Network energy consumption assessment method and apparatus, and device, storage medium and program product

Abstract

Provided in the present disclosure are a network energy consumption assessment method and apparatus, and a device, a storage medium and a program product. The method comprises: acquiring node feature data and node energy consumption data of a network to be subjected to assessment; inputting the node energy consumption data into a first model, so as to obtain node energy consumption, which is outputted by the first model, of each network node within a future preset time period; inputting the node feature data into a second model, so as to obtain total network energy consumption, which is outputted by the second model, of said network within the future preset time period; and on the basis of the energy consumption of each node and the total network energy consumption, assessing the network energy consumption of said network within the future preset time period. The present disclosure can realize network-level energy consumption assessment and analysis, improve the accuracy of an assessment result, adapt to application requirements of a complex network, and achieve a high practicability.

Description

Network energy consumption evaluation method and device, equipment, storage medium and program product

[0001] Cross-reference to related applications

[0002] The present disclosure is based on and claims priority from Chinese Patent Application No. 202411031476.1 filed on July 30, 2024, the entire contents of which are incorporated herein by reference. TECHNICAL FIELD

[0003] The present disclosure relates to the field of artificial intelligence, and in particular, to a network energy consumption evaluation method, device, equipment, storage medium and program product. BACKGROUND

[0004] Network energy consumption evaluation is of great significance to device regulation in the entire network or system. For example, energy network energy consumption prediction and evaluation can be used to guide the operation of various energy devices in the energy network, thereby achieving load balancing and reducing loss.

[0005] A network usually includes multiple network nodes, and each network node can be an independent device. In some related technologies, network energy consumption evaluation focuses on a single device, i.e., a single network node, and simulates and analyzes the energy consumption of each device in the network through mathematical models or physical models, such as power grids, gas networks, and heating networks.

[0006] Since existing network energy consumption evaluation methods focus on a single network node, it is difficult to comprehensively evaluate the energy consumption of the entire network, resulting in inaccurate network energy consumption evaluation results and difficulty in adapting to complex network application requirements. SUMMARY

[0007] Embodiments of the present disclosure provide a network energy consumption evaluation method, device, equipment, storage medium and program product to solve the technical problem of inaccurate evaluation results of existing network energy consumption evaluation methods and difficulty in adapting to complex network application requirements.

[0008] In a first aspect, embodiments of the present disclosure provide a network energy consumption evaluation method, comprising: obtaining node feature data and node energy consumption data of a network to be evaluated, wherein the network to be evaluated includes at least two network nodes; inputting the node energy consumption data into a first target model to obtain node energy consumption of each network node in a future preset time period output by the first target model; inputting the node feature data into a second target model to obtain network total energy consumption of the network to be evaluated in the future preset time period output by the second target model; and based on each node energy consumption and the network total energy consumption, evaluating network energy consumption of the network to be evaluated in the future preset time period.

[0009] In one embodiment, the node energy consumption data includes historical energy consumption data of each network node, the historical energy consumption data including energy consumption data in a plurality of preset historical time periods; the first target prediction model includes a plurality of time windows, an LSTM network layer and a first full connection layer, the LSTM network layer and the first full connection layer being connected, the LSTM network layer including a plurality of LSTM units connected in sequence, each time window being connected with an LSTM unit; each time window is configured to store energy consumption data of each network node in a preset historical time period; the LSTM network layer is configured to extract features from the energy consumption data in each preset historical time period to obtain time sequence features; and the first full connection layer is configured to make a prediction based on the time sequence features to obtain node energy consumption of each network node in a future preset time period.

[0010] In one embodiment, the second target model includes a first graph convolution network layer, a second graph convolution network layer, a third graph convolution network layer, a connection layer, a first convolution layer, a second convolution layer and a second full connection layer connected in sequence, the first graph convolution network layer and the second graph convolution network layer being connected with the connection layer; the first graph convolution network layer is configured to perform convolution processing on the node feature data to obtain first features; the second graph convolution network layer is configured to perform convolution processing on the first features to obtain second features; the third graph convolution network layer is configured to perform convolution processing on the second features to obtain third features; the connection layer is configured to perform connection processing on the first features, the second features and the third features to obtain fourth features; the first convolution layer is configured to extract features from the fourth features to obtain fifth features; the second convolution layer is configured to extract features from the fifth features to obtain sixth features; and the second full connection layer is configured to make a prediction based on the sixth features to obtain total network energy consumption of the network to be evaluated in a future preset time period.

[0011] In one embodiment, based on the node energy consumption and the total network energy consumption, the network energy consumption of the network to be evaluated in the future preset time period is evaluated, including: determining total data volume and total number of network nodes of the network to be evaluated; determining average energy efficiency ratio of the network to be evaluated based on the total data volume and the total network energy consumption; determining high energy consumption nodes in the network to be evaluated based on the node energy consumption, wherein the high energy consumption nodes are network nodes with node energy consumption higher than a preset threshold; determining an energy consumption hotspot proportion of the network to be evaluated based on the number of high energy consumption nodes and the total number of network nodes; and evaluating the network energy consumption of the network to be evaluated in the future preset time period based on the total network energy consumption, the average energy efficiency ratio and the energy consumption hotspot proportion.

[0012] In one embodiment, the node feature data of the network to be evaluated is obtained, including: obtaining network topology structure data of the network to be evaluated and node attribute data of each network node; and performing data preprocessing on the network topology structure data and the node attribute data to obtain the node feature data.

[0013] In an embodiment, before the node feature data and the node energy consumption data of the network to be evaluated are acquired, the method further comprises: acquiring sample node feature data of a sample network, sample network total energy consumption corresponding to the sample node feature data, sample node energy consumption data, and sample node energy consumption corresponding to the sample node energy consumption data; pre-training the first initial model based on the sample node energy consumption data and the sample node energy consumption corresponding to the sample node energy consumption data to obtain a first target model; and pre-training the second initial model based on the sample node feature data and the sample network total energy consumption corresponding to the sample node feature data to obtain a second target model.

[0014] In a second aspect, the embodiments of the present disclosure provide a network energy consumption evaluation device, comprising: an acquisition module configured to acquire node feature data and node energy consumption data of a network to be evaluated, wherein the network to be evaluated comprises at least two network nodes; a node energy consumption prediction module configured to input the node energy consumption data into a first target model to obtain node energy consumption of each network node in a future preset time period output by the first target model; a network energy consumption prediction module configured to input the node feature data into a second target model to obtain network total energy consumption of the network to be evaluated in the future preset time period output by the second target model; and an evaluation module configured to evaluate network energy consumption of the network to be evaluated in the future preset time period based on the node energy consumption of each network node and the network total energy consumption.

[0015] In a third aspect, the embodiments of the present disclosure provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the above network energy consumption evaluation methods when executing the computer program.

[0016] In a fourth aspect, the embodiments of the present disclosure provide a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement any of the above network energy consumption evaluation methods.

[0017] In a fifth aspect, the embodiments of the present disclosure provide a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement any of the above network energy consumption evaluation methods.

[0018] According to the network energy consumption evaluation method, device, equipment, storage medium and program product provided by the embodiment of the present disclosure, the node energy consumption of each network node in a future preset time period is predicted through a node energy consumption prediction model, the total network energy consumption of the network to be evaluated in the future preset time period is predicted through a network energy consumption prediction model, and the network energy consumption of the network to be evaluated in the future preset time period is evaluated according to the node energy consumption of each node and the total network energy consumption. In the process of network energy consumption evaluation, the present disclosure considers not only the influence of a single network node on energy consumption, but also the influence of the whole network on energy consumption, so that network-level energy consumption evaluation and analysis can be realized, the accuracy of the evaluation result is improved, the application requirements of complex networks are met, and the practicability is high. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present disclosure or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the present disclosure, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0020] Fig. 1 is a flow diagram of a network energy consumption evaluation method provided by an embodiment of the present disclosure.

[0021] Fig. 2 is a flow diagram of a network energy consumption evaluation method provided by an embodiment of the present disclosure.

[0022] Fig. 3 is a structural diagram of a network graph provided by an embodiment of the present disclosure.

[0023] Fig. 4 is a structural diagram of a node energy consumption prediction model provided by an embodiment of the present disclosure.

[0024] Fig. 5 is a structural diagram of a network energy consumption prediction model provided by an embodiment of the present disclosure.

[0025] Fig. 6 is a structural diagram of a network energy consumption evaluation device provided by an embodiment of the present disclosure.

[0026] Fig. 7 is a structural diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the technical solutions in the present disclosure will be described clearly and completely below in combination with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present disclosure.

[0028] It should be noted that the embodiments of the present disclosure can involve the use of user data. In actual application, user-specific personal data can be used in the schemes described herein in a manner that complies with the applicable legal and regulatory requirements of the country in which the user is located (e.g., with the explicit consent of the user, with adequate notice to the user, etc.), within the scope permitted by applicable laws and regulations.

[0029] Referring to FIG. 1, FIG. 1 is one of flow diagrams of a network energy consumption evaluation method provided by the embodiments of the present disclosure. In the embodiments of the present disclosure, the network energy consumption evaluation method includes steps S110 to S140, and each step is as follows:

[0030] S110: Obtain node feature data and node energy consumption data of a network to be evaluated.

[0031] In some embodiments, the network to be evaluated includes at least two network nodes, the node feature data is determined based on network topology structure data of the network to be evaluated and node attribute data of each network node, and the node energy consumption data includes historical energy consumption data of each network node.

[0032] Specifically, the relevant data of the network to be evaluated is obtained, and the relevant data includes but is not limited to: network topology structure data of the network to be evaluated, node attribute data of each network node, historical energy consumption data and current energy consumption data of each network node, etc.

[0033] Further, the network topology structure data, each node attribute data, historical energy consumption data and current energy consumption data of each network node, etc. are subjected to data preprocessing such as data cleaning, normalization and feature engineering, to obtain the node feature data and the node energy consumption data.

[0034] S120: Input the node energy consumption data into a first target model to obtain node energy consumption of each network node in a future preset time period.

[0035] Specifically, for the historical energy consumption data of each network node, the historical energy consumption data of the network node is input into a node energy consumption prediction model to obtain node energy consumption of the network node in a future preset time period output by the node energy consumption prediction model. That is, in the present disclosure, the first target model can be a node energy consumption prediction model.

[0036] Specifically, the node energy consumption prediction model is a Long Short-Term Memory (LSTM) model, which can analyze the energy consumption time series data of each network node to predict the node energy consumption of the network node in a future preset time period.

[0037] S130: input the node feature data into the second target model to obtain network total energy consumption of the to-be-evaluated network in a future preset time period output by the second target model.

[0038] Specifically, the second target model can be a network energy consumption prediction model, for example, a Graph Convolutional Network (GCN) model. The GCN model can learn the topology of the network and the interaction between nodes, comprehensively consider node characteristics and connection characteristics, and predict network total energy consumption of the entire network in a future preset time period.

[0039] S140: based on the energy consumption of each node and the network total energy consumption, evaluate network energy consumption of the to-be-evaluated network in a future preset time period.

[0040] In the embodiments of the present disclosure, the network energy consumption of the to-be-evaluated network in a future preset time period refers to multi-dimensional network energy consumption indicators of the to-be-evaluated network in a future preset time period.

[0041] According to the network energy consumption evaluation method provided in the embodiments of the present disclosure, the node energy consumption prediction model is used to predict the node energy consumption of each network node in a future preset time period, the network energy consumption prediction model is used to predict the network total energy consumption of the to-be-evaluated network in a future preset time period, and the network energy consumption of the to-be-evaluated network in a future preset time period is evaluated based on the energy consumption of each node and the network total energy consumption. In the process of network energy consumption evaluation, the present disclosure considers both the influence of a single network node on energy consumption and the influence of the entire network on energy consumption, can realize network-level energy consumption evaluation and analysis, improve the accuracy of the evaluation result, adapt to the application requirements of complex networks, and has high practicability.

[0042] In some embodiments, the node energy consumption data includes historical energy consumption data of each network node, and the historical energy consumption data includes energy consumption data in a plurality of preset historical time periods; the node energy consumption prediction model includes a plurality of time windows, an LSTM network layer, and a first full connection layer, the LSTM network layer and the first full connection layer are connected, the LSTM network layer includes a plurality of LSTM units connected in sequence, and each time window is connected with an LSTM unit; each time window is configured to store energy consumption data of each network node in a preset historical time period; the LSTM network layer is configured to perform feature extraction on the energy consumption data in each preset historical time period to obtain time sequence features; and the first full connection layer is configured to perform prediction based on the time sequence features to obtain node energy consumption of each network node in a future preset time period.

[0043] Please refer to FIG. 2, which is a flowchart of the network energy consumption evaluation method provided in the embodiments of the present disclosure.

[0044] As shown in FIG. 2, the network energy consumption evaluation method includes 5 steps.

[0045] Specifically, relevant data of the network to be evaluated is acquired, which includes but is not limited to network topology data of the network to be evaluated, node attribute data of each network node, historical energy consumption data and current energy consumption data of each network node, etc., and data collection is completed.

[0046] Further, data preprocessing such as data cleaning, normalization and feature engineering is performed on the network topology data, node attribute data of each node, historical energy consumption data and current energy consumption data of each network node, etc., to obtain node feature data and node energy consumption data.

[0047] In this embodiment, one network to be evaluated can be represented by a network graph.

[0048] Please refer to FIG. 3, which is a structural schematic diagram of a network graph provided by the embodiment of the present disclosure.

[0049] Specifically, as shown in FIG. 3, the network to be evaluated can be represented as a network graph G=(V, E), where V is a set of network nodes (i.e. circles in FIG. 3), and E is a set of edges (i.e. straight lines in FIG. 3). Each network node represents an entity in the network (for example, a wired router, a switch or an optical modem, etc.), and the edge represents the connection between the network nodes (for example, network cable connection and wifi bridging, etc.).

[0050] The network topology of the network to be evaluated can be represented by an adjacency matrix A, where the element A ij in the adjacency matrix A is defined as follows:

[0051] Further, after obtaining the adjacency matrix A of the network to be evaluated, the adjacency matrix A is normalized to obtain a normalized adjacency matrix The expression of the normalized adjacency matrix is as follows:

[0052] where I is an identity matrix; D is a diagonal matrix, and each element D ii of the diagonal matrix D represents the degree of the network node i (i.e. the number of edges connected to the network node i).

[0053] In some embodiments of the present disclosure, the normalized adjacency matrix can be used as the network topology data of the network to be evaluated.

[0054] For each network node vi∈V in the network graph, a feature vector is defined to represent the attribute of the network node, and the node attribute data is obtained.

[0055] The attributes of network nodes include device type, device status, device traffic volume, and number of device connections.

[0056] Optionally, the device types include routers, switches, optical modems, etc.

[0057] Optionally, the device status includes on or off.

[0058] Specifically, the feature vector can be denoted as X. i =[x i1 ,x i2 ,...,x in ], where x ij Let be the j-th feature of network node i, and n be the total number of features.

[0059] The feature vectors of all network nodes can be combined to form a feature matrix X, where each row of the feature matrix X corresponds to the feature vector of a network node. The size of the feature matrix X is |V| × n, where |V| is the number of network nodes and n is the total number of features.

[0060] The feature matrix X can be used as node attribute data for the network to be evaluated.

[0061] In this embodiment, the node feature data includes the feature matrix X and the normalized adjacency matrix.

[0062] Furthermore, the process of predicting node energy consumption includes inputting the historical energy consumption data of each network node into the node energy consumption prediction model to obtain the node energy consumption of each network node in the future within a preset time period, as output by the node energy consumption prediction model.

[0063] Please refer to Figure 4, which is a schematic diagram of the structure of the node energy consumption prediction model provided in the embodiments of this disclosure.

[0064] As shown in Figure 4, the node energy consumption prediction model includes multiple time windows, an LSTM network layer, and a first fully connected layer.

[0065] In this embodiment of the disclosure, the LSTM network layer and the first fully connected layer are connected. The LSTM network layer includes a plurality of LSTM units connected in sequence, and each time window is connected to one LSTM unit.

[0066] Specifically, the input data for a single node energy consumption prediction model is the historical energy consumption data of a network node.

[0067] Understandably, since the scale of historical energy consumption data for each network node may vary, it is necessary to normalize the data to improve the convergence speed and prediction accuracy of the node energy consumption prediction model.

[0068] Specifically, the historical energy consumption data includes energy consumption data in a plurality of preset historical time periods, and for the historical energy consumption data of one network node, the historical energy consumption data needs to be serialized to convert the historical energy consumption data into a format of a series of time windows, each time window containing energy consumption data in a preset historical time period as input, and the node energy consumption prediction model can correspondingly output node energy consumption of the network node in a future preset time period.

[0069] Specifically, each time window is used to store energy consumption data of each network node in a preset historical time period.

[0070] The LSTM network layer is used to extract features from the energy consumption data in each preset historical time period to obtain time sequence features.

[0071] Specifically, each LSTM unit in the LSTM network layer can extract features from and learn the energy consumption data in a preset historical time period of one time window, learn time-dependent relationships in the historical energy consumption data, and obtain time sequence features.

[0072] The last LSTM unit in the LSTM network layer is connected with the first fully connected layer, and the first fully connected layer is used to make a prediction based on the time sequence features to obtain node energy consumption of the network node in a future preset time period.

[0073] Further, the process of network energy consumption prediction includes inputting the node feature data into the network energy consumption prediction model to obtain network total energy consumption of the network to be evaluated in a future preset time period output by the network energy consumption prediction model.

[0074] Further, based on the energy consumption of each node and the network total energy consumption, network energy consumption of the network to be evaluated in a future preset time period is evaluated to realize multi-dimensional energy consumption evaluation.

[0075] The network energy consumption evaluation method provided by the embodiments of the present disclosure constructs a node energy consumption prediction model through an LSTM network, so that the node energy consumption prediction model has the ability to analyze time sequence data, can accurately predict future energy consumption changes of each network node, and provides a dynamic energy consumption prediction scheme to provide a scientific basis for energy efficiency management and resource planning.

[0076] In some embodiments, the network energy consumption prediction model comprises a first graph convolutional network layer, a second graph convolutional network layer, a third graph convolutional network layer, a connection layer, a first convolutional layer, a second convolutional layer and a second fully connected layer connected in sequence, and the first graph convolutional network layer, the second graph convolutional network layer and the third graph convolutional network layer are connected with the connection layer respectively; the first graph convolutional network layer is configured to perform convolutional processing on the node feature data to obtain a first feature; the second graph convolutional network layer is configured to perform convolutional processing on the first feature to obtain a second feature; the third graph convolutional network layer is configured to perform convolutional processing on the second feature to obtain a third feature; the connection layer is configured to perform connection processing on the first feature, the second feature and the third feature to obtain a fourth feature; the first convolutional layer is configured to perform feature extraction on the fourth feature to obtain a fifth feature; the second convolutional layer is configured to perform feature extraction on the fifth feature to obtain a sixth feature; and the second fully connected layer is configured to perform prediction based on the sixth feature to obtain the total network energy consumption of the network to be evaluated in a future preset time period.

[0077] After obtaining the node feature data, network energy consumption prediction is needed.

[0078] Referring to FIG. 5, FIG. 5 is a structural schematic diagram of a network energy consumption prediction model provided by an embodiment of the present disclosure.

[0079] As shown in FIG. 5, the network energy consumption prediction model comprises a first graph convolutional network layer, a second graph convolutional network layer, a third graph convolutional network layer, a connection layer, a first convolutional layer, a second convolutional layer and a second fully connected layer connected in sequence, and the first graph convolutional network layer, the second graph convolutional network layer and the third graph convolutional network layer are connected with the connection layer respectively.

[0080] The node feature data comprises a feature matrix X and a normalized adjacency matrix The network energy consumption prediction model first performs graph convolutional operation through a plurality of graph convolutional network layers (including the first graph convolutional network layer, the second graph convolutional network layer and the third graph convolutional network layer), and updates the feature representation of the network node (i.e., the feature matrix X) by using the information of the neighbor network node (i.e., the adjacency matrix ).

[0081] Specifically, the first graph convolutional network layer is configured to perform convolutional processing (i.e., graph convolutional operation) on the node feature data to obtain a first feature; the second graph convolutional network layer is configured to perform convolutional processing on the first feature to obtain a second feature; and the third graph convolutional network layer is configured to perform convolutional processing on the second feature to obtain a third feature.

[0082] Specifically, the graph convolutional operation of the graph convolutional network layer can be represented by the following formula:

[0083] wherein, H (l)For the feature matrix of the lth graph convolution network layer, for the first graph convolution network layer (i.e., the first graph convolution network layer, denoted as H (0) ), there is H (0) =X; is a normalized adjacency matrix; W (l) is the weight matrix of the lth graph convolution network layer; sigma is a nonlinear activation function, specifically a ReLU function, sigma(x)=max(0,x); H (l+1) is the feature matrix of the (l+1)th graph convolution network layer.

[0084] Further, a connection layer is configured to perform connection processing on the first feature, the second feature, and the third feature to obtain a fourth feature, i.e., a fusion feature.

[0085] Further, a first convolution layer is configured to perform feature extraction on the fourth feature to obtain a fifth feature; and a second convolution layer is configured to perform feature extraction on the fifth feature to obtain a sixth feature.

[0086] The first convolution layer performs 1*1 convolution processing on the fourth feature to realize feature extraction and obtain the fifth feature; and the second convolution layer performs 3*3 convolution processing on the fifth feature to realize feature extraction and obtain the sixth feature.

[0087] Further, a second fully connected layer is configured to perform prediction based on the sixth feature to obtain the total network energy consumption of the network to be evaluated within a future preset time period.

[0088] According to the network energy consumption evaluation method provided in the embodiments of the present disclosure, the network energy consumption prediction model is constructed through the graph convolution network layer, and the graph convolution operation is performed through the graph convolution network layer, so that the network energy consumption prediction model can extract and aggregate the feature information of the neighbor nodes, thereby updating the feature representation of each network node. The network energy consumption prediction model can capture the topological dependency of the network graph, i.e., by learning the connection mode between the nodes in the graph to extract effective features. A plurality of graph convolution network layers are stacked, and the output of each layer is based on the feature representation of the previous layer. Through further graph convolution operation, deeper graph structure information and node features are refined. The multi-layer structure enables the network energy consumption prediction model to learn the complex patterns of the network nodes within a plurality of hop neighborhoods, thereby improving the feature extraction effect. At the same time, the outputs of the plurality of graph convolution network layers are spliced through the connection layer, and then connected to the convolution layer for further feature extraction, thereby realizing the analysis of the network topology and identifying the connection and structure patterns that have an important impact on the energy consumption evaluation. The overall energy consumption of the entire network, including the energy consumption of all nodes and connections, can be accurately predicted. Compared with the prior art, the network energy consumption evaluation method provided in the embodiments of the present disclosure can comprehensively analyze the influence of the topological structure of the network and the interaction between the nodes on the energy efficiency, and can provide support for subsequent energy efficiency optimization guidance.

[0089] In some embodiments, based on the per-node energy consumption and the total network energy consumption, the network energy consumption of the network to be evaluated in a future preset time period is evaluated, including: determining the total data amount of the network to be evaluated and the total number of network nodes; based on the total data amount and the total network energy consumption, determining the average energy efficiency ratio of the network to be evaluated; based on the per-node energy consumption, determining high-energy-consumption nodes in the network to be evaluated, wherein the high-energy-consumption nodes are network nodes with node energy consumption higher than a preset threshold; based on the number of high-energy-consumption nodes and the total number of network nodes, determining the energy consumption hotspot proportion of the network to be evaluated; and based on the total network energy consumption, the average energy efficiency ratio, and the energy consumption hotspot proportion, evaluating the network energy consumption of the network to be evaluated in the future preset time period.

[0090] After obtaining the per-node energy consumption and the total network energy consumption, a plurality of evaluation indexes can be calculated based on the per-node energy consumption and the total network energy consumption, each of which evaluates different aspects of the network energy consumption, achieving multi-dimensional energy consumption evaluation.

[0091] In this embodiment, the evaluation indexes can include the total network energy consumption (TNE), the average energy efficiency ratio (AEE), and the energy consumption hotspot proportion (HPR).

[0092] Specifically, the total data amount of the network to be evaluated and the total number of network nodes are determined. Further, based on the total data amount and the total network energy consumption, the average energy efficiency ratio (AEE) of the network to be evaluated is determined. The value of AEE is the ratio of the total data amount to the total network energy consumption, i.e., total data amount / total network energy consumption. The total data amount refers to the total amount of data successfully transmitted by the network during the evaluation period, which can be obtained by adding up the data amounts of all relevant network devices and interfaces during the evaluation period.

[0093] Further, based on the per-node energy consumption, high-energy-consumption nodes in the network to be evaluated are determined. Based on the number of high-energy-consumption nodes and the total number of network nodes, the HPR of the network to be evaluated is determined. The value of HPR is the ratio of the number of high-energy-consumption nodes to the total number of network nodes, i.e., the number of high-energy-consumption nodes / total number of network nodes. The high-energy-consumption nodes are network nodes with node energy consumption higher than a preset threshold. For example, if the node energy consumption of a network node is greater than or equal to 90% of the rated power of the network node, the network node is determined to be a high-energy-consumption node. It can be understood that the preset threshold can be adjusted according to actual needs.

[0094] Further, based on TNE, AEE, and HPR, the network energy consumption of the network to be evaluated in a future preset time period is evaluated.

[0095] TNE is a direct quantification of the total energy consumption of the entire network in a certain time period, which not only considers the energy consumption of each network node, but also comprehensively considers the energy consumption of links in the network, the running state of devices, and other factors. The TNE index can directly reflect the overall energy efficiency of the network and provides a basis for evaluating the energy efficiency of the network.

[0096] AEE is the ratio of total data amount and total energy consumption of the network, reflecting the energy efficiency performance of the network in unit data transmission process. This evaluation index is particularly suitable for measuring data transmission intensive energy networks such as data centers and communication networks, providing an evaluation benchmark for improving data transmission efficiency.

[0097] HPR can be used to identify network nodes or areas with abnormally high energy consumption, and the state of the network is evaluated by the proportion of the number of high energy consumption nodes to the total number of network nodes. HPR not only reveals the uniformity of energy consumption distribution in the network, but also points out potential energy efficiency optimization points in the network. Especially for large and complex networks, this index helps to quickly locate the key areas of energy efficiency management.

[0098] The network energy consumption evaluation method provided by the embodiments of the present disclosure adopts multi-dimensional evaluation indexes to comprehensively evaluate the energy consumption of the network from multiple angles, which can provide multi-aspect decision support for energy efficiency optimization and provide scientific basis and optimization strategy for network system level energy efficiency management, promote the efficient operation of the network, and realize the fine management of energy efficiency. In addition, the network energy consumption evaluation method provided by the embodiments of the present disclosure has high adaptability and can be applied to different types and scales of networks, including but not limited to power grids, heat networks, gas networks, etc., and has strong practicality.

[0099] In some embodiments, the node feature data of the network to be evaluated is obtained, including: obtaining network topology structure data of the network to be evaluated and node attribute data of each network node; and performing data preprocessing on the network topology structure data and the node attribute data of each node to obtain the node feature data.

[0100] In some embodiments, before the node feature data and the node energy consumption data of the network to be evaluated are obtained, the method further includes: obtaining sample node feature data of a sample network, sample network total energy consumption corresponding to the sample node feature data, sample node energy consumption data, and sample node energy consumption corresponding to the sample node energy consumption data; pre-training a first initial model based on the sample node energy consumption data and the sample node energy consumption corresponding to the sample node energy consumption data to obtain a first target model; and pre-training a second initial model based on the sample node feature data and the sample network total energy consumption corresponding to the sample node feature data to obtain a second target model.

[0101] The embodiments of the present disclosure also provide a specific example of a network energy consumption evaluation method, which is given in the form of simulation.

[0102] To simulate the proposed technical solutions and introduce the input content of the method, a set of detailed simulation data needs to be constructed to reflect the network topology, network node characteristics, and energy consumption data, etc. The following is an example of simulation data based on a given network structure (one master switch, two slave switches, one of which is connected in parallel with four routers, and the other is connected in series with four routers).

[0103] Network node set: {master switch, slave switch 1, slave switch 2, router 1, router 2, router 3, router 4, router 5, router 6, router 7, router 8}.

[0104] Edge set: {(master switch, slave switch 1), (master switch, slave switch 2), (slave switch 1, router 1), (slave switch 1, router 2), (slave switch 1, router 3), (slave switch 1, router 4), (slave switch 2, router 5), (router 5, router 6), (router 6, router 7), (router 7, router 8)}.

[0105] Assume that the feature vector of each network node contains the following attributes: device type, device state, average traffic size (Mbps), number of connections, historical energy consumption (kWh). The device type is represented by a number, such as 1 for master switch, 2 for slave switch, and 3 for router. The device state is represented by 0 for off and 1 for on, and the features of each network node are shown in Table 1.

[0106] Table 1

[0107] Specifically, after obtaining the node feature data and node energy consumption data of the network to be evaluated, the historical energy consumption data of each network node is input into the node energy consumption prediction model respectively, and the node energy consumption of each network node in the future preset time period output by the node energy consumption prediction model is obtained.

[0108] Specifically, assume that the historical energy consumption data of router 1 is in kWh, and records the energy consumption of the past 5 time points (i.e. 5 preset historical time periods), the specific data is shown in Table 2. Based on these data, the node energy consumption prediction model based on LSTM network is used to predict the node energy consumption at the next time point (i.e. future preset time period).

[0109] Table 2

[0110] Based on Table 2, the input data of the node energy consumption prediction model can be understood as a data sequence (M): [

[0020] ,

[0022] ,

[0021] ,

[0023] ]; and the output data of the node energy consumption prediction model is represented as (Y):

[0024] , that is, the value output by the node energy consumption prediction model is the node energy consumption at the next time point. For example, the node energy consumption prediction model can predict that the node energy consumption at T6 is 25 kWh.

[0111] Further, the node feature data is input into the network energy consumption prediction model to obtain the network total energy consumption of the network to be evaluated in a future preset time period output by the network energy consumption prediction model.

[0112] Specifically, based on Table 1, a feature matrix X and a normalized adjacency matrix are constructed as node feature data; the feature matrix X and the normalized adjacency matrix are input into the network energy consumption prediction model, and the network energy consumption prediction model predicts the network total energy consumption of the network in a future preset time period according to the input feature matrix X and the normalized adjacency matrix .

[0113] Further, after obtaining each node energy consumption and the network total energy consumption, a plurality of evaluation indexes can be calculated based on each node energy consumption and the network total energy consumption, each evaluation index evaluating different aspects of network energy consumption to achieve multi-dimensional energy consumption evaluation.

[0114] Suppose that the total amount of data successfully transmitted by the network in a certain time (for example, one day) is 1000 GB, and the network total energy consumption (TNE) is 295.25 kWh, then the value of the average energy efficiency ratio (AEE) is 1000 GB / 295.25 kWh≈3.39 GB / kWh.

[0115] Suppose that among all the results output by the node energy consumption prediction model, the energy consumption of 2 network nodes exceeds 80% of the rated power, and they are considered as high energy consumption nodes, then the value of the energy consumption hotspot proportion (HPR) is 2 / 11≈18.18%.

[0116] The embodiments of the present disclosure also provide a network energy consumption evaluation device. Please refer to FIG. 6, which is a structural schematic diagram of a network energy consumption evaluation device provided by the embodiments of the present disclosure. In the embodiments of the present disclosure, the network energy consumption evaluation device comprises an acquisition module 610, a node energy consumption prediction module 620, a network energy consumption prediction module 630 and an evaluation module 640.

[0117] The acquisition module 610 is configured to acquire node feature data and node energy consumption data of a network to be evaluated. The network to be evaluated comprises at least two network nodes.

[0118] The node energy consumption prediction module 620 is configured to input the node energy consumption data into the first target model to obtain node energy consumption of each network node in a future preset time period output by the first target model.

[0119] The network energy consumption prediction module 630 is configured to input the node feature data into the second target model to obtain total network energy consumption of the network to be evaluated in a future preset time period output by the second target model.

[0120] The evaluation module 640 is configured to evaluate the network energy consumption of the network to be evaluated in the future preset time period based on the node energy consumption of each network node and the total network energy consumption.

[0121] In some embodiments, the node energy consumption data includes historical energy consumption data of each network node, and the historical energy consumption data includes energy consumption data in a plurality of preset historical time periods; the first target model includes a plurality of time windows, an LSTM network layer, and a first full connection layer, the LSTM network layer and the first full connection layer are connected, the LSTM network layer includes a plurality of LSTM units connected in sequence, and each time window is connected with an LSTM unit; each time window is configured to store energy consumption data of each network node in a preset historical time period; the LSTM network layer is configured to perform feature extraction on the energy consumption data in each preset historical time period to obtain time sequence features; and the first full connection layer is configured to perform prediction based on the time sequence features to obtain node energy consumption of each network node in a future preset time period.

[0122] In some embodiments, the second target model includes a first graph convolutional network layer, a second graph convolutional network layer, a third graph convolutional network layer, a connection layer, a first convolutional layer, a second convolutional layer, and a second full connection layer connected in sequence, and the first graph convolutional network layer, the second graph convolutional network layer, and the third graph convolutional network layer are connected with the connection layer respectively; the first graph convolutional network layer is configured to perform convolutional processing on the node feature data to obtain first features; the second graph convolutional network layer is configured to perform convolutional processing on the first features to obtain second features; the third graph convolutional network layer is configured to perform convolutional processing on the second features to obtain third features; the connection layer is configured to perform connection processing on the first features, the second features, and the third features to obtain fourth features; the first convolutional layer is configured to perform feature extraction on the fourth features to obtain fifth features; the second convolutional layer is configured to perform feature extraction on the fifth features to obtain sixth features; and the second full connection layer is configured to perform prediction based on the sixth features to obtain the total network energy consumption of the network to be evaluated in the future preset time period.

[0123] In some embodiments, the evaluation module 640 is configured to determine a total data amount and a total number of network nodes of the network to be evaluated, determine an average energy efficiency ratio of the network to be evaluated based on the total data amount and a total energy consumption of the network, determine a high energy consumption node in the network to be evaluated based on an energy consumption of each node, wherein the high energy consumption node is a network node whose energy consumption is higher than a preset threshold, determine a proportion of energy consumption hotspots of the network to be evaluated based on a number of the high energy consumption nodes and the total number of network nodes, and evaluate network energy consumption of the network to be evaluated in a future preset time period based on the total energy consumption of the network, the average energy efficiency ratio, and the proportion of energy consumption hotspots.

[0124] In some embodiments, the acquisition module 610 is configured to acquire network topology data of the network to be evaluated and node attribute data of each network node, and perform data preprocessing on the network topology data and the node attribute data to obtain node feature data.

[0125] In some embodiments, the acquisition module 610 is configured to acquire sample node feature data of a sample network, a total energy consumption of the sample network corresponding to the sample node feature data, sample node energy consumption data, and a sample node energy consumption corresponding to the sample node energy consumption data, pre-train a first initial model based on the sample node energy consumption data and the sample node energy consumption corresponding to the sample node energy consumption data to obtain a first target model, and pre-train a second initial model based on the sample node feature data and the total energy consumption of the sample network corresponding to the sample node feature data to obtain a second target model.

[0126] The embodiments of the present disclosure further provide an electronic device, and FIG. 7 is a structural schematic diagram of the electronic device according to the embodiments of the present disclosure. As shown in FIG. 7, the electronic device can include a processor 710, a communications interface 720, a memory 730, and a communications bus 740. The processor 710, the communications interface 720, and the memory 730 can communicate with each other through the communications bus 740. The processor 710 can invoke a logical instruction in the memory 730 to execute the network energy consumption evaluation method.

[0127] In addition, the logic instructions in the memory 730 described above can be implemented in the form of software function units and sold or used as independent products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present disclosure essentially or in part or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present disclosure. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0128] The embodiments of the present disclosure further provide a non-transitory computer-readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the network energy consumption evaluation method provided by the above-mentioned methods.

[0129] The embodiments of the present disclosure further provide a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the network energy consumption evaluation method provided by the above-mentioned methods.

[0130] The above-mentioned method embodiments of the present disclosure can be combined to form new embodiments without conflict, and the present disclosure does not limit this.

[0131] The above-described device embodiments are only schematic, and the units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement it without creative labor.

[0132] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0133] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for assessing network energy consumption, comprising: Obtain node characteristic data and node energy consumption data of the network to be evaluated, wherein the network to be evaluated includes at least two network nodes; The node energy consumption data is input into the first target model to obtain the node energy consumption of each network node in a future preset time period output by the first target model. The node feature data is input into the second target model to obtain the total network energy consumption of the network to be evaluated in the future preset time period output by the second target model. as well as Based on the energy consumption of each node and the total energy consumption of the network, the network energy consumption of the network to be evaluated is assessed within the preset future time period.

2. The network energy consumption assessment method according to claim 1, wherein, The node energy consumption data includes the historical energy consumption data of each network node, and the historical energy consumption data includes energy consumption data within multiple preset historical time periods; The first target model includes multiple time windows, an LSTM network layer, and a first fully connected layer. The LSTM network layer is connected to the first fully connected layer. The LSTM network layer includes multiple LSTM units connected in sequence. Each time window is connected to one of the LSTM units. Each of the aforementioned time windows is used to store the energy consumption data of each of the aforementioned network nodes within a preset historical time period; The LSTM network layer is used to extract features from the energy consumption data within each preset historical time period to obtain time-series features. The first fully connected layer is used to make predictions based on the time-series features to obtain the node energy consumption of each network node in a future preset time period.

3. The network energy consumption assessment method according to claim 1, wherein, The second target model includes a plurality of graph convolutional network layers, a connection layer, a plurality of convolutional layers, and a second fully connected layer connected in sequence, wherein the plurality of graph convolutional network layers are respectively connected to the connection layer; The multiple graph convolutional network layers are used to perform graph convolution processing by updating the feature matrix of the network nodes using information from neighboring network nodes, and to obtain the convolutional feature vector. The connection layer is used to connect the convolutional feature vectors to obtain a connected convolutional feature vector. The multiple convolutional layers are used to extract features from the concatenated convolutional feature vector to obtain the extracted feature vector; The second fully connected layer is used to predict the extracted feature vectors to obtain the total network energy consumption of the network to be evaluated within the preset future time period.

4. The network energy consumption assessment method according to claim 3, wherein, The plurality of graph convolutional network layers include a first graph convolutional network layer, a second graph convolutional network layer, and a third graph convolutional network layer, and the plurality of convolutional layers include a first convolutional layer, a second convolutional layer, and a second fully connected layer; The first graph convolutional network layer is used to perform convolution processing on the node feature data to obtain the first feature; The second convolutional network layer is used to perform convolution processing on the first feature to obtain the second feature; The third graph convolutional network layer is used to perform convolution processing on the second feature to obtain the third feature; The connection layer is used to connect the first feature, the second feature, and the third feature to obtain the fourth feature; The first convolutional layer is used to extract features from the fourth feature to obtain the fifth feature; The second convolutional layer is used to extract features from the fifth feature to obtain the sixth feature; The second fully connected layer is used to make predictions based on the sixth feature to obtain the total network energy consumption of the network to be evaluated within the preset future time period.

5. The network energy consumption assessment method according to claim 1, wherein, The process of evaluating the network energy consumption of the network to be evaluated within the preset future time period based on the energy consumption of each node and the total network energy consumption includes: Determine the total amount of data and the total number of network nodes in the network to be evaluated; Based on the total data volume and the total network energy consumption, determine the average energy efficiency ratio of the network to be evaluated; Based on the energy consumption of each node, high-energy-consuming nodes in the network to be evaluated are identified, wherein the high-energy-consuming nodes are network nodes whose energy consumption is higher than a preset threshold. Based on the number of high-energy-consuming nodes and the total number of network nodes, the proportion of energy-consuming hotspots in the network to be evaluated is determined; and Based on the total network energy consumption, the average energy efficiency ratio, and the proportion of energy hotspots, the network energy consumption of the network to be evaluated is assessed within the preset future time period.

6. The network energy consumption assessment method according to any one of claims 1-5, wherein, The acquisition of node feature data of the network to be evaluated includes: Obtain the network topology data of the network to be evaluated and the node attribute data of each network node; and The network topology data and the attribute data of each node are preprocessed to obtain the node feature data.

7. The network energy consumption assessment method according to claim 6, wherein the network topology data is a normalized adjacency matrix. The It is obtained after normalizing the adjacency matrix A, wherein the elements A in A are... ij The expression is: The node attribute data mentioned therein is a feature matrix X, and each row of the feature matrix X corresponds to the feature vector of a network node.

8. The network energy consumption assessment method according to claim 7, wherein the node feature data includes the feature matrix X and the normalized adjacency matrix.

9. The network energy consumption assessment method according to any one of claims 1-8, wherein, Before acquiring the node characteristic data and node energy consumption data of the network to be evaluated, the process also includes: Obtain sample node energy consumption data and the corresponding sample node energy consumption; and Based on the energy consumption data of the sample nodes and the energy consumption of the sample nodes corresponding to the energy consumption data of the sample nodes, the first initial model is pre-trained to obtain the first target model.

10. The network energy consumption assessment method according to any one of claims 1-8, wherein, Before acquiring the node characteristic data and node energy consumption data of the network to be evaluated, the process also includes: Obtain the sample node feature data of the sample network, and the total energy consumption of the sample network corresponding to the sample node feature data; and Based on the sample node feature data and the total energy consumption of the sample network corresponding to the sample node feature data, the second initial model is pre-trained to obtain the second target model.

11. A network energy consumption assessment device, comprising: The acquisition module is used to acquire node characteristic data and node energy consumption data of the network to be evaluated, wherein the network to be evaluated includes at least two network nodes; The node energy consumption prediction module is used to input the node energy consumption data into the first target model to obtain the node energy consumption of each network node in a future preset time period output by the first target model. The network energy consumption prediction module is used to input the node feature data into the second target model to obtain the total network energy consumption of the network to be evaluated in the future preset time period output by the second target model. as well as An evaluation module is used to evaluate the network energy consumption of the network to be evaluated within a future preset time period based on the energy consumption of each node and the total network energy consumption.

12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the network energy consumption assessment method as described in any one of claims 1 to 10.

13. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the network energy consumption assessment method as described in any one of claims 1 to 10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the network energy consumption assessment method as described in any one of claims 1 to 10.

Images