CN119762642B - Station line-to-user penetration dynamic rendering method based on energy flow graph - Google Patents

Abstract

The invention belongs to the technical field of power system visualization, and particularly relates to a station line-to-user penetration dynamic rendering method based on an energy flow graph. According to the method, edge computing nodes are arranged at station line-to-customer key points, time sequence data compression is improved, redundant transmission is reduced, and real-time accurate data are provided for dynamic rendering. And then, an improved hierarchical topology modeling algorithm is provided, the predicted power flow is accurately analyzed, and the rendering basis of the flow graph is enabled to be more scientific. And then, generating a local energy flow visual layer according to the user interaction requirement, reducing the calculation pressure, and improving the interaction response speed and the fluency of the energy flow diagram display. And finally, realizing smooth dynamic rendering of the two-dimensional and three-dimensional energy flow diagrams based on WebGPU and loader optimization. The method greatly reduces the calculation pressure, realizes quick interaction, and can smoothly display the dynamic local energy flow diagram.

Description

Station line-to-user penetration dynamic rendering method based on energy flow graph

Technical Field

The invention belongs to the technical field of power system visualization, and particularly relates to a station line-to-user penetration dynamic rendering method based on an energy flow graph.

Background

In order to meet the rapid interaction requirement of users on detailed energy flow information of a local area and reduce the whole-network visual computing pressure, the invention dynamically generates a local energy flow data visual layer according to the user interaction requirement. In the present age, the power system is continuously and rapidly developed, and the complexity of the power grid structure presents an exponential growth situation. The change makes the visual analysis of the running state of the power grid an indispensable important link in the power industry, and the dynamic rendering of the energy flow graph is a core technical requirement.

The existing dynamic rendering method of the energy flow graph exposes a plurality of defects which are difficult to overcome in practical application. In the initial stage of data acquisition, due to the limitation of technical means, real-time data acquisition and processing work cannot be efficiently performed. The data acquisition often has obvious hysteresis, so that the energy flow graph is very slow in the aspect of dynamic update, and the instantaneous change of the real-time state of the power grid cannot be accurately and timely reflected. The power grid topology modeling links are also not fully satisfactory, and an effective strategy matched with the characteristics of the power grid hierarchy structure is lacked. The short board directly causes difficulty in accurately analyzing the power flow, so that the dynamic distribution and flow trend of the power in each part of the power grid cannot be accurately presented in the dynamic rendering process of the power flow graph, and reliable decision basis cannot be provided for operation and maintenance personnel. The problem is also quite serious in the visual computing process. Because the calculation mode is not effectively optimized, when the mass power grid data is faced, the whole-grid visualized calculation pressure such as the Taishan mountain is jacked. When the dynamic display of the two-dimensional and three-dimensional energy flow diagrams is carried out, the phenomena of blocking and frame dropping frequently occur, and the phenomenon is like continuous blocking of the pictures during playing. The dynamic sensing of the power grid energy flow information by the user is seriously disturbed, so that the power grid energy flow information cannot be clearly and continuously mastered, and the capability of the user for carrying out deep analysis and accurate judgment on the power grid running state is greatly hindered.

Disclosure of Invention

Aiming at the technical problems, the invention provides a station line-to-user penetration dynamic rendering method based on an energy flow graph.

In order to achieve the above purpose, the technical scheme adopted by the invention comprises the following steps:

s1, firstly, arranging edge computing nodes at key points of station line households to realize real-time acquisition and distributed processing of data, wherein the distributed processing improves time sequence data compression so as to reduce redundant data transmission;

s2, providing an improved hierarchical topology modeling algorithm, decomposing the power grid topology according to a hierarchy, introducing time sequence-based energy flow prediction, and supporting the combination of real-time data and prediction data;

S3, dynamically generating a visual layer of the local energy flow data according to the user interaction requirement, and reducing the whole-network visual computing pressure, wherein the method is specifically implemented as follows:

S31, firstly, calculating a minimum subgraph G _sub of a user designated area based on layered topology G= (N, epsilon, W) of a power grid, concentrating whole network calculation to a local area of interest of a user, and dynamically adjusting according to real-time interaction;

S32, dynamically generating a local energy flow visualization layer according to the extracted subgraph G _sub, and calculating energy flow intensity S _ij for each edge e _ij∈ε_sub,ε_sub which is an edge set of the subgraph G _sub: wherein P _ij is the power flow of nodes i to j, W _ij is the power transmission capability of edge e _ij, and the energy flow intensity is used for dynamically adjusting the thickness and color of the edge in the layer, and for each node i E N _sub,N_sub the node set of sub-graph G _sub is obtained by The method is characterized by comprising the following steps of aggregating formulas, wherein S _i represents node load ratio, and the colors and the sizes of icons mapped to the nodes, drawing only key edges and the nodes based on the results of S _ij and S _i, and reducing rendering complexity;

S33, finally, through an improved intelligent region sensing algorithm, the fact that region sensing weight omega _i is introduced under real-time interaction, only data in a region of interest N _sub are processed, when user interaction changes, only newly added and reduced node sets delta N _sub are updated for energy flow calculation and caching of generated local layer data is carried out, and a caching result is directly called when a user repeatedly inquires, so that the cost of multiple times of calculation is reduced;

and S4, finally, dynamically rendering based on WebGPU and loader optimization to realize smooth display of the two-dimensional and three-dimensional energy flow diagrams.

Preferably, the step S1 of implementing the distributed processing to improve time series data compression is to fit time series data { (x ₁,y₁),(x₂,y₂),...(x_n,y_n) } to a straight line y=αx+β using a least square method, whereinThe fitting error calculation formula is: Where x _i is the time point, y _i is the data value corresponding to the time point, n is the number of samples of the current segment, and when the error is less than the error tolerance threshold, the segment data is compressed into a starting point, an end point and a fitting parameter.

Preferably, the step S2 of implementing the hierarchical topology modeling algorithm includes:

S21, dividing the power grid topology into a station layer, a line layer, a variable layer and a household layer according to a hierarchy to form a multi-layer topology structure based on a graph, wherein a node set is N= { N ₁,N₂,N₃,N₄ }, N _Station ,N _{Wire (C)} ,N _Variable ,N _Household is a transformer station node, a power transmission station node, a transformer node and a user node respectively, ε= { e _ij|i∈N_l,j∈N_l+1, l=1, 2 and 3} wherein e _ij represents the connection relation between a node i in the hierarchy l and a node j in the hierarchy l+1, and a weighted directed graph G= (N, ε and W) is adopted, wherein a weight matrix W represents the power transmission capacity of the line;

s22, in the hierarchical topology structure, the power flow follows the physical rule of the power grid, and each node meets a power balance equation, P _in,i＝P_out,i+P_loss,i, wherein P _in,i is the input power of a node i, P _out,i is the output power of the node i, P _loss,i is the power loss, and the power loss can be further defined as: wherein P _ij is the power of nodes i to j, R _ij is the resistance of the line, V _i is the voltage of node i, the power transfer relation of each layer of nodes is calculated, and the whole network power flow state is recursively updated;

s23, predicting the power of the node i by adopting LSTM, optimizing the prediction result by utilizing interlayer coupling constraint, enabling the prediction results of the upper layer and the lower layer to be consistent, and enabling an objective function to be: Wherein the method comprises the steps of Is the future power prediction value for node i, i is the number of layers,Is the total predicted input power of all relevant nodes of the next layer;

S24, aiming at the problem that the real-time data and the predicted data have differences, a dynamic data fusion method is adopted, and the fusion power value is as follows: where P _real,i,t+1 is the real-time data, Phi _real,i,φ_pred,i is the credibility of the real-time data and the predicted data respectively.

Preferably, the constraint condition of the objective function in the step S23 is thatEnsuring that the input power of each node is equal to the sum of the output power and the loss power,The output power of each node cannot exceed its design capacity P _capacity,i.

Compared with the prior art, the method has the advantages that by means of edge computing nodes and optimizing time sequence data compression, a more timely and simplified data base is provided for dynamic rendering, rendering content is ensured to reflect the latest state of a power grid, and data transmission is efficient. The improved hierarchical topology modeling algorithm is integrated into energy flow prediction, so that the rendering basis is more scientific and accurate, and the power flow trend can be better presented. When the visual image layer is generated, a user region of interest is focused, data is calculated in an optimized mode and cached, calculation pressure is greatly reduced, quick interaction is achieved, and a dynamic-change local energy flow diagram can be displayed smoothly.

Detailed Description

In order that the above objects, features and advantages of the application may be more clearly understood, a further description of the application will be provided with reference to the following examples. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced otherwise than as described herein, and therefore the present invention is not limited to the specific embodiments of the disclosure that follow.

In the embodiment, with the development of a power system, a power grid structure is increasingly complex, and the requirement for visual analysis of the running state of the power grid is continuously increased, and especially, the dynamic rendering of an energy flow graph becomes a key technical requirement. The traditional dynamic rendering method of the energy flow graph has a plurality of defects. In the data acquisition link, efficient real-time data acquisition and processing cannot be realized, so that the dynamic update of the energy flow graph is slow, and the real-time state change of the power grid cannot be reflected in time. The power grid topology modeling lacks suitability, and is difficult to accurately analyze the power flow, so that the dynamic distribution and flow trend of the power cannot be accurately presented when the power flow diagram is dynamically rendered. In the visual computing process, as the computing mode is not effectively optimized, the whole network visual computing pressure is huge, and when two-dimensional and three-dimensional energy flow diagrams are dynamically displayed, the phenomena of blocking and frame dropping frequently occur, so that the dynamic perception, analysis and judgment of a user on the energy flow information of the power grid are seriously influenced.

In order to realize efficient acquisition and processing of station line-to-customer data in an electric power system and reduce data transmission redundancy, the invention adopts a scheme that edge computing nodes are arranged at station line-to-customer key points and time sequence data compression is improved. In traditional grid data processing, data acquisition efficiency is low and transmission redundancy is serious. According to the invention, edge computing nodes are arranged at key points of the station line change, so that real-time acquisition and distributed processing of data are realized. Fitting the time series data { (x ₁,y₁),(x₂,y₂),...(x_n,y_n) } to a straight line y=αx+β using the least squares method, whereinThe fitting error calculation formula is: Where x _i is the time point, y _i is the data value corresponding to the time point, n is the number of samples of the current segment, and when the error is less than the error tolerance threshold, the segment data is compressed into a starting point, an end point and a fitting parameter. By the mode, a large amount of similar or slowly-changing data can be effectively compressed, the data quantity required to be transmitted is reduced, and the real-time performance of data acquisition is improved.

Then, considering that the existing power grid topology modeling lacks an effective layering strategy, the power flow is difficult to accurately analyze, and in order to accurately construct a power grid topology model and accurately analyze and predict the power flow of the power grid, the invention provides an improved layering topology modeling algorithm. Dividing the power grid topology into stations, lines, transformers and four layers according to a hierarchy to form a multi-layer topology structure based on a graph, wherein a node set is N= { N ₁,N₂,N₃,N₄ }, N _Station ,N _{Wire (C)} ,N _Variable ,N _Household is a transformer station node, a power transmission station node, a transformer node and a user node respectively, ε= { e _ij|i∈N_l,j∈N_l+1, l=1, 2 and 3} wherein e _ij represents a connection relation between a node i in the hierarchy l and a node j in the hierarchy l+1, a weighted directed graph G= (N, ε and W) is adopted, wherein a weight matrix W represents the power transmission capacity of the line, in the hierarchical topology structure, a power flow follows the physical rule of the power grid, each node meets a power balance equation, P _in,i＝P_out,i+P_loss,i, wherein P _in,i is the input power of the node i, P _out,i is the output power of the node i, and P _loss,i is the power loss, and the power loss can be further defined as: wherein P _ij is the power of the nodes i to j, R _ij is the resistance of a line, V _i is the voltage of the node i, the power transfer relation of each layer of nodes is calculated, the state of the power flow of the whole network is recursively updated, then LSTM is adopted to predict the power of the node i, and the prediction result is optimized by utilizing interlayer coupling constraint, so that the prediction results of the upper layer and the lower layer are consistent, and the objective function is as follows: Wherein the method comprises the steps of Is the future power prediction value for node i, i is the number of layers,Is the total predicted input power of all relevant nodes of the next layer, and the constraint condition of the objective function is thatEnsuring that the input power of each node is equal to the sum of the output power and the loss power,Aiming at the problem that the difference exists between real-time data and predicted data, a dynamic data fusion method is adopted, and the fusion power value is as follows: where P _real,i,t+1 is the real-time data, Phi _real,i,φ_pred,i is the credibility of the real-time data and the predicted data respectively. Through the hierarchical topology modeling algorithm, when power flow analysis and prediction are carried out on an urban power grid, power transfer and change trend among different layers can be accurately simulated, and potential problems can be found in advance

And then, in order to meet the rapid interaction requirement of the user on the detailed energy flow information of the local area and reduce the whole-network visual computing pressure, the invention dynamically generates the local energy flow data visual layer according to the user interaction requirement. Firstly, based on the hierarchical topology G= (N, epsilon, W) of a power grid, calculating a minimum sub-graph G _sub of a user-specified area, concentrating the whole-network calculation to a local area of interest of a user, dynamically adjusting according to real-time interaction, then dynamically generating a local energy flow visual layer according to the extracted sub-graph G _sub, and calculating energy flow intensity S _ij of each side e _ij∈ε_sub,ε_sub for a side set of the sub-graph G _sub: wherein P _ij is the power flow of nodes i to j, W _ij is the power transmission capability of edge e _ij, and the energy flow intensity is used for dynamically adjusting the thickness and color of the edge in the layer, and for each node i E N _sub,N_sub the node set of sub-graph G _sub is obtained by The method comprises the steps of aggregating formulas, wherein S _i represents node load proportion, mapping to colors of nodes and sizes of icons, drawing key edges and nodes based on results of S _ij and S _i, reducing rendering complexity, finally, introducing region perception weight omega _i under real-time interaction, processing data in a region of interest N _sub only, updating energy flow calculation only for a newly added and reduced node set delta N _sub when user interaction changes, caching generated local layer data, and directly calling a caching result when a user repeatedly inquires, thereby reducing the cost of multiple times of calculation. By the method, a user can quickly acquire detailed energy flow information of a local area, the interactive response speed is remarkably improved, and meanwhile, the two-dimensional and three-dimensional energy flow diagrams are displayed more smoothly.

Finally, based on WebGPU and loader optimized dynamic rendering, webGPU fully utilizes the parallel processing capability of modern graphics hardware to accelerate visual rendering of the energy flow data. The loader optimizes vertex shaders, fragment shaders, etc. in the graphics rendering pipeline, and adjusts the computing mode of color and transparency according to the energy flow intensity, thereby reducing unnecessary computing overhead. When the two-dimensional and three-dimensional energy flow diagrams of the large-scale power grid are displayed, the phenomenon of blocking and delay can be effectively avoided, and a smooth dynamic rendering effect is realized.

The present invention is not limited to the above-mentioned embodiments, and any equivalent embodiments which can be changed or modified by the technical content disclosed above can be applied to other fields, but any simple modification, equivalent changes and modification made to the above-mentioned embodiments according to the technical substance of the present invention without departing from the technical content of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims (3)

1. A dynamic rendering method for station-line-customer penetration based on energy flow diagrams, characterized by comprising the following steps: S1. First, edge computing nodes are deployed at key points of the station line and substation to realize real-time data acquisition and distributed processing. The distributed processing improves time-series data compression to reduce redundant data transmission. S2. An improved hierarchical topology modeling algorithm is proposed, which decomposes the power grid topology into hierarchical levels and introduces time-series-based energy flow prediction to support the combination of real-time data and prediction data. S3. Then, based on user interaction needs, dynamically generate a visualization layer of local energy flow data to reduce the computational burden of visualization across the entire network. Specifically, this is achieved as follows: S31. First, based on the weighted directed graph G＝(N,ε,W) of the power grid, calculate the minimum subgraph _Gsub of the user-specified region, concentrate the calculation of the entire network to the local region of interest to the user, and dynamically adjust it according to real-time interaction; S32. Next, based on the extracted subgraph G _{_sub}, a local energy flow visualization layer is dynamically generated, where each edge e_{_ij} ∈ _{ε_sub} , and ε_{_sub} is the set of edges of the subgraph G _{_sub} , and its energy flow intensity S _{_ij} is calculated: Where P_{_ij} represents the power flow from node i to j, W _{_ij} represents the power transmission capacity of edge e_{_ij}, and the power flow intensity is used to dynamically adjust the thickness and color of edges in the layer; for each node i∈N_{_sub} , N _{_sub} is the set of nodes in the subgraph G_{_sub}, through... The formula is used for aggregation, where _Si represents the node load percentage, which is mapped to the node's color and icon size; based on the results of _Sij and _Si , only key edges and nodes are drawn, reducing rendering complexity; S33. Finally, through the improved intelligent region perception algorithm, it is ensured that under real-time interaction, the region perception weight _ωi is introduced, and only the data within the region of interest N _sub is processed. When the user interaction changes, the energy flow calculation is updated only for the newly added and reduced node set ΔN _sub and the generated local layer data is cached. When the user queries repeatedly, the cached result is directly called to reduce the overhead of multiple calculations. S4. Finally, dynamic rendering based on WebGPU and Shader optimization enables smooth display of two-dimensional and three-dimensional energy flow graphs; The implementation steps of the hierarchical topology modeling algorithm in step S2 are as follows: S21. Divide the power grid topology into four layers: substation, line, transformer, and user, forming a multi-layer graph-based topology structure. The node set is N = { _N1 , _N2 , _N3 , _N4 }, where _N1 , _N2 , _N3 , and _N4 are substation nodes, transmission station nodes, transformer nodes, and user nodes, respectively. ε = {e_{_ij} | i ∈ _{N_l}, j ∈ N_{_l+1}, l = 1, 2, 3}, where e _{_ij} represents the connection relationship between node i in layer l and node j in layer l+1. A weighted directed graph G = (N, ε, W) is used, where the weight matrix W represents the power transmission capacity of the line. S22. In a hierarchical topology, power flow follows the physical laws of the power grid, and each node satisfies the power balance equation: P_{_in,i} = P_{_out,i} + P_{_loss,i} , where P _{_in,i} is the input power of node i, P_{_out,i} is the output power of node i, and P_{_loss,i} is the power loss, defined as: Where _Pij is the power from node i to j, _Rij is the line resistance, and _Vi is the voltage at node i. Calculate the power transfer relationship of each node and recursively update the power flow state of the entire network. S23. Next, LSTM is used to predict the power of node i. The prediction results are optimized using inter-layer coupling constraints to ensure consistency between the prediction results of upper and lower layers. The objective function is: in Here, l is the predicted future power value of node i, and l is the layer number. It is the total predicted input power of all relevant nodes in the next layer; S24. To address the discrepancy between real-time and predicted data, a dynamic data fusion method is adopted, with a fusion power value of: Where P _{_real,i,t+1} represents real-time data. For predictive data; The credibility of real-time data and predicted data are respectively. 2. The station-line-customer penetration dynamic rendering method based on energy flow diagram according to claim 1, characterized in that, in step S1, the distributed processing improved time-series data compression implementation step is to fit the time-series data {( _x1 , _y1 ), ( _x2 , _y2 ), ... ( _xn , _yn )} to a straight line y = αx + β using the least squares method, where The formula for calculating the fitting error is: Where x _{_i} is the time point, y _{_i} is the data value corresponding to the time point, and n is the number of samples in the current segment. When the error is less than the error tolerance threshold, the data in this segment is compressed into the start point, the end point, and the fitting parameters. 3. The station-line-customer penetration dynamic rendering method based on energy flow diagram according to claim 1, characterized in that the constraint condition of the objective function in step S23 is: Ensure that the input power of each node equals the sum of the output power and the power loss, P _out,i ≤ P _capacity,i The output power of each node cannot exceed its design capacity P_{_capacity,i} .