmission and distribution line state evaluation system applied to the power transmission and distribution line state evaluation method as claimed in any one of claims 1 to 7, characterized by comprising: the data acquisition module is configured to acquire the running state data of the power transmission and distribution line in real time; The model construction module is configured to construct a graph model of the power grid and define a topology cognition entropy index for quantifying the uncertainty of topology identification; an evaluation control module comprising: the feature extraction unit is configured to acquire initial probability distribution of the line state through a graph neural network model based on the running state data; The entropy value calculating unit is configured to combine the initial probability distribution and a graph model of the power grid to calculate topology cognitive entropy; the active detection unit is configured to respond to the fact that the topology cognitive entropy is larger than or equal to a preset safety threshold value, generate a micro-disturbance excitation signal and acquire physical response data; and the causal reconstruction unit is configured to reconstruct a model by utilizing Bayesian causal reasoning based on the physical response data and output a corrected line state evaluation result.

Description

Power transmission and distribution line state evaluation system and method thereof Technical Field The invention relates to the technical field of power system automation and intelligent power grids, in particular to a power transmission and distribution line state evaluation system and a method thereof. Background In an application scene of power transmission and distribution line state evaluation, a power grid dispatching system guarantees the stability and safety of power supply by means of an accurate physical power grid topological structure model and real-time operation data, and a master station system generally needs to combine a node voltage sequence and line tide data to sense the actual operation state of a power grid in real time; For the evaluation of the line state, a passive state estimation architecture based on a static topology file is generally adopted in the prior art, namely, measurement residual errors or model outputs are directly taken as the basis to judge the line state by weighting least square method or conventional deep learning network fitting measurement data, although the scheme has certain feasibility in an ideal environment with a fixed topological structure and stable source load fluctuation, the scheme is excessively dependent on absolute accuracy of static topology data and lacks a quantification mechanism of model self reliability, when encountering a complex dynamic environment with characteristic confusion caused by severe fluctuation of distributed energy sources, a passive algorithm is extremely easy to identify transient characteristics caused by source load fluctuation as topological faults or can not identify model distortion in a 'ash box' state with hidden change of topology, in addition, a false topology assumption which does not accord with physical relations is difficult to be removed by a pure data driving evaluation method, blind acquisition of an error model is easy to occur in the evaluation process, accurate state sensing and fault early warning under a working condition of a high-elasticity power distribution network is difficult to support, therefore, an evaluation mechanism of model cognition quantification capability is established, and the problem of the actual and real and interactive state is required to be accurately verified when the dynamic state with the fact that the dynamic state is deviated from the physical state is difficult to be effectively identified. The above information disclosed in the background section is only for enhancement of understanding of the background of the disclosure and therefore it may include information that does not form the prior art that is already known to those of ordinary skill in the art. Disclosure of Invention In order to solve the technical problems, the invention discloses a power transmission and distribution line state evaluation system and a method thereof, and specifically, the technical scheme of the invention is as follows: a power transmission and distribution line state assessment method, comprising: Constructing a graph model of a power grid comprising a physical power grid topological structure and source load nodes, and defining a topology cognition entropy index for quantifying topology identification uncertainty; acquiring running state data of a power transmission and distribution line in real time, wherein the running state data comprises a node voltage sequence and line tide data; in response to the operational status data update, performing a status assessment procedure comprising: step 1, extracting non-Euclidean space characteristics through a graph neural network model based on the running state data, and acquiring initial probability distribution of a line state; step 2, combining the initial probability distribution with a graph model of the power grid, and calculating topology cognition entropy at the current moment, wherein the topology cognition entropy is used for representing the deviation degree of the digital model and the physical power grid; Step 3, responding to the topology cognition entropy being smaller than a preset safety threshold, judging that the reliability of the model meets the requirement, and outputting a state evaluation result based on the initial probability distribution; Step 4, generating a micro-disturbance excitation signal and injecting the micro-disturbance excitation signal into a physical power grid in response to the topology cognitive entropy being greater than or equal to the preset safety threshold, and observing physical response data of the physical power grid to the micro-disturbance excitation signal; And step 5, reconstructing a graph model of the power grid by utilizing Bayesian causal reasoning based on the physical response data, and outputting a corrected line state evaluation result. Preferably, the graph neural network model includes: A graph roll stacking layer for processing node features; A gating loop unit for handling time seri