A method and system for checking the construction quality of a provincial side power consumption information collection system

Abstract

The application discloses a method and system for checking construction quality of a power side power consumption information collection system, and belongs to the technical field of intelligent power information processing. The method comprises the following steps: generating a graph neural network data sample set of a target power consumption information collection system; generating a graph neural network model according to the graph neural network data sample set, and generating a mirror model of the graph neural network model in a simulation environment of the target power consumption collection system; constructing a twin network checking and evaluating model according to the graph neural network model and the mirror model; determining quality evaluation index values of the target power consumption information collection system according to the twin network checking and evaluating model, performing weighted fusion on the quality evaluation index values to obtain weighted quality evaluation index values, and determining the quality of the target power consumption collection system according to the weighted quality evaluation index values. The application can effectively evaluate the business execution quality of the power consumption information collection system.

Description

Technical Field

[0001] This invention relates to the field of intelligent power information processing technology, and more specifically, to a method and system for quality verification of the construction of a provincial-level power consumption information collection system. Background Technology

[0002] In recent years, facing the new situation and new requirements such as the accelerated integration of the energy revolution and the digital revolution, the in-depth advancement of power market reform, and the increasingly tight power supply and demand situation, the "one body, four wings" development layout has been proposed to actively build a new power system with new energy as the main body. This not only increases the demand for high-frequency and high-quality data from the electricity information collection system, but also places higher demands on the reliability of system control command execution, the real-time performance of load monitoring, the accuracy of user power outage analysis, and the ability to clear and settle electricity consumption and fees. The traditional electricity information collection systems built independently in each province can no longer meet the needs of power dispatching and other business development. Summary of the Invention

[0003] To address the above problems, this invention proposes a method for quality verification of the construction of a provincial-level electricity consumption information collection system, comprising:

[0004] Acquire historical operating data of the target electricity consumption information collection system, and generate a graph neural network data sample set of the target electricity consumption information collection system based on the historical operating data;

[0005] Based on the graph neural network data sample set, a graph neural network model and a mirror model of the graph neural network model are generated. Based on the graph neural network model and the mirror model, a twin network verification and evaluation model is constructed.

[0006] Based on the twin network verification and evaluation model, the quality evaluation index values ​​of the target electricity consumption information collection system are determined. The quality evaluation index values ​​are then weighted and fused to obtain a weighted quality evaluation index value. Based on the weighted quality evaluation index value, the quality of the target electricity consumption collection system is determined.

[0007] Optional, historical operational data, including: time-series monitoring data of network nodes in the electricity consumption information collection system graph for different time periods.

[0008] Optionally, generating a graph neural network model based on the graph neural network data sample set includes:

[0009] Transform the time-series monitoring data in the graph neural network data sample set into graph data format;

[0010] Based on the interconnections of the data flow process in the electricity consumption information collection system, a node adjacency matrix is ​​generated. Based on the adjacency matrix, a node topology connection diagram of the target electricity consumption information collection system is generated, where each node represents a functional module of the target electricity consumption information collection system.

[0011] The graph neural network data sample set, which has been transformed into graph data form, is input to each node in the node topology connection graph according to the correspondence between nodes to obtain the input data of the model.

[0012] The input data is fed into a graph neural network for training to generate a graph neural network model.

[0013] Optionally, constructing a twin network verification and evaluation model includes: determining the similarity value output by the graph neural network model and the mirror model, using the similarity value as the accuracy compensation value of the mirror model, adjusting the parameters of the mirror model, generating a corrected mirror model, and constructing a twin network verification and evaluation model using the corrected mirror model.

[0014] Optionally, the quality assessment index values ​​of the target electricity consumption information collection system are determined, including:

[0015] Determine the timeliness and quality of business execution at each node of the target electricity consumption data acquisition system under different business execution scenarios, and determine the business execution completion indicator value based on the timeliness and quality of business execution.

[0016] Based on the business execution completion indicators, determine the values ​​of each quality assessment indicator.

[0017] Furthermore, this invention also proposes a quality verification system for the construction of a provincial-level electricity consumption information collection system, comprising:

[0018] The data acquisition unit is used to acquire historical operating data of the target electricity consumption information acquisition system, and generate a graph neural network data sample set of the target electricity consumption information acquisition system based on the historical operating data.

[0019] The model generation unit is used to generate a graph neural network model based on the graph neural network data sample set, and generate a mirror model of the graph neural network model in the simulation environment of the target power consumption acquisition system. Based on the graph neural network model and the mirror model, a twin network verification and evaluation model is constructed.

[0020] The evaluation unit is used to determine the quality evaluation index values ​​of the target electricity consumption information collection system according to the twin network verification and evaluation model, perform weighted fusion of the quality evaluation index values ​​to obtain weighted quality evaluation index values, and determine the quality of the target electricity consumption collection system based on the weighted quality evaluation index values.

[0021] Optionally, the historical operating data of the data acquisition unit includes: time-series monitoring data of the network nodes of the electricity information acquisition system graph for different time periods.

[0022] Optionally, the model generation unit generates a graph neural network model based on the graph neural network data sample set, including:

[0023] Transform the time-series monitoring data in the graph neural network data sample set into graph data format;

[0024] Based on the interconnections of the data flow process in the electricity consumption information collection system, a node adjacency matrix is ​​generated. Based on the adjacency matrix, a node topology connection diagram of the target electricity consumption information collection system is generated, where each node represents a functional module of the target electricity consumption information collection system.

[0025] The graph neural network data sample set, which has been transformed into graph data form, is input to each node in the node topology connection graph according to the correspondence between nodes to obtain the input data of the model.

[0026] The input data is fed into a graph neural network for training to generate a graph neural network model.

[0027] Optionally, the evaluation unit is also used to construct a twin network verification evaluation model, including: determining the similarity value output by the graph neural network model and the mirror model, using the similarity value as the accuracy compensation value of the mirror model, adjusting the parameters of the mirror model, generating a corrected mirror model, and constructing the twin network verification evaluation model using the corrected mirror model.

[0028] Optionally, the evaluation unit determines the quality evaluation index values ​​of the target electricity consumption information collection system, including:

[0029] Determine the timeliness and quality of business execution at each node of the target electricity consumption data acquisition system under different business execution scenarios, and determine the business execution completion indicator value based on the timeliness and quality of business execution.

[0030] Based on the business execution completion indicators, determine the values ​​of each quality assessment indicator.

[0031] In another aspect, the present invention also provides a computing device, comprising: one or more processors;

[0032] A processor is used to execute one or more programs;

[0033] When the one or more programs are executed by the one or more processors, the method described above is implemented.

[0034] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the method described above.

[0035] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0036] This invention provides a method for quality verification of a provincial-level electricity consumption information collection system, comprising: acquiring historical operational data of the target electricity consumption information collection system, and generating a graph neural network data sample set for the target electricity consumption information collection system based on the historical operational data; generating a graph neural network model based on the graph neural network data sample set, and generating a mirror model of the graph neural network model in a simulation environment of the target electricity consumption information collection system; constructing a twin network verification and evaluation model based on the graph neural network model and the mirror model; determining various quality evaluation index values ​​of the target electricity consumption information collection system based on the twin network verification and evaluation model; weighting and fusing the various quality evaluation index values ​​to obtain a weighted quality evaluation index value; and determining the quality of the target electricity consumption information collection system based on the weighted quality evaluation index value. This invention can effectively evaluate the business execution quality of an electricity consumption information collection system. Attached Figure Description

[0037] Figure 1 This is a flowchart of the method of the present invention;

[0038] Figure 2 This is a flowchart of an embodiment of the present invention;

[0039] Figure 3 This is a diagram of the graph model training structure according to an embodiment of the present invention;

[0040] Figure 4 This is a schematic diagram illustrating the working principle of the attention mechanism in an embodiment of the present invention.

[0041] Figure 5 This is a flowchart illustrating the similarity between the estimated probability value of the construction quality level of each node and the expected probability value of the construction quality level of each node in the twin network of this invention.

[0042] Figure 6 Structure diagram of the comprehensive weighted quality assessment model of the provincial-level next-generation electricity consumption information collection system of the twin network in this invention;

[0043] Figure 7 This is a diagram of the weighted parameters of the fully connected layer in the comprehensive weighted quality assessment of the provincial-level next-generation electricity consumption information collection system using a twin network, as described in this embodiment of the invention.

[0044] Figure 8 This is a structural diagram of the system of the present invention. Detailed Implementation

[0045] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0046] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0047] Example 1:

[0048] This invention proposes a method for quality verification of the construction of a provincial-level electricity consumption information collection system, such as... Figure 1 As shown, it includes:

[0049] Step 1: Obtain historical operating data of the target electricity consumption information collection system, and generate a graph neural network data sample set of the target electricity consumption information collection system based on the historical operating data;

[0050] Step 2: Based on the graph neural network data sample set, generate a graph neural network model and a mirror model of the graph neural network model; based on the graph neural network model and the mirror model, construct a twin network verification and evaluation model.

[0051] Step 3: Based on the twin network verification and evaluation model, determine the quality evaluation index values ​​of the target electricity consumption information collection system, perform weighted fusion of the quality evaluation index values ​​to obtain the weighted quality evaluation index values, and determine the quality of the target electricity consumption collection system based on the weighted quality evaluation index values.

[0052] Historical operational data includes time-series monitoring data of network nodes in the electricity consumption information collection system at different time periods.

[0053] The generation of a graph neural network model based on the graph neural network data sample set includes:

[0054] Transform the time-series monitoring data in the graph neural network data sample set into graph data format;

[0055] Based on the interconnections of the data flow process of the electricity consumption information collection system, a node adjacency matrix is ​​generated. Based on the adjacency matrix, a node topology connection diagram of the target electricity consumption information collection system is generated, where each node is a functional module of the target electricity consumption information collection system.

[0056] The graph neural network data sample set, which has been transformed into graph data form, is input to each node in the node topology connection graph according to the correspondence between nodes to obtain the input data of the model.

[0057] The input data is fed into a graph neural network for training to generate a graph neural network model.

[0058] The process of constructing a twin network verification and evaluation model includes: determining the similarity value between the outputs of the graph neural network model and the mirror model; using the similarity value as the accuracy compensation value for the mirror model; adjusting the parameters of the mirror model to generate a corrected mirror model; and constructing the twin network verification and evaluation model using the corrected mirror model.

[0059] The determination of various quality assessment index values ​​for the target electricity consumption information collection system includes:

[0060] Determine the timeliness and quality of business execution at each node of the target electricity consumption data acquisition system under different business execution scenarios, and determine the business execution completion indicator value based on the timeliness and quality of business execution.

[0061] Based on the business execution completion indicators, determine the values ​​of each quality assessment indicator.

[0062] The following explanation uses the provincial-level new-generation electricity consumption information collection system as an example, detailing the specific implementation steps, including:

[0063] The specific process is as follows: Figure 2 As shown, it includes:

[0064] A. Based on the new generation of electricity consumption information collection system built at the provincial level, collect historical operation data from the new generation of electricity consumption information collection system at the provincial level to form a graph neural network data sample set;

[0065] Among them, the historical operation data is time-series monitoring data obtained from the graph network node monitoring data in different time periods;

[0066] B. Transform the time-series monitoring data in the graph neural network data sample set into graph data format to form a graph neural network sample dataset;

[0067] The functional modules and terminal devices in the system are abstracted as nodes, and a feature matrix of each node is constructed. The data flow paths within the system are abstracted as edges, and an adjacency matrix is ​​constructed. A graph neural network sample dataset is generated, where n is the number of nodes, f is the dimension of the node monitoring metric, and m is the number of edges in the node. The node feature matrix is ​​constructed as follows:

[0068] For dynamic data, suppose the node data sequence feature In has n time steps Pr1, Pr2, ..., Pr n , forming node features V G Then V G The evolution of data sequence features starting from In can be represented as follows:

[0069] V G =(I n ,P r (1)

[0070] Where Pr is the set of time steps, Pr = {Pr1, Pr2, ..., Pr...} n}

[0071] A single time step Pr in the evolution of data sequence features i (1≤i≤n) includes input In i various types of sequence op i Sequence logic processing f i and output VG i The four aspects are:

[0072] P ri =(I ni ,op i ,f i V Gi (2)

[0073] In the formula, In i =((In1, In2, ..., In) iu ), TB i TE i ) represents the input, In1, In2, ..., In iu Indicates input In i u (u≥1) data sequence attributes, TB i and TE i These represent the start and end times of the input, respectively; op i =(OP) i ,TB' i ,TE' i ) represents various types of sequences, OP i Indicates the name of each type of sequence, TB' i and TE' i f represents the start and end times of each type of sequence, respectively. i In i →VGi This represents a sequence logic processing function from input to output, that is, a concatenation operation of various types of sequences according to time steps, V Gi =((V) G1 V G2 , ..., V Gis ), TB i+1 TE i+1 ) represents the output, V G1 V G2 , ..., V Gis This represents the output of s (s≥1) data sequence attributes, TB i+1 and TE i+1 This indicates the start and end times of the output. Through the above process, the sequence features V of the node are ultimately obtained. G For static data, its content is directly appended to the end of the dynamic data sequence to represent complete node information. Adjacency matrix A G It is formed according to the topology of the data flow link of the electricity information collection system.

[0074] The node features V obtained in the above steps G Normalization is performed using the Z-score method:

[0075]

[0076] In equation (3), For node features V G The mean of all values ​​along a certain dimension For node features V G The variance of all values ​​in a certain dimension; combined with the adjacency matrix of the link topology, forming the graph data required for the data flow link of the electricity information collection system.

[0077] C. Based on the functional sub-modules of the provincial-level new generation electricity consumption information collection system as nodes, they are interconnected according to the data flow process of the provincial-level new generation electricity consumption information collection system. The generated node adjacency matrix represents the correlation between each neighboring node, forming a node topology connection diagram.

[0078] D. Input the graph data of each node in the graph neural network sample dataset into the nodes of the node topology connection graph. The output data of the input node layer enters the graph attention layer, the fully connected layer and the softmax layer in sequence. The probability value of the construction quality level of each node is output. After training the feedback adjustment of the graph attention neural network parameters based on the business execution index data of the graph neural network sample dataset, the parameters are fixed as the graph neural network model for evaluating the construction quality of the provincial side's new generation electricity information collection system.

[0079] The graph model training structure of the electricity information acquisition system based on graph attention neural network is as follows: Figure 3 As shown, graph data is used as input, and the features of each node in the graph data are input into the graph attention layer. Each node aggregates the features of its neighboring nodes based on the adjacency matrix generated by the data flow link topology of the electricity information collection system. The spatial distribution features of various data indicators of the system are extracted by updating the node expression. The intermediate result of fusing the spatial distribution features of the electricity information collection system is then expanded into a one-dimensional vector. The graph neural network model is trained through a fully connected layer and a Softmax function.

[0080] D1. Based on the business execution data of the graph neural network sample dataset, define the business execution completion indicators, and based on the business execution completion indicators, define the construction quality level of each node;

[0081] The business execution completion indicators include response speed level and business execution completion level; the construction quality level of each node is 1-5, and the construction quality level is divided according to the business execution completion indicator level. The maximum and minimum values ​​are normalized and expressed as a percentage.

[0082] D2. Input the transformed graph neural network sample data from step B above into the nodes of the node topology connection graph, and input the output data of each node in the node topology connection graph into the graph attention layer;

[0083] The node feature vectors are continuously input into two graph attention layers. In the graph attention layers, a linear transformation is learned to transform the input features into higher-level feature vectors. This process first uses a common parameter moment W to perform a linear transformation on the features of each node, and then calculates the attention coefficients between the target node and each of its neighboring nodes through an attention mechanism, as shown in formula (4):

[0084]

[0085] In the formula, LeakReLU(·) is the activation function, which can increase the nonlinearity of the model, as shown in formula (5). H is the node feature vector, and FFN is the feedforward neural network. For the vector concatenation operation, β' ij This indicates the importance of the node characteristics of neighbor node j to target node i;

[0086]

[0087] The parameter α is set to 0.2.

[0088] D3. The graph attention layer uses a multi-head attention mechanism to update and adjust the association weights between each neighbor node. The weights represent the influence of the neighbor node's feature data on the target node's feature data. The output is the hidden value of each target node after the influence association between the target node and each neighbor node.

[0089] To integrate the multi-directional features of neighboring nodes into the target node, a multi-head attention mechanism is used to calculate the weights of each neighboring node. First, K different W-parameter matrices are initialized, and the weights of neighboring nodes from multiple angles are calculated. The attention mechanism operates independently for each angle. The multi-head attention mechanism is illustrated in the figure below. Figure 4 As shown. To compare the importance of different nodes using attention weights, the attention coefficients of all neighboring nodes of the target node are normalized using the Softmax function, and the attention weights of each neighboring node are calculated, as shown in equation (6):

[0090]

[0091] Where N i W represents the neighboring nodes of the target node. K This represents the Kth parameter matrix in the multiple parameter matrices initialized for a node.

[0092] The multi-head attention mechanism concatenates the updated node features obtained from each attention mechanism, and then uses an averaging operation to obtain the final target updated node features, as shown in Equation (7):

[0093]

[0094] Where σ(·) represents the nonlinear activation function eLU, its expression in the graph attention neural network is:

[0095]

[0096] In the formula, the parameter θ takes the value of 1 in the graph attention neural network;

[0097] D4. The fully connected layer is expanded in one-dimensional space, and the dimensionality of the feature vector is reduced. After passing through the softmax layer, the probability value of the construction quality level of each target node can be obtained.

[0098] After obtaining the feature vector of the new target node, expand the feature vector of the target node in one-dimensional space, input it into a fully connected layer for feature dimensionality reduction, and feed the dimensionality-reduced vector into the softmax layer to obtain the probability value P of the construction quality level of the node, as shown in formula (9):

[0099]

[0100] Among them W L Let b be the parameter matrix of the fully connected layer. L This is the bias matrix of the fully connected layer.

[0101] D5. Calculate the cross-entropy loss value based on the probability value of the construction quality level of each target node and the actual defined probability value of the construction quality level of each target node. Adjust the feedback graph by paying attention to the parameters of each layer of the neural network, set a threshold for the cross-entropy loss value, and fix the parameters when the cross-entropy loss value is lower than the set threshold. Output a graph neural network model to evaluate the construction quality of the provincial side's new generation of electricity information collection system.

[0102] E. Deploy a graph neural network model based on the provincial-level new-generation electricity consumption information collection system on the headquarters side as a direct deployment model. At the same time, deploy the graph neural network model built on the provincial-level new-generation electricity consumption information collection system as a mirror model in the simulation test environment on the headquarters side. Based on the direct deployment model and the mirror model, construct a twin network verification and evaluation model and output the similarity of the construction quality level probability values ​​of each node.

[0103] Among them, the direct deployment model inputs data from each terminal on the provincial side and time-series monitoring data from each node, while the mirror model inputs simulated terminal data at the level of data collected by the provincial system terminals.

[0104] E1. Deploy a graph neural network model based on the provincial-level new-generation electricity information collection system at the headquarters level as a direct deployment model;

[0105] The direct deployment model acquires real-time time-series monitoring data of each terminal and node in the provincial system.

[0106] E2. Deploy the graph neural network model constructed by the provincial-level new generation electricity consumption information collection system in the headquarters-level simulation test environment as a mirror model;

[0107] The mirror model inputs simulated terminal data on the scale of the provincial-level system terminal data collection.

[0108] E3. Input the data from each terminal on the provincial side and the time-series monitoring data of each node into the direct deployment model, and output the estimated probability value of the construction quality level of each node;

[0109] Based on the estimated probability values ​​of the construction quality level output by each node, the system construction quality level assessment can be achieved.

[0110] E4. Simulate the same amount of terminal data based on the data volume collected by the provincial terminal, input the mirror model, and output the expected probability value of the construction quality level of each node;

[0111] E5. Construct a twin network verification and evaluation model based on the direct deployment model and the mirror model, calculate the similarity between the estimated probability value of the construction quality level of each node and the expected probability value of the construction quality level of each node, and use it as the similarity evaluation value between the construction quality of each node on the provincial side and the construction quality of each node on the headquarters.

[0112] The graph node data sequences from the headquarters side and the provincial side are fed into two identical graph attention neural networks, with the two feature extraction networks having the same structure and parameters. The input data sequences are I... p ={v1,v2,...,v p} and I q ={v1,v2,...,v q}, where I p For the input data sequence on the headquarters side, I q This is the input data sequence from the provincial side. The two data sequences may differ in time length, monitoring data, or human-induced differences. Two identical deep neural networks output a fixed-size vector O. P ∈R M and O q ∈R M The cosine similarity is used to calculate the similarity between two output vectors, as shown in formula (10):

[0113]

[0114] F. Based on the comprehensive weighting of multi-scenario business, a comprehensive evaluation of the construction quality of the provincial-level new-generation electricity consumption information collection system is conducted. That is, under different scenario business execution, the timeliness and quality of business execution at each node are evaluated. Based on the business execution completion indicators, the various indicators of the system are evaluated. The weighting coefficients calculated by twin network are used to weight and integrate the above evaluation indicators to generate a comprehensive weighted quality evaluation index for the provincial-level new-generation electricity consumption information collection system.

[0115] Figure 5 The weight parameters of the fully connected layer are shown, and the structure of the comprehensive weighted quality assessment model of the provincial-level new-generation electricity information collection system is as follows: Figure 6 As shown, the network uses headquarters-side data sequence I' p and provincial side data sequence I' q As input, where I' p For the standardized data sequence on the headquarters side, I' q This is the actual input data sequence from the provincial side. The standardized data sequence is one that allows the system to achieve standard metrics, including data transmission rate, data transmission capacity, and data transmission failure rate at each node. A Siamese neural network (MGAN) is used to extract sequence features, outputting two fixed-size vectors O. ps ∈R M and O qs ∈R M Then, the two feature vectors are concatenated together to form the feature vector L(O). ps O qsThe system's relative quality score is obtained by processing the feature vector using fully connected layers and the Softmax function. The weight parameters of MGAN are fixed; during scoring, only the weight parameters of the fully connected layers need to be learned, such as... Figure 7 As shown.

[0116] Fraction estimation process:

[0117] (1) The input data sequence of each indicator of the system can be calculated as follows:

[0118]

[0119] The first term represents the contribution of the normalized data sequence input to the headquarters-side model. The bias parameter is ignored during the calculation, and w is the weight parameter.

[0120] (2) Calculate the hidden layer output vector z using the following formula:

[0121]

[0122] The first item is also a contribution from the standardized data sequence input to the headquarters-side model.

[0123] The mass fraction is:

[0124]

[0125] The Softmax function maps values ​​to the range (0, 1), taking two decimal places. Let the quality score of the system index be S'. o Then the training loss function for each index of the system is:

[0126]

[0127] Where N is the number of input data sequences.

[0128] G. Use the similarity value as the accuracy compensation weight between mirror model modeling and direct deployment modeling, thereby reducing the differences between the nodes of the provincial-level new generation electricity information collection system in the test environment and the actual operating environment;

[0129] Based on the accuracy compensation weights, calculate the actual scores of each module, i.e., each functional node, in the system;

[0130] The actual score of each functional node is given by the formula (23):

[0131]

[0132] The actual scores of each functional node are summed and averaged:

[0133]

[0134] m represents the number of functional nodes, i.e., the number of nodes in the graph neural network, and the score T. s It can represent the system's final true score;

[0135] Given the complexity and diversity of the distributed architecture used in the provincial-level new-generation electricity consumption information collection system, this invention constructs a graph neural network model based on the terminal data and link topology of the provincial-level new-generation electricity consumption information collection system. This model abstracts each functional module and terminal device in the system as nodes, and the data flow links as edges, outputting the construction quality assessment probability of each node. This method can characterize the spatial correlation between each functional module of the system, and at the same time, it can take the time-series monitoring data of each node as input, so as to continuously assess the construction quality of the provincial-level system over time and avoid the failure of assessment data.

[0136] Given the differences in the construction of the new generation electricity consumption information collection system across various provinces, a mirror model of the headquarters side is typically used to evaluate the provincial construction system. However, due to differences in deployment environments, there are discrepancies between the actual operating data and the ideal operating environment, which seriously affects the accuracy of the assessment of the construction quality of the provincial side. To achieve real-time and fair verification and evaluation of the construction quality of the provincial side, a direct deployment model and a mirror model are constructed based on the graph neural network model of the new generation electricity consumption information collection system of the provincial side. These models are used as twin network verification and evaluation models to calculate the similarity and differences of the construction of each node in the headquarters side and the provincial side system, thereby improving the accuracy of the assessment of the construction quality of the provincial side.

[0137] Given that the selection of traditional system construction quality assessment indicators often lacks fairness and interpretability, this invention proposes a multi-scenario comprehensive weighted assessment method based on specific business execution completion indicators. After obtaining assessment data based on the execution status of each business, the accuracy compensation weights of the headquarters and provincial systems are calculated by combining the twin network verification assessment model, thereby obtaining the multi-scenario comprehensive weighted assessment score of the system, making the assessment results more reliable and authoritative.

[0138] Example 2:

[0139] This invention also proposes a quality verification system 200 for the construction of a provincial-level electricity consumption information collection system, such as... Figure 8 As shown, it includes:

[0140] The data acquisition unit 201 is used to acquire historical operating data of the target electricity consumption information acquisition system, and generate a graph neural network data sample set of the target electricity consumption information acquisition system based on the historical operating data.

[0141] The model generation unit 202 is used to generate a graph neural network model based on the graph neural network data sample set, and generate a mirror model of the graph neural network model in the simulation environment of the target power consumption acquisition system, and construct a twin network verification and evaluation model based on the graph neural network model and the mirror model.

[0142] Evaluation unit 203 is used to determine the quality evaluation index values ​​of the target electricity consumption information collection system according to the twin network verification evaluation model, perform weighted fusion of the quality evaluation index values ​​to obtain weighted quality evaluation index values, and determine the quality of the target electricity consumption collection system according to the weighted quality evaluation index values.

[0143] Historical operational data includes time-series monitoring data of network nodes in the electricity consumption information collection system at different time periods.

[0144] The generation of a graph neural network model based on the graph neural network data sample set includes:

[0145] Transform the time-series monitoring data in the graph neural network data sample set into graph data format;

[0146] Based on the interconnections of the data flow process of the electricity consumption information collection system, a node adjacency matrix is ​​generated. Based on the adjacency matrix, a node topology connection diagram of the target electricity consumption information collection system is generated, where each node is a functional module of the target electricity consumption information collection system.

[0147] The graph neural network data sample set, which has been transformed into graph data form, is input to each node in the node topology connection graph according to the correspondence between nodes to obtain the input data of the model.

[0148] The input data is fed into a graph neural network for training to generate a graph neural network model.

[0149] The evaluation unit 203 is further configured to construct a twin network verification evaluation model, including: determining the similarity value output by the graph neural network model and the mirror model, using the similarity value as the accuracy compensation value of the mirror model, adjusting the parameters of the mirror model, generating a corrected mirror model, and constructing a twin network verification evaluation model using the corrected mirror model.

[0150] The determination of various quality assessment index values ​​for the target electricity consumption information collection system includes:

[0151] Determine the timeliness and quality of business execution at each node of the target electricity consumption data acquisition system under different business execution scenarios, and determine the business execution completion indicator value based on the timeliness and quality of business execution.

[0152] Based on the business execution completion indicators, determine the values ​​of each quality assessment indicator.

[0153] This invention can effectively evaluate the service execution quality of an electricity information collection system.

[0154] Example 3:

[0155] Based on the same inventive concept, this invention also provides a computer device, which includes a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement corresponding method flows or corresponding functions, thereby implementing the steps of the methods in the above embodiments.

[0156] Example 4:

[0157] Based on the same inventive concept, this invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiments.

[0158] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0159] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0160] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0161] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0162] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0163] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for checking the construction quality of a power side power consumption information collection system, characterized in that, include: Acquire historical operating data of the target electricity consumption information collection system, and generate a graph neural network data sample set based on the historical operating data; The historical operational data includes: time-series monitoring data of the network nodes of the electricity consumption information collection system graph monitored in different time periods; Based on the graph neural network data sample set, a graph neural network model and a mirror model of the graph neural network model are generated. Based on the graph neural network model and the mirror model, a twin network verification and evaluation model is constructed. The step of generating a graph neural network model based on the graph neural network data sample set specifically includes: Transform the time-series monitoring data in the graph neural network data sample set into graph data format; Based on the interconnections of the data flow process in the electricity consumption information collection system, a node adjacency matrix is ​​generated. Based on the adjacency matrix, a node topology connection diagram of the target electricity consumption information collection system is generated, where each node represents a functional module of the target electricity consumption information collection system. The graph neural network data sample set, which has been transformed into graph data form, is input to each node in the node topology connection graph according to the correspondence between nodes to obtain the input data of the model. The input data is fed into a graph neural network for training to generate a graph neural network model; The construction of the twin network verification and evaluation model based on the graph neural network model and the mirror model specifically includes: Among them, a graph neural network model based on the provincial electricity consumption information collection system is deployed at the headquarters as a direct deployment model; The direct deployment model acquires real-time time-series monitoring data of each terminal and node in the provincial system. The graph neural network model constructed by the provincial electricity consumption information collection system was deployed in the simulation test environment at the headquarters as a mirror model. The mirror model inputs simulated terminal data on the scale of the provincial-level system terminal data collection. The data from each terminal on the provincial side and the time-series monitoring data of each node are input into the direct deployment model, and the estimated probability value of the construction quality level of each node is output. The system's construction quality level is assessed based on the estimated probability values ​​of the construction quality level output by each node. Based on the data volume collected by the provincial terminal, simulate the terminal data of the same data volume, input the mirror model, and output the expected probability value of the construction quality level of each node. A twin network verification and evaluation model is constructed based on the direct deployment model and the mirror model. The similarity between the estimated probability value of the construction quality level of each node and the expected probability value of the construction quality level of each node is calculated, which serves as the similarity evaluation value between the construction quality of each node on the provincial side and the construction quality of each node on the headquarters. Constructing a twin network verification and evaluation model includes: using the similarity evaluation value as the accuracy compensation value of the mirror model, adjusting the parameters of the mirror model to generate a corrected mirror model, and constructing a twin network verification and evaluation model using the corrected mirror model. Based on the twin network verification and evaluation model, the quality evaluation index values ​​of the target electricity consumption information collection system are determined, the quality evaluation index values ​​of the various quality evaluation index values ​​are weighted and fused to obtain the weighted quality evaluation index values, and the construction quality of the target electricity consumption information collection system is determined based on the weighted quality evaluation index values. The determination of the quality assessment index values ​​of the target electricity consumption information collection system includes: Determine the timeliness and quality of business execution at each node in the target electricity consumption data acquisition system under different business execution scenarios, and determine the business execution completion indicator value based on the timeliness and quality of business execution. Based on the business execution completion target values, determine the values ​​of each quality assessment indicator; The business execution completion indicators include response speed level and business execution completion level.

2. A system for checking the quality of construction of a power consumption information collection system on the provincial side, the system being used to perform the method for checking the quality of construction of a power consumption information collection system on the provincial side according to claim 1, characterized in that, include: The data acquisition unit is used to acquire historical operating data of the target electricity information acquisition system and generate a graph neural network data sample set based on the historical operating data. The model generation unit is used to generate a graph neural network model based on the graph neural network data sample set, and generate a mirror model of the graph neural network model in the simulation environment of the target power consumption acquisition system. Based on the graph neural network model and the mirror model, a twin network verification and evaluation model is constructed. The evaluation unit is used to determine the quality evaluation index values ​​of the target electricity consumption information collection system according to the twin network verification and evaluation model, perform weighted fusion of the quality evaluation index values ​​to obtain weighted quality evaluation index values, and determine the quality of the target electricity consumption collection system based on the weighted quality evaluation index values.

3. A computer device, comprising: include: One or more processors; A processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the method of claim 1 is implemented.

4. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements the method as described in claim 1.

Images