Correlation-based industrial chain power consumption data analysis and prediction method and device

Abstract

The invention discloses a correlation-based industrial chain power consumption data analysis and prediction method and device. The industrial chain power consumption data analysis and prediction method based on correlation comprises the following steps: determining key industrial chain related enterprises of a target area; associating the enterprises with the transaction relationship, and constructing a graph structure corresponding to the key industrial chain; predicting a first electricity load of the key industrial chain at the next moment through a graph neural network model according to the graph structure; and determining a power dispatching strategy corresponding to the key industrial chain according to the first electricity load.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a method and device for analyzing and predicting electricity consumption data of an industrial chain based on correlation. Background Art

[0002] At present, in order to effectively save electricity resources and avoid unnecessary waste, the general approach is to dispatch electricity by region. This approach mainly focuses on allocating electricity according to the actual electricity consumption of each region, that is, regions with greater electricity demand will receive more electricity supply. However, in the process of implementing this strategy, we found that there are often both key and non-key industrial chains in each region. Since key industrial chains have a more urgent need for electricity and are more important, they are usually given a higher priority for power dispatch. Nevertheless, the power dispatch model that relies solely on regional divisions has exposed certain limitations, especially in balancing the electricity demand between key and non-key industrial chains.

[0003] In order to solve the above situation, the existing technology attempts to solve the contradiction between supply and demand by allocating power resources specifically for key and non-key industrial chains, and strives to fully cover the power demand of various industrial chains. At this stage, this kind of resource allocation work mostly relies on traditional empirical judgment and analysis methods based on simulation models. Although these methods have played an important role in the past, with the development of the times and the advancement of technology, their accuracy and reliability have gradually shown their shortcomings. Specifically, empirical judgments are easily affected by subjective factors, while simulation models may lead to prediction deviations due to the inconsistency between the assumed conditions and the actual situation. Therefore, these traditional methods have become increasingly difficult to adapt to the complex and changeable modern power demand environment, especially when faced with the growing requirements for high-precision and high-reliability power services, their shortcomings have become increasingly obvious.

[0004] With respect to the technical problem that the power resource dispatching scheme in the above-mentioned prior art cannot accurately dispatch different industrial chains, no effective solution has been proposed so far. Summary of the invention

[0005] The embodiments of the present application provide a method and device for analyzing and predicting electricity consumption data of an industrial chain based on correlation, so as to at least solve the technical problem that the power resource scheduling scheme existing in the prior art cannot accurately schedule different industrial chains.

[0006] According to one aspect of an embodiment of the present application, a correlation-based method for analyzing and predicting electricity consumption data of an industrial chain is provided, including: determining key industrial chain-related enterprises in a target area; associating enterprises with transaction relationships to construct a graph structure corresponding to the key industrial chain; predicting the first electricity load of the key industrial chain at the next moment based on the graph structure through a graph neural network model; and determining a power dispatching strategy corresponding to the key industrial chain based on the first electricity load.

[0007] According to another aspect of the embodiment of the present application, a correlation-based industrial chain electricity consumption data analysis and prediction device is also provided, including: an enterprise determination module, used to determine key industrial chain-related enterprises in a target area; a graph structure construction module, used to associate enterprises with transaction relationships and construct a graph structure corresponding to the key industrial chain; a load prediction module, used to predict the first power load of the key industrial chain at the next moment according to the graph structure through a graph neural network model; and a strategy determination module, used to determine the power dispatching strategy corresponding to the key industrial chain according to the first power load.

[0008] According to another aspect of the embodiment of the present application, a correlation-based industrial chain electricity consumption data analysis and prediction device is also provided, including: a processor; and a memory, connected to the processor, for providing the processor with instructions for processing the following processing steps: determining key industrial chain-related enterprises in the target area; associating enterprises with transaction relationships to construct a graph structure corresponding to the key industrial chain; predicting the first power load of the key industrial chain at the next moment based on the graph structure through a graph neural network model; and determining the power dispatching strategy corresponding to the key industrial chain based on the first power load.

[0009] According to another aspect of an embodiment of the present application, a computer system is also provided, including a memory, a processor, and a computer program stored in the memory, and the processor executes the computer program to implement the steps of the above method.

[0010] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above method are implemented.

[0011] According to another aspect of the embodiments of the present application, a computer program product is also provided, including a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0012] In the embodiment of the present application, the data processing platform constructs a detailed graph structure by accurately identifying and associating the key industrial chain and its upstream and downstream enterprises in the target area, so that the upstream and downstream enterprises can be associated and the correlation between the characteristic information of the enterprises can be accurately determined. In addition, the technical solution realizes the dynamic prediction of the specific electricity demand of the key industrial chain at the future moment through the graph neural network model, which greatly improves the accuracy and timeliness of the prediction. This prediction method based on real-time data analysis and deep learning overcomes the prediction deviation caused by static assumptions and human intervention in traditional methods, and ensures the efficient and accurate allocation of power resources. In addition, the technical solution realizes the comprehensive consideration and intelligent scheduling of the power demand of the entire region by integrating the power demand forecast of the key industrial chain and the non-key industrial chain, and enhances the overall stability and response speed of the power system. Especially in the face of emergencies or demand fluctuations, based on the dynamic adjustment ability of the graph neural network, it can respond quickly to ensure the continuity of the key industrial chain and the basic power supply of the non-key industrial chain, and improve the adaptability of power dispatching. And then solves the technical problem that the power resource dispatching scheme existing in the prior art cannot accurately dispatch different industrial chains. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0014] Figure 1 is a hardware structure block diagram of a computing device for implementing the method according to Embodiment 1 of the present application;

[0015] Figure 2 is a schematic diagram of the power dispatching system according to Embodiment 1 of the present application;

[0016] Figure 3 It is a flowchart of the method for analyzing and predicting industrial chain electricity consumption data based on correlation according to the first aspect of Example 1 of the present application;

[0017] Figure 4 is a schematic diagram of a graph structure according to Example 1 of the present application;

[0018] Figure 5 is a schematic diagram of a graph neural network model according to Example 1 of the present application;

[0019] Figure 6 It is a flowchart of the method for analyzing and predicting industrial chain electricity consumption data based on correlation according to the first aspect of Example 1 of the present application;

[0020] Figure 7is a schematic diagram of an industrial chain electricity consumption data analysis and prediction device based on correlation according to Example 2 of the present application; and

[0021] Figure 8 It is a schematic diagram of the correlation-based industrial chain electricity consumption data analysis and prediction device described in Example 3 of the present application. DETAILED DESCRIPTION

[0022] In order to enable those skilled in the art to better understand the technical solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only embodiments of a part of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] Example 1

[0025] According to this embodiment, a method embodiment of an industrial chain electricity consumption data analysis and prediction method based on correlation is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0026] The method embodiment provided in this embodiment can be executed in a mobile terminal, a computer terminal, a server or a similar computing device. Figure 1 The hardware structure block diagram of a computing device for implementing a correlation-based industrial chain electricity consumption data analysis and prediction method is shown. Figure 1As shown, the computing device may include one or more processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory for storing data, and a transmission device for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0027] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuitry may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computing device. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0028] The memory can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the industrial chain electricity consumption data analysis and prediction method based on correlation in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, the industrial chain electricity consumption data analysis and prediction method based on correlation of the above-mentioned application is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computing device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0029] The transmission device is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computing device. In one example, the transmission device includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (Radio Frequency, RF) module, which is used to communicate with the Internet wirelessly.

[0030] The display may be, for example, a touch screen liquid crystal display (LCD) that may enable a user to interact with a user interface of the computing device.

[0031] It should be noted that, in some optional embodiments, the above Figure 1 The computing device shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware elements and software elements. Figure 1 This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the computing devices described above.

[0032] Figure 2 Schematic diagram of the power dispatching system according to this embodiment. Figure 2 As shown, the system includes: a data processing platform 10.

[0033] refer to Figure 2 As shown, the technical solution can be used to implement the dispatching of a regional power grid, specifically, the regional power grid can be, for example, a power grid in a city area. The regional power grid is powered by power generation equipment of different energy types, such as photovoltaic equipment, wind power equipment, energy storage equipment and thermal power units.

[0034] The data processing platform 10 can communicate with photovoltaic equipment, wind power equipment, energy storage equipment, thermal power units and power loads through the Internet of Things, so as to obtain the power supply power of photovoltaic equipment, wind power equipment, energy storage equipment and thermal power units, as well as the power consumption of power loads. Power loads can include residential power, industrial power, commercial power, agricultural power, and transportation power, etc.

[0035] The data processing platform 10 can also achieve a balance between the power consumption and power supply of the power load of the regional power grid through power dispatching. Among them, compared with photovoltaic equipment and wind power equipment, thermal power units are relatively stable and controllable. In addition, energy storage devices can be charged during flat and valley periods and discharged during peak and peak periods, thereby realizing power dispatching in a "peak shaving and valley filling" manner. In addition, when the power supply of the power supply equipment of the regional power grid cannot meet the power load, electricity can be purchased from the national power grid. The electricity price of the national power grid also varies according to the peak and valley, with the valley price being lower and the peak price being higher.

[0036] Among them, the power load can be divided into industrial, civil, commercial and other categories. In this embodiment, the power load is simply divided into two categories: key industrial chain power and non-key industrial chain power. Among them, non-key industrial chain power is the remaining power after excluding enterprises related to the key industrial chain, including civil, industrial and commercial power.

[0037] It should be noted that the data processing platform 10 in the system is applicable to the hardware structure described above.

[0038] Under the above operating environment, according to the first aspect of this embodiment, a method for analyzing and predicting industrial chain electricity consumption data based on correlation is provided. The method comprises: Figure 2 The data processing platform 10 shown in FIG. Figure 3 A schematic diagram showing the process of the method is shown in FIG. Figure 3 As shown, the method includes:

[0039] S302: Identify key industry chain related enterprises in the target area;

[0040] S304: Associating enterprises with transaction relationships to construct a graph structure corresponding to the key industrial chain;

[0041] S306: predicting the first power load of the key industrial chain at the next moment according to the graph structure through the graph neural network model; and

[0042] S308: Determine a power dispatching strategy corresponding to the key industrial chain according to the first power load.

[0043] Specifically, the data processing platform 10 determines the key industrial chain corresponding to the target area and the enterprises related to the key industrial chain by crawling the enterprise information and industry information related to the target area on the Internet. For example, the key industrial chain of the target area includes key industrial chain 1 and key industrial chain 2, wherein key industrial chain 1 can be, for example, new energy vehicles, and industrial chain 2 can be, for example, real estate. After that, the data processing platform 10 classifies the related enterprises of each key industrial chain into upstream enterprises, midstream enterprises and downstream enterprises. Taking the key industrial chain corresponding to new energy vehicles as an example, upstream enterprises can be enterprises related to the production of raw materials (such as battery materials), midstream enterprises can be enterprises related to the production of parts (batteries, motors and electronic controls, etc.), and downstream enterprises can be enterprises related to the production of complete vehicle manufacturing or charging facilities.

[0044] Further, the data processing platform 10 determines the upstream enterprises and midstream enterprises with transaction relationships, and determines the midstream enterprises and downstream enterprises with transaction relationships. Then, the data processing platform 10 associates the upstream enterprises with the midstream enterprises with transaction relationships, and associates the midstream enterprises with the downstream enterprises with transaction relationships, and constructs a graph structure corresponding to the key industrial chain. One key industrial chain corresponds to one graph structure.

[0045] For example, the upstream enterprises of key industry chain 1 include E 1 ~E 4 , midstream companies include E 5 ~E 7 , downstream enterprises include E 8 and E 9 .

[0046] Among them, the relationship between upstream and midstream enterprises: E 1 Respectively with E 5 and E 6 have a transaction relationship, the data processing platform 10 will 1 Respectively with E 5 and E 6 Make an association; E 2 Respectively with E 5 and E 6 have a transaction relationship, the data processing platform 10 will 2 Respectively with E 5 and E 6 Make an association; E 3 and E 4 Both with E 7 have a transaction relationship, the data processing platform 10 will 3 and E 4 Both with E 7 association.

[0047] For the relationship between midstream and downstream enterprises: E 6 With E 8 If there is a transaction relationship, the data processing platform 10 will 5 With E 8 Make an association; E 6 Respectively with E 8 and E 9 have a transaction relationship, the data processing platform 10 will 6 Respectively with E 8 and E 9 Make an association; E 7 With E 9 If there is a transaction relationship, the data processing platform 10 will 7 With E 9 Make an association.

[0048] Thus, the data processing platform 10 converts the enterprise E 1 ~E 9 As a graph node N 1 ~N 9 , and according to Enterprise E 1 ~E 9 The relationship between them is constructed as Figure 4 The graph structure shown.

[0049] Furthermore, the data processing platform 10 is provided with a pre-trained graph neural network model. The graph neural network model is used to predict the power load of each key industrial chain at the next moment. Thus, the data processing platform 10 inputs the graph structure corresponding to the key industrial chain into the graph neural network model, processes the graph structure through the graph neural network model, and predicts the power load (i.e., the first power load) of each key industrial chain at the next moment.

[0050] Furthermore, the data processing platform 10 determines the non-critical industrial chains in the target area except the critical industrial chains. Then, the data processing platform predicts the power load of the non-critical industrial chain at the next moment according to the preset power prediction model and the power load of the non-critical industrial chain in the preset period. The power prediction model can be, for example, a pre-trained neural network model.

[0051] Furthermore, the data processing platform 10 determines a power dispatching strategy for the key industrial chain and the non-key industrial chain based on the first power load corresponding to the key industrial chain and the predicted power load corresponding to the non-key industrial chain, and dispatches power resources for enterprises in the key industrial chain and the non-key industrial chain through the power dispatching strategy.

[0052] As described in the background technology, the existing technology attempts to solve the contradiction between supply and demand by allocating power resources specifically for key and non-key industrial chains, and strives to fully cover the power demand of various industrial chains. At this stage, this type of resource allocation work mostly relies on traditional empirical judgments and analysis methods based on simulation models. Although these methods have played an important role in the past, with the development of the times and the advancement of technology, their accuracy and reliability have gradually shown their shortcomings. Specifically, empirical judgments are easily affected by subjective factors, while simulation models may cause prediction deviations due to the inconsistency between the assumed conditions and the actual situation. Therefore, these traditional methods have become increasingly difficult to adapt to the modern complex and changeable power demand environment, especially when faced with the growing requirements for high-precision and high-reliability power services, their shortcomings have become increasingly obvious.

[0053] In view of the technical problems described above, through the technical solutions of the embodiments of the present application, the data processing platform accurately identifies and associates the key industrial chains and their upstream and downstream enterprises in the target area, and constructs a detailed graph structure, so that the upstream and downstream enterprises can be associated, and the correlation between the characteristic information of the enterprises can be accurately determined. In addition, the technical solution realizes the dynamic prediction of the specific electricity demand of the key industrial chain at the future moment through the graph neural network model, which greatly improves the accuracy and timeliness of the prediction. This prediction method based on real-time data analysis and deep learning overcomes the prediction deviation caused by static assumptions and human intervention in traditional methods, and ensures the efficient and accurate allocation of power resources. In addition, the technical solution realizes the comprehensive consideration and intelligent scheduling of the power demand of the entire region by integrating the power demand forecast of the key industrial chain and the non-key industrial chain, and enhances the overall stability and response speed of the power system. Especially in the face of emergencies or demand fluctuations, based on the dynamic adjustment capability of the graph neural network, it can respond quickly to ensure the continuity of the key industrial chain and the basic power supply of the non-key industrial chain, and improve the adaptability of power dispatching. Thus, the technical problem that the power resource dispatching scheme existing in the prior art cannot be accurately dispatched for different industrial chains is solved.

[0054] Optionally, enterprises with transaction relationships are associated to construct a graph structure corresponding to the key industrial chain, including: obtaining electricity consumption information related to the enterprise as the attribute of the corresponding graph node, wherein the electricity consumption information includes current load, current load variance, long-term average load, long-term average load variance and temperature; calculating the electricity consumption correlation between enterprises with transaction relationships, and using the electricity consumption correlation as the weight of the connection edges between graph nodes; and connecting the graph nodes through connection edges with corresponding weights to construct a graph structure.

[0055] Specifically, taking the key industrial chain 1 as an example, the data processing platform 10 obtains the information related to the enterprise E 1 ~E 9The relevant electricity consumption information includes current load, current load variance, long-term average load, long-term average load variance and temperature. i The power usage information includes the current load i,1 , current load variance en i,2 , long-term average load i,3 , long-term average load variance i,4 and temperature i,5 As the corresponding graph node N i Attributes and generate graph node N i Vector EN i =[en i,1 ,en i,2 ,en i,3 ,en i,4 ,en i,5 ] T , i=1~9.

[0056] Furthermore, the data processing platform 10 obtains the historical electricity load of each enterprise, calculates the electricity consumption correlation between enterprises with transaction relationships based on the historical electricity load, and uses the electricity consumption correlation as the weight of the connection edges between graph nodes, and then connects the graph nodes through connection edges with corresponding weights to construct a graph structure.

[0057] For example, Enterprise E 1 and E 5 If there is a transaction relationship, the data processing platform 10 will 1 and E 5 The historical electricity load of the enterprise E 1 and E 5 When constructing the graph structure corresponding to the key industrial chain 1, the enterprise E 1 and E 5 As a graph node N 1 and N 5 , and connect them through connecting edges, and connect enterprise E 1 and E 5 The electricity consumption correlation is used as the graph node N 1 and N 5 The weight of the connecting edge.

[0058] Thus, the data processing platform 10 will be connected with the enterprise E having a transaction relationship according to the above 1 and E 5 The corresponding graph node N 1 and N 5 By connecting with weighted edges, other enterprises with transaction relationships are taken as graph nodes and connected in the same way, so that according to the graph node N 1 ~N9 And the corresponding connecting edges to build the graph structure (such as Figure 4 shown).

[0059] Therefore, this technical solution cleverly quantifies the degree of mutual influence between enterprises by calculating the electricity consumption correlation as the weight of the connecting edges between graph nodes, so that the constructed graph structure can not only reflect the static transaction relationship, but also dynamically reflect the linkage of electricity demand, enhance the prediction ability of the model, and thus improve the accuracy and efficiency of power dispatch.

[0060] Optionally, the operation of calculating the electricity consumption correlation between enterprises with a transaction relationship and using the electricity consumption correlation as the weight of the connecting edges between graph nodes includes: determining the second electricity load of the enterprise at each moment within a preset period; and determining the electricity consumption correlation between the enterprises according to the second electricity load at each moment through a preset correlation calculation formula.

[0061] Specifically, the data processing platform 10 obtains the historical power load (i.e., the second power load) of the enterprise at each moment in a preset period. The preset period is h moments before the current moment. i The second electrical load S at the kth moment of the preset cycle i,k . Where k = 1 ~ h.

[0062] Furthermore, the data processing platform 10 calculates the enterprise E in the preset period according to the correlation calculation formula. x and Enterprise E y The power consumption correlation between x,y The correlation calculation formula is:

[0063]

[0064] For example, the data processing platform 10 calculates the enterprise E with transaction relationship according to the correlation calculation formula. 1 and E 5 Electricity usage correlation:

[0065]

[0066] Thus, the data processing platform 10 calculates the electricity consumption correlation between enterprises having transaction relationships according to the correlation calculation formula.

[0067] Therefore, this technical solution uses the calculated electricity consumption correlation as the weight of the connection edge between graph nodes, so that the constructed graph structure not only has the function of describing transaction relationships, but also can reflect the mutual influence of electricity demand between enterprises. Such a graph structure provides information-rich input for the graph neural network model, enabling it to more accurately predict the distribution of electricity demand in the future.

[0068] Optionally, the operation of predicting the first power load of the key industrial chain at the next moment is performed through a graph neural network model according to the graph structure, including: determining the power probability value of the graph node according to the graph structure through a graph neural network model, wherein the power probability value is used to predict the power load range of each enterprise in the key industrial chain at the next moment; and calculating the first power load of the key industrial chain according to the center value of the power load range and the corresponding power probability value by using a weighted summation method.

[0069] Specifically, the data processing platform 10 inputs the graph structure corresponding to the key industrial chain into the preset graph neural network model, processes the graph structure through the graph neural network model, and thus determines the graph node N at the next moment. i The corresponding power consumption probability value p of the power load interval i,1 ~p i,9 The power load ranges include: 1000KWH~900KWH, 900KWH~800KWH, ..., 200KWH~100KWH. i,1 Used to indicate enterprise E i The power load consumed at the next moment is in the power load range of 1000KWH to 900KWH; p i,2 Used to indicate enterprise E i The power load consumed at the next moment is in the power load interval of 900KWH to 800KWH; ...; p i,9 Used to indicate enterprise E i The power load consumed at the next moment is in the power load interval of 200KWH to 100KWH.

[0070] Then the data processing platform 10 connects the graph node N to the graph neural network model. i The corresponding power consumption probability value p i,1 ~p i,9 For comparison, from p i,1 ~p i,9 Determine the maximum power consumption probability value P i and will be compared with the maximum power consumption probability value P i The corresponding power load interval is determined as graph node N i The final corresponding power load interval IN i .

[0071] Furthermore, the data processing platform 10 determines each enterprise E in the key industrial chain i Power load range IN i After that, take each enterprise E i Power load range IN i The median value MNi , and obtain the power consumption probability value P corresponding to the power load interval i For example, the power load range of a certain enterprise is 1000KWH~900KWH, and the middle value is 950KWH.

[0072] Thus, the data processing platform 10 uses a weighted summation method according to the power load interval IN i The central value MN i And the corresponding power consumption probability value P i , calculate the power load of the key industrial chain (ie, the first power load) AN:

[0073]

[0074] Therefore, this technical solution predicts the power consumption probability value of each enterprise at the next moment by inputting the graph structure corresponding to the key industrial chain into the preset graph neural network model, thereby determining the power load range of each enterprise. Therefore, this technical solution uses the characteristics of the graph neural network model that can accurately analyze the relationship between enterprises, making the prediction result more accurate.

[0075] Optionally, the operation of determining the power consumption probability value of the graph node according to the graph structure through the graph neural network model includes: extracting features of the graph nodes through the graph convolution module of the graph neural network model to generate feature information; and classifying the feature information using the multi-layer perceptron of the graph neural network model to determine the power consumption probability value of the graph node corresponding to the power load interval.

[0076] Specifically, refer to Figure 5 As shown, the graph neural network model includes m graph convolution modules and a multi-layer perceptron. The data processing platform 10 inputs the graph structure into the graph neural network model, and then the graph neural network model extracts features of each graph node in the graph structure in turn through the m graph convolution modules to generate feature information.

[0077] More specifically, for example, the graph structure constructed by the data processing platform 10 is a graph structure A, and the graph neural network model determines the graph node N in the graph structure A through the graph convolution module 1. i The adjacent nodes connected to the graph node N are connected according to the vector of the adjacent node and the weight of the edge connected to it. i Perform feature extraction operations to map node N i Update and generate a new graph structure A' (ie, feature information).

[0078] For example, with graph node N 1 The number of connected adjacent nodes is N 5 and N 6 , so the graph neural network model determines N through the graph convolution module 15 Vector EN 5 and N 1 With N 5 The weight of the connecting edge is a 1,5 , and determine N 6 Vector EN 6 and N 1 With N 6 The weight of the connecting edge is a 1,6 Then the graph neural network model passes through the graph convolution module 1 according to N 5 Vector EN 5 and weight a 1,5 , and N 6 Vector EN 6 and weight a 1,6 Weighted sum calculation graph node N 1 The updated vector EN 1 ', where EN 1 The calculation formula of ' is:

[0079] EN 1 '=EN 5 ×a 1,5 +EN 6 ×a 1,6 .

[0080] For example, the adjacent nodes connected to graph node N5 are N 1 、N 2 and N 8 , so the graph neural network model determines N through the graph convolution module 1 1 Vector EN 1 and N 1 With N 5 The weight of the connecting edge is a 1,5 , determine N 2 Vector EN 2 and N 2 With N 5 The weight of the connecting edge is a 2,5 , and determine the vector EN of N8 8 and N 5 With N 8 The weight of the connecting edge is a 5,8 Then the graph neural network model passes through the graph convolution module 1 according to N 1 Vector EN 1 and weight a 1,5 , N 2 Vector EN 2 and weight a 2,5 , and N 8 Vector EN 8 and weight a 5,8 Weighted sum calculation graph node N5 The updated vector EN 5 ', where EN 5 The calculation formula of ' is:

[0081] EN 5 '=EN 1 ×a 1,5 +EN 2 ×a 1,6 +EN 8 ×a 5,8 .

[0082] Thus, the graph neural network model updates the graph node N through the graph convolution module 1 according to the above 1 and N 5 The vector operation of the other graph nodes is updated, and the vectors of the other graph nodes are updated according to the updated vector EN i 'Generate a new graph structure A'.

[0083] Furthermore, the graph neural network model inputs the new graph structure A' into the graph convolution module 2, and the graph convolution module 2 updates the graph structure A' according to the above-mentioned method of updating the graph structure, and outputs the updated graph structure A". And so on, the graph neural network model outputs the updated graph structure A"' through the graph convolution module m.

[0084] Furthermore, the graph neural network model inputs the graph structure A'' (i.e., feature information) into a multi-layer perceptron, and the multi-layer perceptron classifies the graph structure A'' (i.e., feature information) to determine the graph node N i The power consumption probability value P corresponding to the power load interval i .

[0085] Therefore, this technical solution extracts features of the graph structure through multiple graph convolution modules, so as to capture deeper feature information and make the obtained feature information more accurate.

[0086] Optionally, the operation of determining a power dispatching strategy corresponding to the key industrial chain according to the first power load includes: determining a power dispatching strategy corresponding to the key industrial chain according to the first power load based on a genetic algorithm.

[0087] Specifically, the data processing platform 10 determines the number of each power supply device such as photovoltaic devices, wind power devices, energy storage devices and thermal power units in the regional power grid. 1 PV devices, respectively Total n 2 wind power equipment, respectively marked as Total n 3 thermal power units, respectively marked as Total n 4energy storage devices, respectively marked as

[0088] Afterwards, the data processing platform 10 determines the power supply of each photovoltaic device and wind power device and other new energy equipment according to the meteorological forecast data through the preset first power prediction model. And the data processing platform 10 determines the power supply of the thermal power unit according to the power supply history data of the thermal power unit through the preset second power prediction model. And the data processing platform 10 determines the power supply of the energy storage device according to the power supply history data of the energy storage device through the preset third power prediction model. And the data processing platform 10 determines the power supply of the national power grid according to the power supply history data of the national power grid through the preset fourth power prediction model. Among them, the first power prediction model, the second power prediction model, the third power prediction model and the fourth power prediction model are all pre-trained according to the sample data.

[0089] More specifically, for new energy equipment, the data processing platform 10 collects weather forecast data for the next moment. The weather forecast data includes data on sunshine intensity, cloud cover, temperature, wind speed, temperature, and air pressure. The data processing platform 10 then inputs the weather forecast data for the next moment into a preset first power forecast model, so that the first power forecast model outputs the next moment of the photovoltaic equipment according to the weather forecast data. Power generated And wind power equipment Power generated

[0090] For the thermal power generation unit, the data processing platform 10 collects the historical power supply data of the thermal power generation unit within the sample period, and then inputs the historical power supply data into the second power prediction model to predict the thermal power generation unit through the second power prediction model. Power supply

[0091] For energy storage devices, the data processing platform 10 collects historical power supply data of the energy storage devices within a sample period, and then inputs the historical power supply data into a third power prediction model to predict the energy storage devices. Power supply

[0092] For the national power grid, the data processing platform 10 collects the historical power supply data of the national power grid within the sample period, and then inputs the historical power supply data into the fourth power prediction model to predict the power supply PB of the national power grid B through the fourth power prediction model.

[0093] Further, the data processing platform 10 constructs power supply constraints related to each power supply device according to the weight and power supply of each power supply device. The weight value is 0 or 1, indicating whether the power supply device is selected for power supply. 0 indicates that the power supply device is not selected for power supply, and 1 indicates that the power supply device is selected for power supply.

[0094] For photovoltaic equipment Total power supply PW 1 :

[0095]

[0096] in Photovoltaic equipment The weight of j 1 =1~n 1 .

[0097] For wind power equipment Total power supply PW 2 :

[0098]

[0099] in Wind power equipment The weight of j 2 =1~n 2 .

[0100] For thermal power plants Total power supply PW 3 :

[0101]

[0102] in Represents thermal power unit The weight of j 3 =1~n 3 .

[0103] For energy storage devices Total power supply PW 4 :

[0104]

[0105] in Energy storage equipment The weight of j 4 =1~n 4 .

[0106] For the total power supply power PW of the national power grid B 5 :

[0107] PW 5 =W(B)×PB.

[0108] Where W(B) represents the weight of the national power grid B.

[0109] The constraints are: PW 1 +PW 2 +PW 3 +PW 4 +PW 5 ≥AN. AN is the power load of the key industrial chain.

[0110] Furthermore, the data processing platform 10 constructs a cost calculation model based on the operating cost coefficient of each power supply device and the power supply cost. More specifically, the data processing platform 10 constructs a cost calculation model based on the preset operating cost coefficient θ of the photovoltaic device. 1 and the total power supply PW of the photovoltaic equipment 1 , calculate the power supply cost Fa of the photovoltaic equipment: Fa = θ 1 *PW 1 Then the data processing platform 10 calculates the operating cost coefficient θ of the wind power equipment according to the preset 2 and the total power supply PW of wind power equipment 2 , calculate the power supply cost Fb of wind power equipment: Fb = θ 2 *PW 2 Then the data processing platform 10 calculates the operating cost coefficient θ of the thermal power unit according to the preset 3 And the total power supply power PW of thermal power units 3 , calculate the power supply cost Fc of the thermal power unit: Fc = θ 3 *PW 3 Then the data processing platform 10 calculates the operating cost coefficient θ of the energy storage device according to the preset operating cost coefficient θ 4 and the total power supply power PW of the energy storage device 4 , calculate the power supply cost Fd of the energy storage device: Fd = θ 4 *PW 4 The data processing platform 10 then calculates the power price Ce at the next moment and the total power supply power PW of the national power grid. 5 , calculate the power supply cost Fe of the national power grid: Fe = Ce*PW 5 .

[0111] Furthermore, the data processing platform 10 constructs a power supply cost calculation model based on the power supply cost Fa of the photovoltaic equipment, the power supply cost Fb of the wind power equipment, the power supply cost Fc of the thermal power unit, the power supply cost Fd of the energy storage equipment and the power supply cost Fe of the national power grid, wherein the power supply cost calculation model is: FX=Fa+Fb+Fc+Fd+Fe.

[0112] Furthermore, the data processing platform 10 solves the cost calculation model based on the genetic algorithm under the premise of meeting the constraint conditions, thereby determining the power supply selection of each power supply device corresponding to the minimum cost. The genetic algorithm is as follows:

[0113] First, the data processing platform 10 initializes the initial population of the genetic algorithm, where the initial population includes y initial individuals The initial individual is used to indicate the weight of each power supply device, where the weight is 0 or 1.

[0114] And among them

[0115]

[0116]

[0117] in For the initial individual Photovoltaic equipment The weight of j 1 =1~n 1 , r = 1 ~ y; For the initial individual Wind power equipment The weight of j 2 =1~n 2 ; For the initial individual Thermal power units The weight of j 3 =1~n 3 ; For the initial individual Energy storage equipment The weight of j 4 =1~n 4 ; For the initial individual The weight of national grid B.

[0118] The data processing platform 10 then uses the cost calculation model as the initial individual The fitness model of . Thus, the data processing platform 10 converts the initial individual Substitute them into the cost calculation model FX=Fa+Fb+Fc+Fd+Fe respectively.

[0119] The data processing platform 10 then determines the fitness The selection probability of each initial individual is calculated according to the fitness, and then the first number of initial individuals with higher selection probability are selected as the first individual, thereby generating the first population. Individuals with higher fitness have a greater probability of being selected. The method for calculating the selection probability can be, for example, a roulette wheel selection method.

[0120] The data processing platform 10 then determines a second number of individuals from the first population according to a preset crossover rate, performs a crossover operation on the individuals to generate new individuals (ie, second individuals), and thus the second individuals form a second population.

[0121] The data processing platform 10 then determines a third number of individuals from the second population according to a preset mutation rate, performs mutation operations on the individuals, generates new individuals (ie, the third individual), and thus the third individuals form a third population.

[0122] At this point, the first round of iteration of the genetic algorithm is completed.

[0123] For the second round of iteration, the data processing platform 10 inputs the third individuals in the third population into the cost calculation model to determine the fitness of each third individual, and then the data processing platform 10 performs selection operations, crossover operations, and mutation operations on the third individuals in accordance with the steps in the first round of iteration, thereby generating a new population. At this point, the second round of iteration of the genetic algorithm is completed.

[0124] The data processing platform 10 thus performs multiple rounds of iterative calculations according to the above steps until a preset number of iterations is reached.

[0125] For example, the pre-set number of iterations is three, so the data processing platform 10 uses the power supply equipment with a weight of 1 among the individuals corresponding to the minimum cost determined in the third round of iteration as the power supply equipment of the key industrial chain, thereby determining the power dispatching strategy corresponding to the key industrial chain.

[0126] Therefore, this technical solution determines the dispatching strategy corresponding to the power supply equipment of the key industrial chain according to the minimum power supply cost, so as to provide sufficient power supply with minimum loss and improve the utilization rate of power resources. In addition, this technical solution obtains the optimal dispatching scheme of each power supply equipment based on the minimum power supply cost through genetic algorithm, so as to flexibly dispatch each power supply equipment.

[0127] Optionally, the method further includes: determining a power dispatching strategy corresponding to the non-critical industrial chain based on a genetic algorithm according to the power load of the non-critical industrial chain.

[0128] Specifically, after the data processing platform 10 determines the power supply equipment of the key industrial chain, it uses the remaining power supply equipment (i.e., the power supply equipment with a weight of 0 in the individuals corresponding to the minimum cost) as the power supply equipment for power dispatching of the non-key industrial chain. Among them, the national power grid can supply power to both the key industrial chain and the non-key industrial chain at the same time, that is, after the national power grid serves as the power supply equipment of the key industrial chain, it can still serve as the power supply equipment of the non-key industrial chain. Therefore, the data processing platform 10 determines that the power supply equipment corresponding to the non-key industrial chain includes: 1 PV devices, respectively Where L 1 ≤n 1 ; Total L 2 wind power equipment, respectively marked as Where L 2 ≤n 2 ; Total L 3 thermal power units, respectively marked as Where L 3 ≤n 3 ; Total L 4 energy storage devices, respectively marked as Where L 4 ≤n 4 .

[0129] Further, the data processing platform 10 constructs power supply constraints related to each power supply device according to the weights and power supplies of each power supply device corresponding to the non-critical industrial chain. The weight value is 0 or 1, indicating whether the power supply device is selected for power supply. 0 indicates that the power supply device is not selected for power supply, and 1 indicates that the power supply device is selected for power supply.

[0130] For photovoltaic equipment Total power supply PV 1 :

[0131]

[0132] in Photovoltaic equipment The weight, v 1 =1~L 1 .

[0133] For wind power equipment Total power supply PV 2 :

[0134]

[0135] in Wind power equipment The weight, v 2=1~L 2 .

[0136] For thermal power plants Total power supply PV 3 :

[0137]

[0138] in Represents thermal power unit The weight, v 3 =1~L 3 .

[0139] For energy storage devices Total power supply PV 4 :

[0140]

[0141] in Energy storage equipment The weight, v 4 =1~L 4 .

[0142] For the total power supply PV of the national grid B 5 :

[0143] PV 5 =W(B)×PB.

[0144] Where W(B) represents the weight of the national power grid B.

[0145] The constraints are: PV 1 +PV 2 +PV 3 +PV 4 +PV 5 ≥NAN. NAN is the power load of non-critical industrial chains.

[0146] Furthermore, the data processing platform 10 constructs a cost calculation model based on the operating cost coefficients of each power supply device and the total power supply cost. More specifically, the data processing platform 10 constructs a cost calculation model based on the preset operating cost coefficients θ of the photovoltaic devices. 1 and the total power supply PV of the photovoltaic equipment 1 , calculate the power supply cost Za of the photovoltaic equipment: Za = θ 1 *PV 1 Then the data processing platform 10 calculates the operating cost coefficient θ of the wind power equipment according to the preset 2 Total power supply of PV and wind power equipment 2 , calculate the power supply cost Zb of wind power equipment: Zb = θ 2 *PV2 Then the data processing platform 10 calculates the operating cost coefficient θ of the thermal power unit according to the preset 3 and the total power supply PV of thermal power units 3 , calculate the power supply cost Zc of the thermal power unit: Zc = θ 3 *PV 3 Then the data processing platform 10 calculates the operating cost coefficient θ of the energy storage device according to the preset operating cost coefficient θ 4 Total power supply PV and energy storage equipment 4 , calculate the power supply cost Zd of the energy storage device: Zd = θ 4 *PV 4 The data processing platform 10 then calculates the power price Ce at the next moment and the total power supply PV of the national power grid. 5 , calculate the power supply cost Ze of the national power grid: Ze=Ce*PV 5 .

[0147] Furthermore, the data processing platform 10 constructs a power supply cost calculation model based on the power supply cost Za of the photovoltaic equipment, the power supply cost Zb of the wind power equipment, the power supply cost Zc of the thermal power unit, the power supply cost Zd of the energy storage equipment and the power supply cost Ze of the national power grid, wherein the power supply cost calculation model is: ZX=Za+Zb+Zc+Zd+Ze.

[0148] Furthermore, the data processing platform 10 solves the cost calculation model based on the genetic algorithm under the premise of meeting the constraint conditions, thereby determining the power supply selection of each power supply device corresponding to the minimum cost. The genetic algorithm is as follows:

[0149] First, the data processing platform 10 initializes the initial population of the genetic algorithm, where the initial population includes y initial individuals The initial individual is used to indicate the weight of each power supply device, where the weight is 0 or 1.

[0150] And among them

[0151]

[0152] in For the initial individual Photovoltaic equipment The weight, v 1 =1~L 1 , r = 1 ~ y; For the initial individual Wind power equipment The weight, v 2 =1~L 2 ; For the initial individual Thermal power units The weight, v 3 =1~L 3 ; For the initial individual Energy storage equipment The weight, v 4 =1~L 4 ; For the initial individual The weight of national grid B.

[0153] The data processing platform 10 then uses the cost calculation model as the initial individual The fitness model of . Thus, the data processing platform 10 converts the initial individual Substitute them into the cost calculation model ZX=Za+Zb+Zc+Zd+Ze respectively.

[0154] The data processing platform 10 then determines the fitness The selection probability of each initial individual is calculated according to the fitness, and then the first number of initial individuals with higher selection probability are selected as the first individual, thereby generating the first population. Individuals with higher fitness have a greater probability of being selected. The method for calculating the selection probability can be, for example, a roulette wheel selection method.

[0155] The data processing platform 10 then determines a second number of individuals from the first population according to a preset crossover rate, performs a crossover operation on the individuals to generate new individuals (ie, second individuals), and thus the second individuals form a second population.

[0156] The data processing platform 10 then determines a third number of individuals from the second population according to a preset mutation rate, performs mutation operations on the individuals, generates new individuals (ie, the third individual), and thus the third individuals form a third population.

[0157] At this point, the first round of iteration of the genetic algorithm is completed.

[0158] For the second round of iteration, the data processing platform 10 inputs the third individuals in the third population into the cost calculation model to determine the fitness of each third individual, and then the data processing platform 10 performs selection operations, crossover operations, and mutation operations on the third individuals in accordance with the steps in the first round of iteration, thereby generating a new population. At this point, the second round of iteration of the genetic algorithm is completed.

[0159] The data processing platform 10 thus performs multiple rounds of iterative calculations according to the above steps until a preset number of iterations is reached.

[0160] For example, the pre-set number of iterations is three, so the data processing platform 10 uses the power supply equipment with a weight of 1 among the individuals corresponding to the minimum cost determined in the third round of iteration as the power supply equipment of the non-critical industrial chain, thereby determining the power dispatching strategy corresponding to the non-critical industrial chain.

[0161] Therefore, this technical solution dispatches power supply equipment according to the power demand of key industrial chains first, and then uses the remaining power supply equipment to determine the power dispatch strategy of non-key industrial chains. This realizes the priority of providing power dispatch for key industrial chains. And through genetic algorithms, power dispatch strategies are formulated for key industrial chains and non-key industrial chains, making the dispatch more accurate.

[0162] In summary, reference Figure 6 As shown, the data processing platform 10 performs power dispatching for key industrial chains in the following order:

[0163] S601: The data processing platform 10 determines key industry chain related enterprises in the target area;

[0164] S602: The data processing platform 10 associates enterprises with transaction relationships and constructs a graph structure corresponding to the key industrial chain;

[0165] S603: The data processing platform 10 obtains electricity consumption information related to the enterprise as an attribute of a corresponding graph node;

[0166] S604: The data processing platform 10 determines the second power load of the enterprise at each time within a preset period;

[0167] S605: The data processing platform 10 determines the power consumption correlation between the enterprises according to the second power load at each time through a preset correlation calculation formula;

[0168] S606: The data processing platform 10 connects the graph nodes through connection edges with corresponding weights to construct a graph structure;

[0169] S607: The data processing platform 10 extracts features from the graph nodes through the graph convolution module of the graph neural network model to generate feature information;

[0170] S608: The data processing platform 10 uses the multi-layer perceptron of the graph neural network model to classify the feature information and determine the power consumption probability value corresponding to the power load interval of the graph node;

[0171] S609: The data processing platform 10 calculates the first power load of the key industrial chain according to the central value of the power load interval and the corresponding power consumption probability value by using a weighted summation method;

[0172] S610: The data processing platform 10 determines a power dispatching strategy corresponding to the key industrial chain according to the first power load.

[0173] In addition, according to a second aspect of this embodiment, a computer system is provided, including a memory, a processor, and a computer program stored in the memory, and the processor executes the computer program to implement the steps of the above method.

[0174] In addition, according to a third aspect of this 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 steps of the above method are implemented.

[0175] In addition, according to a fourth aspect of this embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0176] Thus, according to this embodiment, the data processing platform constructs a detailed graph structure by accurately identifying and associating the key industrial chains and their upstream and downstream enterprises in the target area, so that the upstream and downstream enterprises can be associated and the correlation between the characteristic information of the enterprises can be accurately determined. In addition, this technical solution realizes the dynamic prediction of the specific electricity demand of the key industrial chain at the future moment through the graph neural network model, which greatly improves the accuracy and timeliness of the prediction. This prediction method based on real-time data analysis and deep learning overcomes the prediction deviation caused by static assumptions and human intervention in traditional methods, and ensures the efficient and accurate allocation of power resources. In addition, this technical solution realizes the comprehensive consideration and intelligent scheduling of the power demand of the entire region by integrating the power demand forecast of the key industrial chain and the non-key industrial chain, and enhances the overall stability and response speed of the power system. Especially in the face of emergencies or demand fluctuations, based on the dynamic adjustment capability of the graph neural network, it can respond quickly to ensure the continuity of the key industrial chain and the basic power supply of the non-key industrial chain, and improve the adaptability of power dispatching. Thus, the technical problem that the power resource dispatching scheme existing in the prior art cannot accurately dispatch different industrial chains is solved.

[0177] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0178] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0179] Example 2

[0180] Figure 7 FIG. 7 shows an apparatus 700 for analyzing and predicting industrial chain electricity consumption data based on correlation according to this embodiment, and the apparatus 700 corresponds to the method described in the first aspect of Embodiment 1. Figure 7 As shown, the device 700 includes: an enterprise determination module 710, which is used to determine the key industrial chain-related enterprises in the target area; a graph structure construction module 720, which is used to associate enterprises with transaction relationships and construct a graph structure corresponding to the key industrial chain; a load prediction module 730, which is used to predict the first power load of the key industrial chain at the next moment according to the graph structure through a graph neural network model; and a strategy determination module 740, which is used to determine the power dispatching strategy corresponding to the key industrial chain according to the first power load.

[0181] Optionally, the graph structure construction module 720 includes: an information acquisition submodule, used to obtain electricity consumption information related to the enterprise as the attribute of the corresponding graph node, wherein the electricity consumption information includes current load, current load variance, long-term average load, long-term average load variance and temperature; a first calculation submodule, used to calculate the electricity consumption correlation between enterprises with transaction relationships, and use the electricity consumption correlation as the weight of the connection edge between the graph nodes; and a structure construction submodule, used to connect the graph nodes through connection edges with corresponding weights to construct a graph structure.

[0182] Optionally, the first calculation submodule includes: a first determination unit, used to determine the second power load of the enterprise at each moment within a preset period; and a second determination unit, used to determine the power consumption correlation between enterprises according to the second power load at each moment through a preset correlation calculation formula.

[0183] Optionally, the load prediction module 730 includes: a first determination submodule, used to determine the power consumption probability value of the graph node according to the graph structure through a graph neural network model, wherein the power consumption probability value is used to predict the power load range of each enterprise in the key industrial chain at the next moment; and a second calculation submodule, used to calculate the first power load of the key industrial chain according to the central value of the power load range and the corresponding power consumption probability value by using a weighted summation method.

[0184] Optionally, the first determination submodule includes: an information generation unit, used to extract features of graph nodes and generate feature information through a graph convolution module of a graph neural network model; and a third determination unit, used to classify the feature information using a multi-layer perceptron of the graph neural network model to determine the power consumption probability value of the graph node corresponding to the power load interval.

[0185] Optionally, the strategy determination module 740 includes: a strategy determination submodule, configured to determine a power dispatching strategy corresponding to the key industrial chain according to the first power load based on a genetic algorithm.

[0186] Optionally, the device 700 further includes: a second strategy determination module, configured to determine a power dispatching strategy corresponding to a non-critical industrial chain according to the power load of the non-critical industrial chain based on a genetic algorithm.

[0187] Thus, according to this embodiment, the data processing platform constructs a detailed graph structure by accurately identifying and associating the key industrial chains and their upstream and downstream enterprises in the target area, so that the upstream and downstream enterprises can be associated and the correlation between the characteristic information of the enterprises can be accurately determined. In addition, this technical solution realizes the dynamic prediction of the specific electricity demand of the key industrial chain at the future moment through the graph neural network model, which greatly improves the accuracy and timeliness of the prediction. This prediction method based on real-time data analysis and deep learning overcomes the prediction deviation caused by static assumptions and human intervention in traditional methods, and ensures the efficient and accurate allocation of power resources. In addition, this technical solution realizes the comprehensive consideration and intelligent scheduling of the power demand of the entire region by integrating the power demand forecast of the key industrial chain and the non-key industrial chain, and enhances the overall stability and response speed of the power system. Especially in the face of emergencies or demand fluctuations, based on the dynamic adjustment capability of the graph neural network, it can respond quickly to ensure the continuity of the key industrial chain and the basic power supply of the non-key industrial chain, and improve the adaptability of power dispatching. Thus, the technical problem that the power resource dispatching scheme existing in the prior art cannot accurately dispatch different industrial chains is solved.

[0188] Example 3

[0189] Figure 8The device 800 for analyzing and predicting industrial chain electricity consumption data based on correlation according to this embodiment is shown, and the device 800 corresponds to the method according to the first aspect of embodiment 1. Figure 8 As shown, the device 800 includes: a processor 810; and a memory 820, which is connected to the processor 810 and is used to provide the processor 810 with instructions for processing the following processing steps: determining key industrial chain-related enterprises in the target area; associating enterprises with transaction relationships to construct a graph structure corresponding to the key industrial chain; predicting the first power load of the key industrial chain at the next moment based on the graph structure through a graph neural network model; and determining the power dispatching strategy corresponding to the key industrial chain based on the first power load.

[0190] Optionally, enterprises with transaction relationships are associated to construct a graph structure corresponding to the key industrial chain, including: obtaining electricity consumption information related to the enterprise as the attribute of the corresponding graph node, wherein the electricity consumption information includes current load, current load variance, long-term average load, long-term average load variance and temperature; calculating the electricity consumption correlation between enterprises with transaction relationships, and using the electricity consumption correlation as the weight of the connection edges between graph nodes; and connecting the graph nodes through connection edges with corresponding weights to construct a graph structure.

[0191] Optionally, the operation of calculating the electricity consumption correlation between enterprises with a transaction relationship and using the electricity consumption correlation as the weight of the connecting edges between graph nodes includes: determining the second electricity load of the enterprise at each moment within a preset period; and determining the electricity consumption correlation between the enterprises according to the second electricity load at each moment through a preset correlation calculation formula.

[0192] Optionally, the operation of predicting the first power load of the key industrial chain at the next moment is performed through a graph neural network model according to the graph structure, including: determining the power probability value of the graph node according to the graph structure through a graph neural network model, wherein the power probability value is used to predict the power load range of each enterprise in the key industrial chain at the next moment; and calculating the first power load of the key industrial chain according to the center value of the power load range and the corresponding power probability value by using a weighted summation method.

[0193] Optionally, the operation of determining the power consumption probability value of the graph node according to the graph structure through the graph neural network model includes: extracting features of the graph nodes through the graph convolution module of the graph neural network model to generate feature information; and classifying the feature information using the multi-layer perceptron of the graph neural network model to determine the power consumption probability value of the graph node corresponding to the power load interval.

[0194] Optionally, the operation of determining a power dispatching strategy corresponding to the key industrial chain according to the first power load includes: determining a power dispatching strategy corresponding to the key industrial chain according to the first power load based on a genetic algorithm.

[0195] Optionally, the memory 820 is also used to provide the processor 810 with instructions for processing the following processing steps: determining a power dispatching strategy corresponding to a non-critical industrial chain based on the power load of the non-critical industrial chain based on a genetic algorithm.

[0196] Thus, according to this embodiment, the data processing platform constructs a detailed graph structure by accurately identifying and associating the key industrial chains and their upstream and downstream enterprises in the target area, so that the upstream and downstream enterprises can be associated and the correlation between the characteristic information of the enterprises can be accurately determined. In addition, this technical solution realizes the dynamic prediction of the specific electricity demand of the key industrial chain at the future moment through the graph neural network model, which greatly improves the accuracy and timeliness of the prediction. This prediction method based on real-time data analysis and deep learning overcomes the prediction deviation caused by static assumptions and human intervention in traditional methods, and ensures the efficient and accurate allocation of power resources. In addition, this technical solution realizes the comprehensive consideration and intelligent scheduling of the power demand of the entire region by integrating the power demand forecast of the key industrial chain and the non-key industrial chain, and enhances the overall stability and response speed of the power system. Especially in the face of emergencies or demand fluctuations, based on the dynamic adjustment capability of the graph neural network, it can respond quickly to ensure the continuity of the key industrial chain and the basic power supply of the non-key industrial chain, and improve the adaptability of power dispatching. Thus, the technical problem that the power resource dispatching scheme existing in the prior art cannot accurately dispatch different industrial chains is solved.

[0197] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0198] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0199] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0200] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0201] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0202] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0203] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for analyzing and predicting electricity consumption data of an industrial chain based on correlation, characterized in that: include: Identify key industry chain-related enterprises in the target area; Associating enterprises with transaction relationships to construct a graph structure corresponding to the key industrial chain; Predicting the first power load of the key industrial chain at the next moment according to the graph structure through a graph neural network model; and A power dispatching strategy corresponding to the key industrial chain is determined according to the first power load.

2. The method according to claim 1, characterized in that The operations of associating enterprises with transaction relationships and constructing a graph structure corresponding to the key industrial chain include: Acquire electricity usage information related to the enterprise as an attribute of a corresponding graph node, wherein the electricity usage information includes current load, current load variance, long-term average load, long-term average load variance, and temperature; Calculating the electricity consumption correlation between enterprises having a transaction relationship, and using the electricity consumption correlation as the weight of the connection edge between the graph nodes; and The graph nodes are connected through connection edges with corresponding weights to construct the graph structure.

3. The method according to claim 2, characterized in that The operation of calculating the electricity consumption correlation between enterprises having a transaction relationship and using the electricity consumption correlation as the weight of the connection edge between the graph nodes includes: Determine a second power load of the enterprise at each time within a preset period; and The power consumption correlation between the enterprises is determined according to the second power load at each moment through a preset correlation calculation formula.

4. The method according to claim 2, characterized in that: Predicting the operation of the first power load of the key industrial chain at the next moment according to the graph structure through the graph neural network model, including: Determining the power consumption probability value of the graph node according to the graph structure through the graph neural network model, wherein the power consumption probability value is used to predict the power load range of each enterprise in the key industrial chain at the next moment; and The first power load of the key industrial chain is calculated according to the central value of the power load interval and the corresponding power consumption probability value by using a weighted summation method.

5. The method according to claim 4, characterized in that The operation of determining the power usage probability value of the graph node according to the graph structure through the graph neural network model includes: Extracting features of the graph nodes through the graph convolution module of the graph neural network model to generate feature information; and The feature information is classified and processed using a multi-layer perceptron of the graph neural network model to determine the power consumption probability value of the graph node corresponding to the power load interval.

6. The method according to claim 1, characterized in that The operation of determining the power dispatching strategy corresponding to the key industrial chain according to the first power load includes: Based on the genetic algorithm and according to the first power load, a power dispatching strategy corresponding to the key industrial chain is determined.

7. The method according to claim 1, characterized in that Also includes: Based on the genetic algorithm and according to the power load of the non-critical industrial chain, a power dispatching strategy corresponding to the non-critical industrial chain is determined.

8. A device for analyzing and predicting electricity consumption data of an industrial chain based on correlation, characterized in that: include: Enterprise identification module, used to identify key industry chain related enterprises in the target area; A graph structure building module, used to associate enterprises with transaction relationships and build a graph structure corresponding to the key industrial chain; A load prediction module, configured to predict the first power load of the key industrial chain at the next moment according to the graph structure through a graph neural network model; and A strategy determination module is used to determine the power dispatching strategy corresponding to the key industrial chain according to the first power load.

9. A device for analyzing and predicting electricity consumption data of an industrial chain based on correlation, characterized in that: include: processor; as well as A memory, connected to the processor, configured to provide the processor with instructions for processing the following processing steps: Identify key industry chain-related enterprises in the target area; Associating enterprises with transaction relationships to construct a graph structure corresponding to the key industrial chain; Predicting the first power load of the key industrial chain at the next moment according to the graph structure through a graph neural network model; and A power dispatching strategy corresponding to the key industrial chain is determined according to the first power load.

10. A computer system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method of claim 1.

Images