Intelligent power protection scene building method based on digital twinning

Abstract

The invention belongs to the field of three-dimensional modeling of computer mapping, and particularly relates to an intelligent power protection scene building method based on digital twinning. The method comprises the following steps: firstly, comprehensively collecting equipment operation, state and power grid topology data; an improved graph neural network is combined with multi-scale and time sequence characteristic modeling, and the equipment relation is deeply analyzed. The method is characterized in that based on a digital twinborn model, an equipment fault probability, fault propagation and load fluctuation are comprehensively considered to construct a power protection scene. The influence of the scene on the stability of the power grid is accurately evaluated by dynamically simulating the evolution of the power grid. The method can dynamically adjust the scene according to the actual working condition, effectively improves the scientificity and effectiveness of the construction of the power guarantee scene, guarantees the stable operation of a power grid, and provides powerful support for the power guarantee work of a power system.

Description

Technical Field

[0001] The present invention belongs to the field of three-dimensional modeling of computer graphics, and in particular to a method for building an intelligent power protection scene based on digital twins. Background Art

[0002] With the rapid development of social economy and the continuous advancement of science and technology, various industries have increasingly higher requirements for the stability and reliability of power supply, and the importance of intelligent power supply scenario construction technology in the field of power systems has become increasingly prominent. In the current operation of the power system, traditional power supply methods mainly rely on manual experience and simple monitoring methods. In the face of increasingly complex power grid structures and diversified power demand, these traditional methods have gradually shown many defects. For example, when the power grid topology changes or faces emergencies, traditional power supply measures are difficult to adjust strategies quickly and accurately, resulting in low power supply efficiency and poor results. The existing power supply scenario construction lacks systematicity and intelligence. In most cases, it is impossible to fully utilize a large amount of equipment operation data, equipment status information, and power grid topology data for in-depth analysis and accurate prediction. This makes it often impossible to fully consider the impact of various potential factors on power grid stability when constructing power supply scenarios, making it difficult to formulate the optimal power supply plan. At the same time, there is a lack of effective dynamic evaluation methods for evaluating the impact of power supply scenarios on power grid stability. Traditional evaluation methods are mostly based on static models or simple empirical formulas. They cannot track changes in grid operating conditions in real time, cannot accurately predict the dynamic response characteristics of the grid under different power protection scenarios, and cannot promptly detect possible stability problems. Furthermore, with the continuous expansion of the scale of the grid and the sharp increase in the number of devices, the interaction between devices has become extremely complex, and traditional technologies have difficulty in effectively modeling and analyzing such complex interactions. For example, the interaction between devices has multi-scale and time-series characteristics, and existing methods have difficulty integrating these characteristics into power protection scenario construction and stability assessment. Summary of the invention

[0003] In response to the technical problems existing in the background technology, the present invention proposes a method for building an intelligent power protection scenario based on digital twins, which has a reasonable design, simple method, strong theoretical basis and can be implemented.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is: comprising the following steps:

[0005] S1, first collect the equipment operation data, equipment status and power grid topology data;

[0006] S2. Secondly, we use improved graph neural network and topological data to build a model. Taking into account the different scales and timing issues of the interactions between devices, we introduce multi-scale and timing characteristics. The final node feature representation formula can be integrated into the following form: in is the final node feature representation, ω l is the weight of the l-th layer feature, For the timing characteristics considered, For the multiple scales considered;

[0007] S3. Finally, based on the digital twin model, the method generates multiple power supply scenarios, simulates the operating conditions of the generated power supply scenarios, and dynamically evaluates the impact of the power supply scenarios on the stability of the power grid; the specific implementation of the power supply scenarios and operating condition simulation in step S3 is:

[0008] S31. First, the improved dynamic failure probability formula is used to calculate the failure probability of each device, taking into account the device operating status, environmental conditions and the impact of failure propagation of adjacent devices: The dynamic correction factor ψ i (t) including load fluctuation ΔL i , voltage deviation V i deviation and temperature change ΔT i , the correlation factor φ i (t) Reflect the impact of failures in adjacent equipment;

[0009] S32, based on the equipment failure probability P(f i ,t) Generate the initial faulty device set: F k (t)={irand()<P(f i ,t)}, expand the initial fault set and consider the fault propagation to neighboring devices: Where N(i) is the set of neighbor devices of device i, ξ ij (t) is the impact factor of the failure of device j due to the propagation of the failure of device i;

[0010] S33, combined with the generated fault set F k ′(t) and load fluctuation, define the power protection scenario: S k (t) = {F k ′(t),L t +ΔL t}, where L t Load distribution at time t, ΔL t is the charge fluctuation in the scenario, ensuring that the scenario includes dynamic adjustment of fault and load conditions;

[0011] S34. Finally, each scenario is dynamically simulated, and the evolution of the grid operation status is simulated through time series integration: Where P i (t) is the active power of node i at time t, Q i(t) is reactive power, V i (t), V j (t) The voltage amplitude between nodes i and j at time t, G ij , B ij is the real and imaginary part of the admittance between nodes i and j, θ ij (t) is the voltage phase difference between node i and node j, ΔP i (t), ΔQ i (t) is the dynamic power correction term of node i at time t, which dynamically updates the voltage and phase to simulate the evolution of power conservation.

[0012] Preferably, in step S1, the equipment operation data includes voltage, current, power, and load; the equipment status includes equipment working time, temperature, humidity, and vibration; and the grid topology data includes grid topology and switch status data.

[0013] Preferably, the specific steps of modeling in step S2 are:

[0014] S21. First, define the power system graph as G = (V, E), where V is the device node set, E represents the connection relationship between devices, and the characteristic of node i is expressed as The adjacency matrix of the graph is A, and the time step is t;

[0015] S22. In the multi-scale graph convolution layer, the adjacency matrix A of different scales is used. l To handle the interactions between power devices, for each scale l, the update formula of node i is: Among them, Nl(i) represents the neighbor set of node i in the adjacency relationship of the lth layer, is the dynamic attention coefficient calculated by the graph attention mechanism, W l (l) is the weight matrix of the lth layer, is the bias term, σ is the ReLU function;

[0016] S23. Combined with time series modeling to capture the dynamic behavior of power equipment and networks, based on the features of graph convolution output, LSTM modeling time dependency is introduced. The time series modeling formula of node i at time step t is: in is the state of node i at the previous time step, is the input feature of node i at the current time step;

[0017] S24. Finally, the final node feature representation is obtained by combining multi-scale and temporal characteristics:

[0018] Preferably, the dynamic attention coefficient calculated by the graph attention mechanism in step S22 The calculation formula is: in is the dynamic attention coefficient between node i and node j at layer l, and a is the learned attention weight vector.

[0019] As a preference, the influencing factor of the fault in step S32 is Among them, P(f j ,t) is the dynamic failure probability of j at time t, d ij is the distance between devices i and j.

[0020] Preferably, the formula for dynamically updating the voltage and phase in step S34 is: in It represents the dynamic change rate of phase and voltage, and α and β are the update step sizes.

[0021] Compared with the prior art, the advantages and positive effects of the present invention are that the introduction of multi-scale and time series characteristics in the modeling process can accurately handle complex relationships between equipment, which is difficult to achieve with traditional technologies. In terms of fault probability calculation, the improved formula that integrates multiple factors makes the prediction more accurate. The present invention realizes the dynamic construction and evaluation of power supply protection scenarios based on the digital twin model, which can be adjusted in real time according to the working conditions. Traditional technologies are mostly static and lack dynamic adaptability. The present invention effectively improves the scientific nature of power supply protection scenario construction and the ability to ensure grid stability. DETAILED DESCRIPTION

[0022] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0024] Embodiment, the present invention aims at the problem that when constructing the existing power protection scenario, the impact of various potential factors on the stability of the power grid is often not fully considered, and overcomes the defects of the existing modeling technology to carry out reasonable power protection scenario construction and dynamic power grid stability evaluation. In the method for building an intelligent power protection scenario based on digital twins of the present invention, data collection is first performed to collect the operating data, equipment status and power grid topology data of the equipment. The equipment operation data includes voltage, current, power, and load. The equipment status includes the working time, temperature, humidity, and vibration of the equipment. The power grid topology data includes power grid topology and switch status data, so as to fully understand the structure and connection relationship of the power grid. By collecting these multi-type and all-round data, a data foundation is provided for subsequent precise modeling and accurate analysis.

[0025] In order to effectively handle the complex interaction between devices, the traditional method is difficult to consider the defects of multi-scale and timing characteristics. This paper uses improved graph neural network and topological data for modeling, taking into account the different scales and timing problems of the interaction between devices, and introduces multi-scale and timing characteristics. The specific implementation is as follows: First, define the power system graph as G = (V, E), where V is the device node set, E represents the connection relationship between devices, and the feature of node i is represented as The adjacency matrix of the graph is A, and the time step is t; in the multi-scale graph convolution layer, the adjacency matrix A of different scales is used l To handle the interactions between power devices, for each scale l, the update formula of node i is: Among them, Nl(i) represents the neighbor set of node i in the adjacency relationship of the lth layer, is the dynamic attention coefficient calculated by the graph attention mechanism, W l (l) is the weight matrix of the lth layer, is the bias term, σ is the ReLU function; combined with time series modeling to capture the dynamic behavior of power equipment and networks, based on the features of graph convolution output, LSTM modeling time dependency is introduced, and the time series modeling formula of node i at time step t is: in is the state of node i at the previous time step, is the input feature of node i at the current time step; finally, the final node feature representation is obtained by combining multi-scale and time series characteristics: Existing technologies often ignore multi-scale and time-series characteristics when dealing with device interactions, resulting in an inaccurate understanding of device relationships. However, the solution of the present invention can carefully depict the complex interactions between devices and provide a model foundation that is more in line with the actual situation for building power conservation scenarios.

[0026] In order to accurately calculate the probability of equipment failure and improve the single and incomplete situation of traditional failure probability calculation methods, this invention comprehensively considers temperature changes, load fluctuations, and the impact of neighboring equipment failure propagation, so as to timely discover the situation where the risk of equipment failure increases significantly and make preparations for power protection in advance. First, the improved dynamic failure probability formula is used to calculate the failure probability of each device, taking into account the equipment operating status, environmental conditions, and the impact of neighboring equipment failure propagation: The dynamic correction factor ψ i (t) including load fluctuation ΔL i , voltage deviation V i deviation and temperature change ΔT i , the correlation factor φ i (t) reflects the impact of failure of adjacent equipment; based on the failure probability P(f i ,t) Generate the initial faulty device set: F k (t)={irand()<P(f i ,t)}, expand the initial fault set and consider the fault propagation to neighboring devices: Where N(i) is the set of neighbor devices of device i, ξ ij (t) is the impact factor of device j failing due to the fault propagation of device i Among them, P(f j ,t) is the dynamic failure probability of j at time t, d ij is the distance between devices i and j; combined with the generated fault set F k ′(t) and load fluctuation, define the power protection scenario: S k (t) = {F k ′(t),L t +ΔL t}, where L t Load distribution at time t, ΔL t It is the charge fluctuation in the scenario, ensuring that the scenario includes dynamic adjustment of faults and load conditions; finally, each scenario is dynamically simulated, simulating the evolution of the grid operation state through time series integration:

[0027] Where P i (t) is the active power of node i at time t, Q i (t) is reactive power, V i (t), V j (t) The voltage amplitude between nodes i and j at time t, where in represents the dynamic change rate of phase and voltage, α, β are the update step size, G ij , B ijis the real and imaginary part of the admittance between nodes i and j, θ ij (t) is the voltage phase difference between node i and node j, ΔP i (t), ΔQ i (t) is the dynamic power correction term of node i at time t, dynamically updates the voltage and phase, and simulates the evolution of power supply protection. The present invention solves the deficiencies in data, models, dynamics, etc. in the existing power supply protection scene construction, so as to achieve more efficient, accurate, and dynamic power supply protection scene construction, ensure the stable operation of the power system under various working conditions, and meet the strict requirements for power supply reliability.

[0028] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for building an intelligent power conservation scenario based on digital twins, characterized in that: The following steps are involved: S1, first collect the equipment operation data, equipment status and power grid topology data; S2. Secondly, we use improved graph neural network and topological data to build a model. Taking into account the different scales and timing issues of the interactions between devices, we introduce multi-scale and timing characteristics. The final node feature representation formula can be integrated into the following form: in is the final node feature representation, ω l is the weight of the l-th layer feature, For the timing characteristics considered, For the multiple scales considered; S3. Finally, based on the digital twin model, the method generates multiple power supply scenarios, simulates the operating conditions of the generated power supply scenarios, and dynamically evaluates the impact of the power supply scenarios on the stability of the power grid; the specific implementation of the power supply scenarios and operating condition simulation in step S3 is: S31. First, the improved dynamic failure probability formula is used to calculate the failure probability of each device, taking into account the device operating status, environmental conditions and the impact of failure propagation of adjacent devices: The dynamic correction factor ψ i (t) including load fluctuation ΔL i , voltage deviation V i deviation and temperature change ΔT i , the correlation factor φ i (t) Reflect the impact of failures in adjacent equipment; S32, based on the equipment failure probability P(f i ,t) Generate the initial faulty device set: F k (t)={irand()<P(f i ,t)}, expand the initial fault set and consider the fault propagation to neighboring devices: Where N(i) is the set of neighbor devices of device i, ξ ij (t) is the impact factor of the failure of device j due to the propagation of the failure of device i; S33, combined with the generated fault set F k ′(t) and load fluctuation, define the power protection scenario: S k (t) = {F k ′(t),L t +ΔL t }, where L t Load distribution at time t, ΔL t is the charge fluctuation in the scenario, ensuring that the scenario includes dynamic adjustment of fault and load conditions; S34. Finally, each scenario is dynamically simulated, and the evolution of the grid operation status is simulated through time series integration: Where P i (t) is the active power of node i at time t, Q i (t) is reactive power, V i (t), V j (t) The voltage amplitude between nodes i and j at time t, G ij , B ij is the real and imaginary part of the admittance between nodes i and j, θ ij (t) is the voltage phase difference between node i and node j, ΔP i (t), ΔQ i (t) is the dynamic power correction term of node i at time t, which dynamically updates the voltage and phase to simulate the evolution of power conservation.

2. According to a method for building an intelligent power conservation scenario based on digital twins according to claim 1, it is characterized in that: In step S1, the equipment operation data includes voltage, current, power, and load; the equipment status includes the equipment working time, temperature, humidity, and vibration; and the grid topology data includes grid topology and switch status data.

3. According to the method for building an intelligent power conservation scenario based on digital twins according to claim 1, it is characterized in that: The specific steps of modeling in step S2 are: S21. First, define the power system graph as G = (V, E), where V is the device node set, E represents the connection relationship between devices, and the characteristic of node i is expressed as The adjacency matrix of the graph is A, and the time step is t; S22. In the multi-scale graph convolution layer, the adjacency matrix A of different scales is used. l To handle the interactions between power devices, for each scale l, the update formula of node i is: Among them, Nl(i) represents the neighbor set of node i in the adjacency relationship of the lth layer, is the dynamic attention coefficient calculated by the graph attention mechanism, W l (l) is the weight matrix of the lth layer, is the bias term, σ is the ReLU function; S23. Combined with time series modeling to capture the dynamic behavior of power equipment and networks, based on the features of graph convolution output, LSTM modeling time dependency is introduced. The time series modeling formula of node i at time step t is: in is the state of node i at the previous time step, is the input feature of node i at the current time step; S24. Finally, the final node feature representation is obtained by combining multi-scale and temporal characteristics:

4. According to the method for building an intelligent power conservation scenario based on digital twins according to claim 3, it is characterized in that: The dynamic attention coefficient calculated by the graph attention mechanism in step S22 The calculation formula is: in is the dynamic attention coefficient between node i and node j at layer l, and a is the learned attention weight vector.

5. According to a method for building an intelligent power conservation scenario based on digital twins according to claim 1, it is characterized in that: The influencing factor of the fault in step S32 Among them, P(f j ,t) is the dynamic failure probability of j at time t, d ij is the distance between devices i and j.

6. According to a method for building an intelligent power conservation scenario based on digital twins according to claim 1, it is characterized in that: The formula for dynamically updating the voltage and phase in step S34 is: in It represents the dynamic change rate of phase and voltage, and α and β are the update step sizes.

Images