Link scheduling strategy determination method and device, computer equipment, medium and product

Abstract

The invention relates to a link scheduling strategy determination method and device, computer equipment, a medium and a product. The method comprises the following steps: acquiring current scheduling state information of each device in a scheduling system of a target power system in a current time period; determining a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents a conflict relationship between different scheduling links in the scheduling system in the current time period, and the target utility function represents satisfaction of the scheduling system in the current time period to different link scheduling strategies; and based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, determining a link scheduling strategy of the scheduling system in a future time period. By adopting the method, the link scheduling efficiency and accuracy of the scheduling system in the power system can be improved.

Description

Technical Field

[0001] The present application relates to the field of smart grid technology, and in particular to a link scheduling strategy determination method, device, computer equipment, medium and product. Background Art

[0002] Power dispatching is an effective management method to ensure the safe and stable operation of the power grid, reliable external power supply, and orderly progress of various power production tasks. As the scale of the power grid continues to grow, the topological structure of the power grid dispatching system becomes more and more complex. Link dispatching based on the power grid system is generally used for tasks such as load forecasting and optimizing power resources.

[0003] Current link scheduling methods are usually implemented using architectures such as parameterized functions and multi-layer neural networks or convolutional neural networks. However, the above methods ignore the consideration of network topology in the power grid's scheduling system and need to be adjusted according to the scheduling scenario, which reduces the efficiency and accuracy of link scheduling. Summary of the invention

[0004] Based on this, it is necessary to provide a link scheduling strategy determination method, device, computer equipment, medium and product that can improve the efficiency and accuracy of link scheduling in the dispatching system in the power system in response to the above technical problems.

[0005] In a first aspect, the present application provides a method for determining a link scheduling strategy, comprising:

[0006] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0007] Determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0008] Based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future period is determined.

[0009] In one embodiment, determining the target link conflict graph and target utility function of the scheduling system according to the current scheduling state information includes:

[0010] Determine, according to the current scheduling state information, association information between the scheduling links in the scheduling system in the current period, and a scheduling amount corresponding to each scheduling link;

[0011] Constructing a target link conflict graph of the scheduling system according to the association information between the scheduling links;

[0012] According to the target link conflict graph and the scheduling amount corresponding to each scheduling link, a target utility function of the scheduling system is constructed.

[0013] In one of the embodiments, the link scheduling prediction model includes a feature extraction network, a decoding network and a solution network;

[0014] The link scheduling prediction model is based on the target link conflict graph and the target utility function, and the link scheduling strategy of the scheduling system in the future period is determined, including:

[0015] Inputting the target link conflict graph into the feature extraction network to obtain conflict features of the target link conflict graph;

[0016] Inputting the conflict feature and the target link conflict graph into the decoding network to obtain a new conflict graph; wherein the new conflict graph is obtained by embedding the conflict feature into the target link conflict graph;

[0017] The new conflict graph and the target utility function are input into the solution network, and the solution network solves the target utility function according to the new conflict graph, and outputs the link scheduling strategy of the scheduling system in the future period.

[0018] In one of the embodiments, the feature extraction network includes a convolutional layer and a wavelet transform layer;

[0019] The step of inputting the target link conflict graph into the feature extraction network to obtain the conflict features of the target link conflict graph includes:

[0020] Inputting the target link conflict graph into the convolution layer to obtain preliminary features of the target link conflict graph output by the convolution layer represented in the frequency domain;

[0021] The preliminary features are input into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0022] In one embodiment, the step of inputting the preliminary features into the wavelet transform layer to obtain the conflict features of the target link conflict graph includes:

[0023] Dividing the preliminary feature into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency;

[0024] respectively determining a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature;

[0025] The first frequency feature, the second frequency feature, the first attention weight, and the second attention weight are input into the wavelet transform layer to obtain the conflict feature of the target link conflict graph.

[0026] In one embodiment, the link scheduling prediction model is constructed in the following manner:

[0027] Acquire a historical link conflict graph and a historical utility function of each device in the scheduling system in a reference period, and a link scheduling strategy of the scheduling system in a comparison period; wherein the comparison period is a period after the reference period;

[0028] Based on the spectral graph neural network, an initial link scheduling prediction model is constructed;

[0029] Inputting the historical link conflict graph and the historical utility function into the initial link scheduling prediction model to obtain the prediction scheduling strategy of the scheduling system in the comparison period;

[0030] Based on the cosine similarity function, constructing a first loss function according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period;

[0031] Using a greedy algorithm, a reward function is determined according to the historical utility function;

[0032] Constructing a second loss function according to the reward function and the prediction scheduling strategy;

[0033] The initial link scheduling prediction model is trained using the first loss function and the second loss function to obtain a link scheduling prediction model.

[0034] In a second aspect, the present application also provides a link scheduling strategy determination device, including:

[0035] An information acquisition module is used to obtain the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0036] A function determination module, configured to determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0037] The strategy determination module is used to determine the link scheduling strategy of the scheduling system in the future period based on the link scheduling prediction model, according to the target link conflict graph and the target utility function.

[0038] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0039] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0040] Determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0041] Based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future period is determined.

[0042] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0043] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0044] Determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0045] Based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future period is determined.

[0046] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0047] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0048] Determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0049] Based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future period is determined.

[0050] The above-mentioned link scheduling strategy determination method, device, computer equipment, medium and product obtain the current scheduling status information of each device in the scheduling system of the target power system in the current time period; determine the target link conflict graph and target utility function of the scheduling system based on the current scheduling status information; since the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current time period, it is equivalent to considering the conflict between each scheduling link, and since the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current time period, it is equivalent to comprehensively considering the scheduling feedback of the entire scheduling system; further, by introducing a link scheduling prediction model, and based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future time period is determined, thereby further ensuring the accuracy and effectiveness of the determined link scheduling strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 A schematic diagram of a flow chart of a method for determining a link scheduling strategy in an embodiment;

[0053] Figure 2 A schematic diagram of a process for determining a target link conflict graph and a target utility function in one embodiment;

[0054] Figure 3 A schematic diagram of a process for determining a link scheduling strategy in an embodiment;

[0055] Figure 4 A schematic diagram of a process for determining conflicting features in one embodiment;

[0056] Figure 5 A schematic diagram of a flow chart for determining conflicting features in another embodiment;

[0057] Figure 6 A schematic diagram of a process for building a link scheduling prediction model in one embodiment;

[0058] Figure 7A schematic diagram of a flow chart of a method for determining a link scheduling strategy in another embodiment;

[0059] Figure 8 A structural block diagram of a link scheduling strategy determination device in one embodiment;

[0060] Fig. 9 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0062] The link scheduling strategy determination method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the link scheduling strategy determination method provided in the embodiment of the present application can be executed by a computer device, which can be a server or a terminal with powerful computing capabilities.

[0063] In one embodiment, Figure 1 As shown, a link scheduling strategy determination method is provided, which is described by taking the method applied to a server as an example, and specifically includes the following steps:

[0064] S101, obtaining current dispatching status information of each device in the dispatching system of the target power system in the current period.

[0065] The scheduling status information of each device is the scheduling status of each device in the scheduling process. The current scheduling status information of each device is the scheduling status information of each device in the current time period.

[0066] Optionally, the current dispatching status information in the current period uploaded by each device may be obtained from the dispatching system of the target power system.

[0067] S102, determining a target link conflict graph and a target utility function of the scheduling system according to current scheduling state information.

[0068] Among them, the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period.

[0069] Optionally, based on the conflict graph construction method in graph theory and according to the current scheduling information, a target link conflict graph representing the conflict relationship between different scheduling links in the scheduling system in the current period can be constructed; at the same time, based on the target utility function construction rule and according to the current scheduling status information, a target utility function representing the satisfaction of the scheduling system with different link scheduling strategies in the current period can be constructed.

[0070] S103, based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, determine the link scheduling strategy of the scheduling system in the future period.

[0071] Among them, the link scheduling prediction model is a model used to predict the link scheduling strategy of the scheduling system in the future period, that is, to find the optimal scheduling link of the scheduling system. In an embodiment of the present application, the link scheduling prediction model is constructed based on a spectral graph neural network, and the link scheduling prediction model includes a feature extraction network, a decoding network, and a solution network, wherein the feature extraction network can be a multi-scale spectral feature extractor, which can be specifically used to extract conflict information in the target link conflict graph; the decoding network can be a decoder for constructing a consistency loss mechanism, which can be specifically used to embed conflict information into the target link conflict graph; the solution network can be a local greedy MWIS (Maximum Weight Independent Set) solver for link scheduling, which can be specifically used to solve the target utility function.

[0072] Optionally, the target link conflict graph and the target utility function can be input into the link scheduling prediction model, so that the link scheduling prediction model solves the target utility function according to the target link conflict graph based on the model parameters, and uses the output of the link scheduling prediction model as the link scheduling strategy of the scheduling system in the future time period.

[0073] In the above-mentioned link scheduling strategy determination method, the current scheduling status information of each device in the scheduling system of the target power system in the current time period is obtained; based on the current scheduling status information, the target link conflict graph and the target utility function of the scheduling system are determined; since the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current time period, it is equivalent to considering the conflict between each scheduling link, and since the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current time period, it is equivalent to comprehensively considering the scheduling feedback of the entire scheduling system; further, by introducing a link scheduling prediction model, and based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future time period is determined, thereby further ensuring the accuracy and effectiveness of the determined link scheduling strategy.

[0074] Optionally, in one embodiment, Figure 2As shown, a method for determining a target link conflict graph and a target utility function is provided, which refines the above S102 and specifically includes the following steps:

[0075] S201, according to the current scheduling state information, determining the association information between the scheduling links in the scheduling system in the current period and the scheduling amount corresponding to each scheduling link.

[0076] The association relationship between the scheduling links includes but is not limited to a conflict relationship.

[0077] Optionally, information representing the conflict relationship between the scheduling links in the current scheduling state information may be extracted as association information between the scheduling links in the scheduling system in the current period; further, the scheduling amount between the scheduling links may be extracted from the current scheduling state information.

[0078] S202: construct a target link conflict graph of the scheduling system according to the association information between the scheduling links.

[0079] Optionally, a target link conflict graph of the scheduling system may be constructed based on information representing the conflict relationship between each scheduling link in the association information between each scheduling link. Specifically, the target link conflict graph consists of nodes and lines between nodes, where each node corresponds to a link in the scheduling system, and lines between nodes represent the mutual influence between corresponding links.

[0080] S203: construct a target utility function of the scheduling system according to the target link conflict graph and the scheduling amount corresponding to each scheduling link.

[0081] Optionally, in order to quantitatively describe the link conflicts in the scheduling system, we can first use a tuple To describe the system state of the scheduling system; among them, represents the target link conflict graph, Indicates the link queue. represents the scheduling amount corresponding to each scheduling link, Represents the overall system state of the scheduling system. At this point, the target utility function of the scheduling system is constructed. Specifically, the target utility function can be expressed as: ,in, Represents a link subset. By constructing the objective utility function, the link scheduling problem is transformed into the problem of finding the largest independent set in graph theory.

[0082] Optionally, consider the target link conflict graph and the target utility function: ,in is the set of nodes in the target link conflict graph, that is, each scheduling link in the scheduling system. The optimal link scheduling problem can be realized by selecting a node itself. Specifically, the optimal link scheduling problem can be expressed as:

[0083] (1)

[0084] In this embodiment, by introducing the association information between the scheduling links in the scheduling system in the current time period and the scheduling amount corresponding to each scheduling link, it is ensured that the constructed target link conflict graph can accurately represent the conflict relationship between different scheduling links in the scheduling system in the current time period, and at the same time, it is ensured that the constructed target utility function can accurately represent the satisfaction of the scheduling system with different link scheduling strategies in the current time period.

[0085] Optionally, the link scheduling prediction model includes a feature extraction network, a decoding network and a solution network; on this basis, in one embodiment, Figure 3 As shown, a method for determining a link scheduling strategy is provided, which refines the above S103 and specifically includes the following steps:

[0086] S301, inputting the target link conflict graph into a feature extraction network to obtain conflict features of the target link conflict graph.

[0087] The conflict feature represents the conflict information of the conflict relationship between the links in the conflict graph.

[0088] Optionally, the target link conflict graph may be input into a feature extraction network so that the feature extraction network processes the conflict graph according to model parameters of the network model, and uses the output of the feature extraction network as the conflict feature of the conflict graph.

[0089] S302, inputting the conflict characteristics and the target link conflict graph into the decoding network to obtain a new conflict graph.

[0090] The new conflict graph is obtained by embedding the conflict features into the conflict graph.

[0091] Optionally, the conflict features and the target link conflict graph can be input into a decoding network so that the decoding network encodes the conflict graph, converts the graph structure of the target link conflict graph into scalar nodes, and embeds the conflict features into the converted scalar nodes. Further, the scalar nodes embedded with the conflict features are converted into a graph structure as a new conflict graph.

[0092] S303, inputting the new conflict graph and the target utility function into the solution network, obtaining the solution network to solve the target utility function according to the new conflict graph, and outputting the link scheduling strategy of the scheduling system in the future period.

[0093] Optionally, the new conflict graph and the target utility function can be input into the solution network, so that the solution network adopts a local greedy algorithm to solve the target utility function according to the new conflict graph, and obtains the node set when the target utility function is maximum, that is, the link set in the new conflict graph. Furthermore, the output of the solution network is used as the link scheduling strategy of the scheduling system in the future period.

[0094] In the embodiment of the present application, by introducing an extraction network, the accuracy of the conflict features in the extracted conflict graph is guaranteed; at the same time, by introducing a decoding network, the conflict features are accurately embedded into the conflict graph; further, by introducing a solution network, the accuracy of the result obtained by solving the target utility function according to the new conflict graph is guaranteed.

[0095] Optionally, the feature extraction network includes a convolution layer and a wavelet transform layer; based on the above embodiment, in one embodiment, Figure 4 As shown, a conflict feature determination method is provided, which refines the above S301 and specifically includes the following steps:

[0096] S401, input the target link conflict graph into the convolution layer to obtain the preliminary features of the conflict graph output by the convolution layer represented in the frequency domain.

[0097] Optionally, the target link conflict graph may be input into a convolutional layer so that the convolutional layer performs feature extraction processing on the conflict graph, and the output of the convolutional layer is used as a preliminary feature of the conflict graph represented in the frequency domain.

[0098] Alternatively, convolutional layers can be constructed from spectral filters. In this case, the adjacency matrix of the conflict graph can be constructed ,in Represents an edge In the embodiment of the present application, the degree matrix is ​​introduced and used Therefore, the normalized adjacency matrix with self-loops is:

[0099] (2)

[0100] in, is the identity matrix.

[0101] Furthermore, spectral convolution can be expressed as:

[0102] (3)

[0103] Therefore, the preliminary features of the convolutional layer output can be expressed as:

[0104] (4)

[0105] in, is the diagonal matrix of the kernel of the graph convolution. For example, the convolution kernel is ,in, is the normalized Laplacian operator The eigenvalues ​​of are sorted in ascending order; is the spectral basis, is the corresponding eigenvector; is a Parameterized diagonal spectral filter.

[0106] S402, inputting the preliminary features into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0107] Optionally, the preliminary features may be input into a wavelet transform layer, so that the wavelet transform layer performs wavelet transform on the preliminary features, and the output of the wavelet transform layer is used as the conflict feature of the conflict graph.

[0108] The embodiment of the present application introduces a convolution layer to accurately and quickly obtain the preliminary features of the target link conflict graph represented in the frequency domain; further, by introducing a wavelet transform layer, the accuracy of the conflict features of the determined conflict graph is guaranteed.

[0109] Optionally, in one embodiment, Figure 5 As shown, a method for determining conflict features is provided, which refines the above S402 and specifically includes the following steps:

[0110] S501, dividing the preliminary features into first frequency features and second frequency features.

[0111] The first frequency is higher than the second frequency. In the embodiment of the present application, the first frequency is defined as a high frequency, and the second frequency is defined as a low frequency; the first frequency feature may be a high frequency feature in the preliminary feature, and the second frequency feature may be a low frequency feature in the preliminary feature.

[0112] Optionally, in order to further capture multi-scale conflict features, the diagonal values ​​of small indices can be classified as low frequency, and the diagonal values ​​of large indices can be classified as high frequency. The spectral basis of corresponds to low-frequency components and high-frequency components. Low-frequency components and high-frequency components convey different structural insights in the new conflict graph. Therefore, the spectral basis is first divided into two groups:

[0113] (5)

[0114] in, and They are low-frequency components and high-frequency components respectively; As a hyperparameter, the threshold between low and high frequencies is defined. and are the weights assigned to low frequencies and high frequencies, respectively, indicating their importance.

[0115] S502, respectively determine a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature.

[0116] Among them, the first attention weight represents the importance of the first frequency feature; the second attention weight represents the importance of the second frequency feature.

[0117] Optionally, a re-parameterization technique may be used to determine a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature.

[0118] (6)

[0119] in, , are the first attention weight and the second attention weight respectively.

[0120] In addition, to ensure and are positive and comparable. In the embodiment of the present application, an attention-based method can be used to normalize them through a softmax function.

[0121] S503, inputting the first frequency feature, the second frequency feature, the first attention weight and the second attention weight into a wavelet transform layer to obtain a conflict feature of a target link conflict graph.

[0122] Optionally, the process of wavelet transforming by the wavelet transform layer can be expressed by the following formula:

[0123] (7)

[0124] in, is the basis for wavelet transform; is the normalized Laplacian operator The characteristic vector of is a scaling matrix whose heat kernel is determined by the hyperparameter to zoom in or out. By simply Replace with Get. Smaller indices in the graph wavelet correspond to low-frequency components, and vice versa. In general, using graph wavelets The base architecture can be expressed as:

[0125] (8)

[0126] Furthermore, the conflict characteristics of the conflict graph can be expressed as:

[0127] (9)

[0128] The embodiment of the present application ensures the accuracy of the determined first attention weight and the second attention weight by introducing the first frequency feature and the second frequency feature; further, through the wavelet transform layer, the accuracy of the conflict feature of the determined target link conflict graph is ensured.

[0129] Optionally, in one embodiment, Figure 6 As shown, a method for constructing a link scheduling prediction model is provided, which specifically includes the following steps:

[0130] S601, obtaining a historical link conflict graph and a historical utility function of each device in the scheduling system in a reference period, and a link scheduling strategy of the scheduling system in a comparison period.

[0131] The comparison period is the period after the reference period.

[0132] S602: Build an initial link scheduling prediction model based on the spectral graph neural network.

[0133] Optionally, an initial link scheduling prediction model can be constructed based on the architecture of the spectral graph neural network.

[0134] S603, inputting the historical link conflict graph and the historical utility function into the initial link scheduling prediction model to obtain the predicted scheduling strategy of the scheduling system in the comparison period.

[0135] S604 , constructing a first loss function based on the cosine similarity function and according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period.

[0136] The cosine similarity function is a function for calculating cosine similarity. The first loss function characterizes the difference between the conflict features associated with the embedding of each node and the actual node conflict features.

[0137] Optionally, the first loss function can be represented by constraining node features so that it is closely aligned with the conflict structure of the graph, thereby effectively guiding the learning process, and its loss function form is mean square error loss. Based on the cosine similarity function, according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period, the process of constructing the first loss function can be expressed by the following formula:

[0138] (10)

[0139] in, is the cosine similarity function; and They represent the node embedding-related conflict characteristics and the actual node conflict characteristics, respectively, which can be determined by the prediction scheduling strategy; Represents the adjacency matrix.

[0140] S605, using a greedy algorithm to determine a reward function based on the historical utility function.

[0141] Optionally, a greedy algorithm is used. According to the historical utility function, the process of determining the reward function can be expressed by the following formula:

[0142] (11)

[0143] in, It is generated by selecting the node with the largest historical utility function through a greedy algorithm.

[0144] S606: Construct a second loss function according to the reward function and the prediction scheduling strategy.

[0145] Optionally, according to the reward function and the prediction scheduling strategy, the process of constructing the second loss function can be expressed by the following formula:

[0146] (12)

[0147] S607: Use the first loss function and the second loss function to train the initial link scheduling prediction model to obtain the link scheduling prediction model.

[0148] Optionally, the sum of the first loss function and the second loss function can be used as the target loss function. Further, the target loss function is used to train the initial link scheduling prediction model, and when the target loss function reaches a preset threshold, a trained link scheduling prediction model is obtained.

[0149] The embodiment of the present application ensures the effectiveness and accuracy of the training process of the initial link scheduling prediction model by introducing the first loss function and the second loss function, thereby ensuring the accuracy and effectiveness of the obtained link scheduling prediction model.

[0150] Figure 7 FIG. 2 is a flow chart of a method for determining a link scheduling strategy in another embodiment. Based on the above embodiment, this embodiment provides an optional example of a method for determining a link scheduling strategy. Figure 7 The specific implementation process is as follows:

[0151] S701, obtaining current dispatching status information of each device in the dispatching system of the target power system in the current period.

[0152] S702, according to the current scheduling state information, determine the association information between the scheduling links in the scheduling system in the current period, and the scheduling amount corresponding to each scheduling link.

[0153] S703: construct a target link conflict graph of the scheduling system according to the association information between the scheduling links.

[0154] S704: construct a target utility function of the scheduling system according to the target link conflict graph and the scheduling amount corresponding to each scheduling link.

[0155] S705, input the target link conflict graph into the convolution layer of the feature extraction network in the link scheduling prediction model to obtain preliminary features of the conflict graph output by the convolution layer represented in the frequency domain.

[0156] S706, input the preliminary features into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0157] Optionally, the preliminary features are divided into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency; a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature are respectively determined; the first frequency feature, the second frequency feature, the first attention weight and the second attention weight are input into a wavelet transform layer to obtain conflict features of the target link conflict graph.

[0158] S707, input the conflict characteristics and the target link conflict graph into the decoding network in the link scheduling prediction model to obtain a new conflict graph.

[0159] The new conflict graph is obtained by embedding the conflict features into the conflict graph.

[0160] S708, input the new conflict graph and the utility function into the solution network in the link scheduling prediction model, obtain the solution network to solve the target utility function according to the new conflict graph, and output the link scheduling strategy of the scheduling system in the future period.

[0161] The specific process of the above S701-S708 can be found in the description of the above method embodiment. The implementation principle and technical effect are similar and will not be repeated here.

[0162] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0163] Based on the same inventive concept, the embodiment of the present application also provides a link scheduling strategy determination device for implementing the link scheduling strategy determination method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more link scheduling strategy determination device embodiments provided below can refer to the limitations of the link scheduling strategy determination method above, and will not be repeated here.

[0164] In an exemplary embodiment, Figure 8 As shown, a link scheduling strategy determination device 800 is provided, comprising: an information acquisition module 810, a function determination module 820 and a strategy determination module 830, wherein:

[0165] The information acquisition module 810 is used to obtain the current dispatching status information of each device in the dispatching system of the target power system in the current period.

[0166] The function determination module 820 is used to determine the target link conflict graph and target utility function of the scheduling system according to the current scheduling status information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period.

[0167] The strategy determination module 830 is used to determine the link scheduling strategy of the scheduling system in the future period based on the link scheduling prediction model, according to the target link conflict graph and the target utility function.

[0168] The link scheduling strategy determination device obtains the current scheduling status information of each device in the scheduling system of the target power system in the current time period; determines the target link conflict graph and target utility function of the scheduling system based on the current scheduling status information; since the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current time period, it is equivalent to considering the conflict between each scheduling link, and since the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current time period, it is equivalent to comprehensively considering the scheduling feedback of the entire scheduling system; further, by introducing a link scheduling prediction model, and based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future time period is determined, thereby further ensuring the accuracy and effectiveness of the determined link scheduling strategy.

[0169] In one embodiment, the function determination module 820 is specifically used to:

[0170] According to the current scheduling status information, the correlation information between the scheduling links in the scheduling system in the current period and the scheduling amount corresponding to each scheduling link are determined; according to the correlation information between the scheduling links, the target link conflict graph of the scheduling system is constructed; according to the target link conflict graph and the scheduling amount corresponding to each scheduling link, the target utility function of the scheduling system is constructed.

[0171] In one embodiment, the link scheduling prediction model includes a feature extraction network, a decoding network, and a solution network; the strategy determination module 830 includes:

[0172] The feature extraction unit is used to input the target link conflict graph into the feature extraction network to obtain the conflict features of the target link conflict graph.

[0173] The decoding unit is used to input the conflict feature and the target link conflict graph into the decoding network to obtain a new conflict graph; wherein the new conflict graph is obtained by embedding the conflict feature into the conflict graph.

[0174] The function solving unit is used to input the new conflict graph and the utility function into the solving network, so that the solving network solves the target utility function according to the new conflict graph and outputs the link scheduling strategy of the scheduling system in the future period.

[0175] In one embodiment, the feature extraction network includes a convolution layer and a wavelet transform layer; the feature extraction unit includes:

[0176] The first extraction subunit is used to input the target link conflict graph into the convolution layer to obtain the preliminary features of the conflict graph output by the convolution layer represented in the frequency domain.

[0177] The second extraction subunit is used to input the preliminary features into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0178] In one embodiment, the second extraction subunit is specifically used for:

[0179] The preliminary features are divided into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency; a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature are determined respectively; the first frequency feature, the second frequency feature, the first attention weight and the second attention weight are input into a wavelet transform layer to obtain conflict features of a target link conflict graph.

[0180] In one embodiment, the link scheduling strategy determining device 800 is further configured to:

[0181] Obtain the historical link conflict graph and historical utility function of each device in the scheduling system in the reference period, as well as the link scheduling strategy of the scheduling system in the comparison period; wherein the comparison period is the period after the reference period; construct an initial link scheduling prediction model based on the spectral graph neural network;

[0182] The historical link conflict graph and the historical utility function are input into the initial link scheduling prediction model to obtain the predicted scheduling strategy of the scheduling system in the comparison period; based on the cosine similarity function, the first loss function is constructed according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period; the reward function is determined according to the historical utility function using a greedy algorithm; the second loss function is constructed according to the reward function and the predicted scheduling strategy; the initial link scheduling prediction model is trained using the first loss function and the second loss function to obtain the link scheduling prediction model.

[0183] Each module in the above-mentioned link scheduling strategy determination device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0184] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig. 9As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a link scheduling strategy determination method is implemented.

[0185] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0186] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0187] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0188] According to the current scheduling state information, determine the target link conflict graph and target utility function of the scheduling system; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0189] Based on the link scheduling prediction model, the link scheduling strategy of the scheduling system in the future period is determined according to the target link conflict graph and the target utility function.

[0190] In one embodiment, when the processor executes the computer program to determine the target link conflict graph and the target utility function of the scheduling system according to the current scheduling state information, the processor further implements the following steps:

[0191] According to the current scheduling status information, the correlation information between the scheduling links in the scheduling system in the current period and the scheduling amount corresponding to each scheduling link are determined; according to the correlation information between the scheduling links, the target link conflict graph of the scheduling system is constructed; according to the target link conflict graph and the scheduling amount corresponding to each scheduling link, the target utility function of the scheduling system is constructed.

[0192] In one embodiment, the link scheduling prediction model includes a feature extraction network, a decoding network, and a solution network; when the processor executes the computer program to determine the link scheduling strategy of the scheduling system in the future period based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the processor also implements the following steps:

[0193] The target link conflict graph is input into the feature extraction network to obtain the conflict features of the target link conflict graph; the conflict features and the target link conflict graph are input into the decoding network to obtain a new conflict graph; wherein the new conflict graph is obtained by embedding the conflict features into the conflict graph; the new conflict graph and the utility function are input into the solution network to obtain the solution network to solve the target utility function according to the new conflict graph, and the output is the link scheduling strategy of the scheduling system in the future period.

[0194] In one embodiment, the feature extraction network includes a convolution layer and a wavelet transform layer; when the processor executes a computer program to input the target link conflict graph into the feature extraction network and obtains the conflict features of the target link conflict graph, the following steps are also implemented:

[0195] The target link conflict graph is input into the convolution layer to obtain the preliminary features of the conflict graph output by the convolution layer in the frequency domain; the preliminary features are input into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0196] In one embodiment, when the processor executes the computer program to input the preliminary features into the wavelet transform layer and obtains the conflict features of the target link conflict graph, the following steps are also implemented:

[0197] The preliminary features are divided into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency; a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature are determined respectively; the first frequency feature, the second frequency feature, the first attention weight and the second attention weight are input into a wavelet transform layer to obtain conflict features of a target link conflict graph.

[0198] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0199] Obtain the historical link conflict graph and historical utility function of each device in the scheduling system in the reference period, as well as the link scheduling strategy of the scheduling system in the comparison period; wherein the comparison period is the period after the reference period; construct an initial link scheduling prediction model based on the spectral graph neural network; input the historical link conflict graph and the historical utility function into the initial link scheduling prediction model to obtain the predicted scheduling strategy of the scheduling system in the comparison period; based on the cosine similarity function, construct a first loss function according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period; adopt a greedy algorithm to determine the reward function according to the historical utility function; construct a second loss function according to the reward function and the predicted scheduling strategy; adopt the first loss function and the second loss function to train the initial link scheduling prediction model to obtain the link scheduling prediction model.

[0200] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0201] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0202] According to the current scheduling state information, determine the target link conflict graph and target utility function of the scheduling system; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0203] Based on the link scheduling prediction model, the link scheduling strategy of the scheduling system in the future period is determined according to the target link conflict graph and the target utility function.

[0204] In one embodiment, when the processor executes the computer program to determine the target link conflict graph and the target utility function of the scheduling system according to the current scheduling state information, the processor further implements the following steps:

[0205] According to the current scheduling status information, the correlation information between the scheduling links in the scheduling system in the current period and the scheduling amount corresponding to each scheduling link are determined; according to the correlation information between the scheduling links, the target link conflict graph of the scheduling system is constructed; according to the target link conflict graph and the scheduling amount corresponding to each scheduling link, the target utility function of the scheduling system is constructed.

[0206] In one embodiment, the link scheduling prediction model includes a feature extraction network, a decoding network, and a solution network; when the processor executes the computer program to determine the link scheduling strategy of the scheduling system in the future period based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the processor also implements the following steps:

[0207] The target link conflict graph is input into the feature extraction network to obtain the conflict features of the target link conflict graph; the conflict features and the target link conflict graph are input into the decoding network to obtain a new conflict graph; wherein the new conflict graph is obtained by embedding the conflict features into the conflict graph; the new conflict graph and the utility function are input into the solution network to obtain the solution network to solve the target utility function according to the new conflict graph, and the output is the link scheduling strategy of the scheduling system in the future period.

[0208] In one embodiment, the feature extraction network includes a convolution layer and a wavelet transform layer; when the processor executes a computer program to input the target link conflict graph into the feature extraction network and obtains the conflict features of the target link conflict graph, the following steps are also implemented:

[0209] The target link conflict graph is input into the convolution layer to obtain the preliminary features of the conflict graph output by the convolution layer in the frequency domain; the preliminary features are input into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0210] In one embodiment, when the processor executes the computer program to input the preliminary features into the wavelet transform layer and obtains the conflict features of the target link conflict graph, the following steps are also implemented:

[0211] The preliminary features are divided into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency; a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature are determined respectively; the first frequency feature, the second frequency feature, the first attention weight and the second attention weight are input into a wavelet transform layer to obtain conflict features of a target link conflict graph.

[0212] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0213] Obtain the historical link conflict graph and historical utility function of each device in the scheduling system in the reference period, as well as the link scheduling strategy of the scheduling system in the comparison period; wherein the comparison period is the period after the reference period; construct an initial link scheduling prediction model based on the spectral graph neural network; input the historical link conflict graph and the historical utility function into the initial link scheduling prediction model to obtain the predicted scheduling strategy of the scheduling system in the comparison period; based on the cosine similarity function, construct a first loss function according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period; adopt a greedy algorithm to determine the reward function according to the historical utility function; construct a second loss function according to the reward function and the predicted scheduling strategy; adopt the first loss function and the second loss function to train the initial link scheduling prediction model to obtain the link scheduling prediction model.

[0214] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0215] Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period;

[0216] According to the current scheduling state information, determine the target link conflict graph and target utility function of the scheduling system; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period;

[0217] Based on the link scheduling prediction model, the link scheduling strategy of the scheduling system in the future period is determined according to the target link conflict graph and the target utility function.

[0218] In one embodiment, when the processor executes the computer program to determine the target link conflict graph and the target utility function of the scheduling system according to the current scheduling state information, the processor further implements the following steps:

[0219] According to the current scheduling status information, the correlation information between the scheduling links in the scheduling system in the current period and the scheduling amount corresponding to each scheduling link are determined; according to the correlation information between the scheduling links, the target link conflict graph of the scheduling system is constructed; according to the target link conflict graph and the scheduling amount corresponding to each scheduling link, the target utility function of the scheduling system is constructed.

[0220] In one embodiment, the link scheduling prediction model includes a feature extraction network, a decoding network, and a solution network; when the processor executes the computer program to determine the link scheduling strategy of the scheduling system in the future period based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the processor also implements the following steps:

[0221] The target link conflict graph is input into the feature extraction network to obtain the conflict features of the target link conflict graph; the conflict features and the target link conflict graph are input into the decoding network to obtain a new conflict graph; wherein the new conflict graph is obtained by embedding the conflict features into the conflict graph; the new conflict graph and the utility function are input into the solution network to obtain the solution network to solve the target utility function according to the new conflict graph, and the output is the link scheduling strategy of the scheduling system in the future period.

[0222] In one embodiment, the feature extraction network includes a convolution layer and a wavelet transform layer; when the processor executes a computer program to input the target link conflict graph into the feature extraction network and obtains the conflict features of the target link conflict graph, the following steps are also implemented:

[0223] The target link conflict graph is input into the convolution layer to obtain the preliminary features of the conflict graph output by the convolution layer in the frequency domain; the preliminary features are input into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

[0224] In one embodiment, when the processor executes the computer program to input the preliminary features into the wavelet transform layer and obtains the conflict features of the target link conflict graph, the following steps are also implemented:

[0225] The preliminary features are divided into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency; a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature are determined respectively; the first frequency feature, the second frequency feature, the first attention weight and the second attention weight are input into a wavelet transform layer to obtain conflict features of a target link conflict graph.

[0226] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0227] Obtain the historical link conflict graph and historical utility function of each device in the scheduling system in the reference period, as well as the link scheduling strategy of the scheduling system in the comparison period; wherein the comparison period is the period after the reference period; construct an initial link scheduling prediction model based on the spectral graph neural network; input the historical link conflict graph and the historical utility function into the initial link scheduling prediction model to obtain the predicted scheduling strategy of the scheduling system in the comparison period; based on the cosine similarity function, construct a first loss function according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period; adopt a greedy algorithm to determine the reward function according to the historical utility function; construct a second loss function according to the reward function and the predicted scheduling strategy; adopt the first loss function and the second loss function to train the initial link scheduling prediction model to obtain the link scheduling prediction model.

[0228] It should be noted that the data involved in this application (including but not limited to data used for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0229] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0230] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0231] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the present application. It should be noted that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for determining a link scheduling strategy, characterized in that: The method comprises: Obtaining the current dispatching status information of each device in the dispatching system of the target power system in the current period; Determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period; Based on the link scheduling prediction model, according to the target link conflict graph and the target utility function, the link scheduling strategy of the scheduling system in the future period is determined.

2. The method according to claim 1, characterized in that Determining the target link conflict graph and target utility function of the scheduling system according to the current scheduling state information includes: Determine, according to the current scheduling state information, association information between the scheduling links in the scheduling system in the current period, and a scheduling amount corresponding to each scheduling link; Constructing a target link conflict graph of the scheduling system according to the association information between the scheduling links; According to the target link conflict graph and the scheduling amount corresponding to each scheduling link, a target utility function of the scheduling system is constructed.

3. The method according to claim 1, characterized in that The link scheduling prediction model includes a feature extraction network, a decoding network and a solution network; The link scheduling prediction model is based on the target link conflict graph and the target utility function, and the link scheduling strategy of the scheduling system in the future period is determined, including: Inputting the target link conflict graph into the feature extraction network to obtain conflict features of the target link conflict graph; Inputting the conflict feature and the target link conflict graph into the decoding network to obtain a new conflict graph; wherein the new conflict graph is obtained by embedding the conflict feature into the target link conflict graph; The new conflict graph and the target utility function are input into the solution network, and the solution network solves the target utility function according to the new conflict graph, and outputs the link scheduling strategy of the scheduling system in the future period.

4. The method according to claim 3, characterized in that The feature extraction network includes a convolution layer and a wavelet transform layer; The step of inputting the target link conflict graph into the feature extraction network to obtain the conflict features of the target link conflict graph includes: Inputting the target link conflict graph into the convolution layer to obtain preliminary features of the target link conflict graph output by the convolution layer represented in the frequency domain; The preliminary features are input into the wavelet transform layer to obtain the conflict features of the target link conflict graph.

5. The method according to claim 4, characterized in that The step of inputting the preliminary features into the wavelet transform layer to obtain the conflict features of the target link conflict graph includes: Dividing the preliminary feature into a first frequency feature and a second frequency feature; wherein the first frequency is higher than the second frequency; respectively determining a first attention weight corresponding to the first frequency feature and a second attention weight corresponding to the second frequency feature; The first frequency feature, the second frequency feature, the first attention weight, and the second attention weight are input into the wavelet transform layer to obtain the conflict feature of the target link conflict graph.

6. The method according to claim 1, characterized in that The link scheduling prediction model is constructed in the following way: Acquire a historical link conflict graph and a historical utility function of each device in the scheduling system in a reference period, and a link scheduling strategy of the scheduling system in a comparison period; wherein the comparison period is a period after the reference period; Based on the spectral graph neural network, an initial link scheduling prediction model is constructed; Inputting the historical link conflict graph and the historical utility function into the initial link scheduling prediction model to obtain the prediction scheduling strategy of the scheduling system in the comparison period; Based on the cosine similarity function, constructing a first loss function according to the predicted scheduling strategy and the link scheduling strategy of the scheduling system in the comparison period; Using a greedy algorithm, a reward function is determined according to the historical utility function; Constructing a second loss function according to the reward function and the prediction scheduling strategy; The initial link scheduling prediction model is trained using the first loss function and the second loss function to obtain a link scheduling prediction model.

7. A link scheduling strategy determination device, characterized in that: The device comprises: An information acquisition module is used to obtain the current dispatching status information of each device in the dispatching system of the target power system in the current period; A function determination module, configured to determine a target link conflict graph and a target utility function of the scheduling system according to the current scheduling state information; wherein the target link conflict graph represents the conflict relationship between different scheduling links in the scheduling system in the current period, and the target utility function represents the satisfaction of the scheduling system with different link scheduling strategies in the current period; The strategy determination module is used to determine the link scheduling strategy of the scheduling system in the future period based on the link scheduling prediction model, according to the target link conflict graph and the target utility function.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images