Load prediction method and device for energy management system, equipment and storage medium
By acquiring and classifying load correlation factors in the energy management system and building and training corresponding load prediction models, the problem of low prediction accuracy in the prior art is solved, and higher prediction accuracy and stability are achieved.
Patent Information
- Application Number
- CN202411977045.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-31
- Publication Date
- 2025-05-30
AI Technical Summary
The prior art ignores the impact of dynamically changing data in power load prediction, and traditional time series prediction methods fail to effectively utilize the correlation between power load data, resulting in limited prediction accuracy.
By obtaining all related factors of load in the energy management system, classifying them according to the correspondence between factors and power load data, building and training the load prediction model corresponding to each type of related factors, and finally summing the various predicted values to obtain the total load prediction value.
The accuracy and stability of load prediction are improved, and compared with traditional methods, the accuracy and stability of predictions is improved by more than 20%, and the deviation is controlled at about 10%.
Smart Images
Figure CN120069154A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of energy management systems, and specifically, to a load prediction method, device, equipment, and storage medium for an energy management system. Background Art
[0002] Under the background of the national energy intensity and total amount control, and the major policies of carbon peak and carbon neutrality, and under the background that more requirements are put forward for the management of the power demand side and power load management in the power aspect, it is required that all walks of life using electricity at the terminal can understand, predict, and control their own electricity loads. The energy management system has become an essential information system for enterprises and large systems, which can help enterprises and large systems effectively manage all aspects of the purchase, storage, processing, conversion, transmission, distribution, and terminal consumption of their energy. Among them, the effective management of power load is the most basic and important part, and the effective and accurate prediction of power load has become one of the main problems. This problem needs to be solved as soon as possible in order to effectively guide enterprises and large systems in load management and respond to the requirements of the demand side platform.
[0003] There are mainly two existing load prediction methods: one is the prediction method based on historical data, but this method ignores the impact of dynamic change data on the load, especially the impact of data such as weather, human flow, enterprise orders, and production volume on the load is very large; in recent years, statistical and artificial intelligence technologies have been widely used. The application statistical prediction models often used for power load prediction include autoregressive model (Autoregressive Model), moving average model (Moving Average Model), autoregressive moving average model (Autoregressive Moving Average Model), and autoregressive integrated moving average model (Autoregressive Integrated Moving Average Model). These models do not require high computing requirements, but when compared with artificial intelligence-based models, the prediction accuracy of these models is lower. On the other hand, artificial intelligence technologies, such as artificial neural networks (Artificial Neural Networks) and support vector regression (Support Vector Regression, SVR), although showing more accurate performance, are more expensive in terms of computing cost. More importantly, the above algorithms do not consider the correlation between consecutive power load series data. Power load data, as a classic time series data, has a high degree of correlation between consecutive power load time series. Therefore, traditional time series prediction methods ignore the correlation between consecutive power load data with distance characteristics in the time dimension, resulting in limited prediction accuracy. Summary of the Invention
[0004] An embodiment of the present application provides a method, apparatus, device, and storage medium for load forecasting in an energy management system.
[0005] In a first aspect of the embodiments of the present application, a method for load forecasting in an energy management system is provided, including:
[0006] Obtain all associated factors of the load in the energy management system;
[0007] Classify all associated factors according to the corresponding relationship between the associated factors and the power load data;
[0008] Construct and train a load forecasting model corresponding to each type of associated factor respectively;
[0009] According to each trained load forecasting model, determine the load forecasting value corresponding to each type of associated factor, and sum up the load forecasting values corresponding to each type of associated factor to obtain the total load forecasting value of the energy management system.
[0010] In an optional embodiment of the present application, all associated factors of the load in the energy management system include the number of system orders, production volume, number of personnel, equipment type, equipment operating conditions, equipment efficiency, production plan, outdoor weather conditions, holidays, day, night, and power quality.
[0011] In an optional embodiment of the present application, classifying all associated factors according to the corresponding relationship between the associated factors and the power load data includes:
[0012] Obtain the power load data distribution in the energy management system, and determine the first power load data level, the second power load data level, and the third power load data level from large to small according to the power load data distribution;
[0013] The first type of associated factors corresponding to the first power load data level includes the number of orders and production volume, the second type of associated factors corresponding to the second power load data level includes equipment type, equipment operating conditions, equipment efficiency, and outdoor weather conditions; the third type of associated factors corresponding to the third power load data level includes the number of personnel, outdoor weather conditions, and holidays. The associated factors other than the first type of associated factors, the second type of associated factors, and the third type of associated factors among all associated factors are used as the fourth type of associated factors.
[0014] In an optional embodiment of the present application, constructing a load forecasting model corresponding to each type of associated factor respectively includes:
[0015] The load prediction models corresponding to the first - type associated factors, the second - type associated factors, and the third - type associated factors are each any one of the long - short - term memory network model, the gated recurrent unit network model, and the one - dimensional convolutional neural network model, and the load prediction model corresponding to the fourth - type associated factors is an artificial neural network.
[0016] In an alternative embodiment of the present application, training the load prediction models corresponding to each type of associated factor respectively includes:
[0017] Taking the order quantity and production volume as inputs and the load value as the output, training the load prediction model corresponding to the first - type associated factors;
[0018] Taking the equipment type, equipment operating conditions, equipment efficiency, and outdoor weather conditions as inputs and the load value as the output, training the load prediction model corresponding to the second - type associated factors;
[0019] Taking the number of personnel, outdoor weather conditions, and holidays as inputs and the load value as the output, training the load prediction model corresponding to the third - type associated factors;
[0020] Taking the associated factors other than the first - type associated factors, the second - type associated factors, and the third - type associated factors among all associated factors as inputs and the load value as the output, training the load prediction model corresponding to the fourth - type associated factors.
[0021] In an alternative embodiment of the present application, the method further includes:
[0022] For the first - target - type associated factors in each type of associated factors where the complexity of the associated factors exceeds a first preset threshold, splitting the load prediction model corresponding to each first - target - type associated factor into multiple sub - models, and training each sub - model to predict the load through the trained sub - models.
[0023] In an alternative embodiment of the present application, the method further includes:
[0024] For the second - target - type associated factors in each type of associated factors where the data distribution uniformity of the associated factors exceeds a second preset threshold, splitting the load prediction model corresponding to each second - target - type associated factor into multiple sub - models, and training each sub - model to predict the load through the trained sub - models.
[0025] In the second aspect of the embodiments of the present application, there is provided a load prediction device for an energy management system, including:
[0026] An acquisition module, configured to acquire all associated factors of the load in the energy management system;
[0027] A classification module, configured to classify all associated factors according to the correspondence between the associated factors and the power load data;
[0028] A construction module, configured to respectively construct and train a load prediction model corresponding to each category of associated factors;
[0029] A determination module, configured to determine the load prediction value corresponding to each category of associated factors according to each trained load prediction model, and sum up the load prediction values corresponding to each category of associated factors to obtain the total load prediction value of the energy management system.
[0030] In a third aspect of the embodiments of the present application, a computer device is provided, including: a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the above methods for load prediction of an energy management system are implemented.
[0031] In a fourth aspect of the embodiments of the present application, 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 any one of the above methods for load prediction of an energy management system are implemented.
[0032] The above technical solutions provided by the embodiments of the present application have at least some or all of the following advantages compared with the prior art:
[0033] The method for load prediction of an energy management system described in the embodiments of the present application obtains all associated factors of the load in the energy management system; classifies all associated factors according to the correspondence between the associated factors and the power load data; respectively constructs and trains a load prediction model corresponding to each category of associated factors; determines the load prediction value corresponding to each category of associated factors according to each trained load prediction model, and sums up the load prediction values corresponding to each category of associated factors to obtain the total load prediction value of the energy management system, classifies the associated factors of the load, and respectively constructs a load prediction model for each category of associated factors, which can solve the problems of being too complex and having low accuracy existing in the existing prediction methods. Description of the Drawings
[0034] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:
[0035] Figure 1 It is a flowchart of the method for load prediction of an energy management system provided by an embodiment of the present application;
[0036] Figure 2 It is a comparison diagram of predicted power consumption load and actual power consumption load provided by an embodiment of the present application;
[0037] Figure 3 Schematic structural diagram of a load forecasting device for an energy management system provided by an embodiment of the present application;
[0038] Figure 4 Schematic structural diagram of a computer device provided by an embodiment of the present application. Specific embodiments
[0039] In order to make the technical solutions and advantages in the embodiments of the present application clearer and more understandable, the following further details the exemplary embodiments of the present application with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than an exhaustive list of all embodiments. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.
[0040] Please refer to Figure 1 , the load forecasting method for an energy management system provided by the embodiments of the present application includes the following steps 100 to 400:
[0041] S100, obtain all associated factors of the load in the energy management system;
[0042] S200, classify all associated factors according to the corresponding relationship between the associated factors and the power load data;
[0043] S300, respectively construct and train a load forecasting model corresponding to each type of associated factor;
[0044] S400, determine the load forecasting value corresponding to each type of associated factor according to each trained load forecasting model, and sum up the load forecasting values corresponding to each type of associated factor to obtain the total load forecasting value of the energy management system.
[0045] In an optional embodiment of the present application, all associated factors of the load in the energy management system include the number of system orders, production volume, number of personnel, equipment type, equipment operating conditions, equipment efficiency, production plan, outdoor weather conditions, holidays, day, night, and power quality.
[0046] In an optional embodiment of the present application, classifying all associated factors according to the corresponding relationship between the associated factors and the power load data includes:
[0047] Obtain the power load data distribution in the energy management system, and determine the first power load data level, the second power load data level, and the third power load data level from large to small according to the power load data distribution;
[0048] The first type of associated factors corresponding to the first-level power load data include order quantity and production volume. The second type of associated factors corresponding to the second-level power load data include equipment type, equipment operating conditions, equipment efficiency, and outdoor weather conditions. The third type of associated factors corresponding to the third-level power load data include the number of personnel, outdoor weather conditions, and holidays. The associated factors other than the first type, the second type, and the third type of associated factors among all associated factors are used as the fourth type of associated factors.
[0049] In an optional embodiment of the present application, the main work of model construction is to split the complex model of the overall enterprise or large system according to the actual power load data distribution and in combination with the model organizational structure of the electricity-consuming unit. The model is split into one by one sub-models with relatively clear associated factors and relatively stable data sets. For the sub-models with relatively unclear associated factors, the traditional prediction algorithm of artificial neural network can still be used to improve their accuracy. There are many associated factors of power load, such as system order quantity, production volume, number of personnel, equipment type, equipment operating conditions, equipment efficiency, production plan, outdoor weather conditions, holidays, day and night, power quality, etc. A load prediction model with order quantity and production volume as the main associated factors is constructed for the production system; a load prediction model with equipment type, equipment operating conditions, equipment efficiency, and outdoor weather conditions as the main associated factors is constructed for the auxiliary production system; a load prediction model with the number of personnel, outdoor weather conditions, holidays, etc. as the main associated factors is constructed for the affiliated system. Secondary models can be defined under each classification, and finally the overall prediction result is obtained by aggregating each sub-model.
[0050] In an optional embodiment of the present application, load prediction models corresponding to each type of associated factors are constructed respectively, including:
[0051] The load prediction models corresponding to the first type, the second type, and the third type of associated factors are each any one of the long-short term memory network model, the gated recurrent unit network model, and the one-dimensional convolutional neural network model. The load prediction model corresponding to the fourth type of associated factors is an artificial neural network.
[0052] In an alternative embodiment of the present application, in the initial stage of the development of load forecasting, the mathematical algorithm theory is mainly used as the modeling basis. Traditional load forecasting treats power load data as time series data for processing, and the forecasting methods mainly include time series method, regression analysis method, grey prediction method, Kalman filtering method, etc. Under the current development trend, the types and quantities of power loads and their external influencing factors are increasing continuously. Coupled with the enhanced initiative and passive uncertainty of the load side brought about by the increasing proportion of new energy power generation, traditional load forecasting models are difficult to meet the high requirements of the new power system. Relying on its capabilities such as non-linear fitting, artificial intelligence technology has made breakthroughs in load modeling and forecasting, load optimization, etc. The intelligent algorithm model based on artificial intelligence technology can better capture the non-linear characteristics of current power loads, greatly improving the load forecasting accuracy and becoming the mainstream model of current load forecasting. For the recurrent neural network unit, the output of the recurrent neural network comprehensively considers the previous sequence features and the current sequence features. Such a special structure can effectively avoid the problems of information storage and reuse. Therefore, using the recurrent neural network as the basic unit to process time series tasks such as power load forecasting will endow the model with unique advantages. At present, long short-term memory network and gated recurrent unit have emerged in various scientific research fields and are maturely applied in engineering tasks. The conventional recurrent neural network is prone to losing the feature information from a long time ago and experiencing the phenomenon of gradient disappearance. Both the long short-term memory network and the gated recurrent unit can make up for this defect by designing a "gating" structure. Among them, the former is mainly composed of a forgetting gate, an input gate, cell state update, and an output gate. The forgetting gate discards the hidden cell state of the previous layer with a certain probability, thereby reducing some redundant information. For the one-dimensional convolutional neural network unit, the convolutional neural network has strong learning ability for data features. Its one-dimensional form can filter the time series signal in the form of a specific receptive field size and generate a feature map. Each feature map can be regarded as the convolution operation of different filters on the time series signal of the current time step. Due to the advantage of weight sharing, a convolutional kernel can traverse the local features of the time series signal using a preset kernel size, which greatly improves the efficiency of the feature extraction process and avoids the phenomenon of overfitting of the model. The basic convolutional unit includes a convolutional layer and an activation function.
[0053] In an alternative embodiment of the present application, the load forecasting models corresponding to each type of associated factor are trained separately, including:
[0054] Taking the order quantity and production volume as inputs and the load value as the output, training to obtain the load forecasting model corresponding to the first type of associated factor;
[0055] Taking the equipment type, equipment operating conditions, equipment efficiency, and outdoor weather conditions as inputs and the load value as the output, training to obtain the load forecasting model corresponding to the second type of associated factor;
[0056] Taking the number of personnel, outdoor weather conditions, and holidays as inputs and the load value as the output, a load prediction model corresponding to the third type of associated factors is trained.
[0057] Taking the associated factors other than the first type of associated factors, the second type of associated factors, and the third type of associated factors among all the associated factors as inputs and the load value as the output, a load prediction model corresponding to the fourth type of associated factors is trained.
[0058] In the load prediction method for an energy management system of the present application, the test data sources are extensive, including data from actual real projects and data from experiments. Most of the data comes from the actual application scenarios of real projects. The data volume is large, and there is sufficient accumulation in terms of both breadth and depth, solving the problem of low accuracy of existing load prediction methods.
[0059] In an optional embodiment of the present application, the method further includes:
[0060] For the first target type of associated factors whose complexity of the associated factors in each type of associated factors exceeds the first preset threshold, the load prediction model corresponding to each first target type of associated factors is split into multiple sub-models, and each sub-model is trained to predict the load through the trained sub-models.
[0061] In an optional embodiment of the present application, the method further includes:
[0062] For the second target type of associated factors whose data distribution uniformity of the associated factors in each type of associated factors exceeds the second preset threshold, the load prediction model corresponding to each second target type of associated factors is split into multiple sub-models, and each sub-model is trained to predict the load through the trained sub-models.
[0063] Table 1 below is an example of the load prediction method for an energy management system of the present application. By comparing the actual monthly overall load data and the overall load prediction data, the deviation of the energy consumption data predicted according to the load prediction method for an energy management system of the present application from the actual energy consumption data is controlled at about 10%, that is, the deviation is not large, and the prediction result can be further optimized after accumulating more historical data.
[0064] Table 1
[0065] Time <![CDATA[P 预测 (kWh)]]> <![CDATA[P 实际 (kWh)]]> Deviation January 441936 413897 6.77％ February 331452 337038 -1.66％ March 405108 407786 -0.66％ April 423522 422250 0.30％ May 430637 421430 2.18％ June 593549 633593 -6.32％ July 619802 672529 -7.84％ August 846876 812040 4.29％ September 639200 671923 -4.87％ October 534599 565833 -5.52％ November 429829 435138 -1.22％ December 325843 365377 -10.82％ Subtotal 6022353 6158833 4.01％
[0066] The load prediction method for an energy management system of the present application can improve the accuracy rate of the load prediction result to more than 90%, and can maintain a relatively stable output under various external environments and different internal working conditions throughout the year. There is more than a 20% improvement in both the prediction accuracy and prediction stability compared to traditional other methods.
[0067] Table 2 below is another example of the load forecasting method for the energy management system in this application. The data used comes from 5 air compressors of an enterprise. Among them, the rated power of No. 1-4 air compressors is 315 Kw, the nominal volume flow is 52.2 m 3 / min, the input specific power is 7 Kw / (m3 / min), and the rated working pressure is 0.8 MPa; the rated power of the 5# air compressor is 160 Kw, and the nominal volume flow is 28.92 m 3 / min, the input specific power is 6.5 Kw / (m 3 / min), and the rated working pressure is 0.8 MPa. This application mainly considers the impact of factors such as gas supply on demand, adjusting the operation strategies of No. 1-5 air compressors, and increasing pressure on the prediction data, as Figure 2 shown.
[0068] Table 2
[0069]
[0070]
[0071] For the load forecasting method of the energy management system in this application, the data set is split into multiple sub-data sets according to the power load distribution, and then multiple sub-forecasting models are established respectively for forecasting. The sub-forecasting models can be split again according to the complexity and data distribution. This can solve problems such as uneven distribution of power load data, complexity of a single model, and large computational volume, and can simplify complex problems and multi-dimensionalize a single model, thereby improving the forecasting accuracy.
[0072] It should be understood that although the steps in the flowchart are shown sequentially according to the arrows, these steps do not necessarily need to be executed sequentially in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the figure may include multiple sub-steps or multiple stages. These sub-steps or stages do not necessarily need to be executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages does not necessarily need to be sequential, but can be executed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.
[0073] Please refer to Figure 3 , an embodiment of this application provides a load forecasting device 300 for the energy management system, including:
[0074] An acquisition module 310, configured to acquire all associated factors of the load in the energy management system;
[0075] A classification module 320 is configured to classify all associated factors according to the corresponding relationship between the associated factors and the power load data;
[0076] A construction module 330 is configured to respectively construct and train a load prediction model corresponding to each category of associated factors;
[0077] A determination module 340 is configured to determine the load prediction value corresponding to each category of associated factors according to each trained load prediction model, sum up the load prediction values corresponding to each category of associated factors, and obtain the total load prediction value of the energy management system.
[0078] The load prediction device for an energy management system of the present application solves the problems of being too complex and having low accuracy existing in the existing prediction methods; on the premise of making full use of the existing prediction technologies, it can greatly improve the accuracy, effectiveness and reliability of the overall load prediction.
[0079] For the specific limitations of the above device 300, reference can be made to the limitations on the load prediction method for an energy management system in the foregoing, which will not be elaborated here. Each module in the above device 300 can be implemented in whole or in part by software, hardware and their combination. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory of the computer device in the form of software, so as to facilitate the processor to call and execute the operations corresponding to the above respective modules.
[0080] In one embodiment, a computer device is provided, and the internal structure diagram of the computer device can be as Figure 4 shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, 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 database of the computer device is used to store data. The network 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, it implements a load prediction method for an energy management system as described above. It includes: including a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, it implements any step in the load prediction method for an energy management system as described above.
[0081] 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, it can implement any step in the load prediction method for an energy management system as described above.
[0082] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.
[0083] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or the combination of blocks.
[0084] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or the combination of blocks.
[0085] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable devices provide steps for implementing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or the combination of blocks.
[0086] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they know the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present application.
[0087] Obviously, those skilled in the art can make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalent technologies, this application is also intended to cover these changes and modifications.
Claims
1. A load forecasting method for an energy management system, characterized in that: The method comprises: Obtain all relevant factors of load in energy management system; Classify all the related factors according to the corresponding relationship between the related factors and the power load data; Construct and train the load forecasting model corresponding to each type of correlation factor respectively; According to each trained load forecasting model, the load forecast value corresponding to each type of correlation factor is determined, and the load forecast value corresponding to each type of correlation factor is added up to obtain the total load forecast value of the energy management system.
2. The method according to claim 1, characterized in that All the associated factors of the load in the energy management system include the number of system orders, output, number of personnel, equipment type, equipment operating conditions, equipment efficiency, production plan, outdoor weather conditions, holidays, daytime, nighttime and power quality.
3. The method according to claim 2, characterized in that According to the corresponding relationship between the correlation factors and the power load data, all correlation factors are classified, including: Obtaining the power load data distribution in the energy management system, and determining a first power load data level, a second power load data level, and a third power load data level from large to small according to the power load data distribution; The first type of correlation factors corresponding to the first power load data level include order quantity and output, the second type of correlation factors corresponding to the second power load data level include equipment type, equipment operating conditions, equipment efficiency and outdoor weather conditions; the third type of correlation factors corresponding to the third power load data level include the number of personnel, outdoor weather conditions and holidays. All correlation factors except the first type of correlation factors, the second type of correlation factors and the third type of correlation factors are regarded as the fourth type of correlation factors.
4. The method according to claim 3, characterized in that The load forecasting model corresponding to each type of correlation factor is constructed separately, including: The load forecasting models corresponding to the first, second and third types of correlation factors are any one of the long-short term memory network model, the gated recurrent network model and the one-dimensional convolutional neural network model. The load forecasting model corresponding to the fourth type of correlation factor is an artificial neural network.
5. The method according to claim 3, characterized in that: The load forecasting model corresponding to each type of correlation factor is trained separately, including: With order quantity and output as input and load value as output, the load forecasting model corresponding to the first type of correlation factors is trained; The load forecasting model corresponding to the second type of correlation factors is trained by taking equipment type, equipment operating condition, equipment efficiency and outdoor weather conditions as input and load value as output; With the number of personnel, outdoor weather conditions and holidays as input and the load value as output, the load forecasting model corresponding to the third type of correlation factors is trained; With the correlation factors except the first type of correlation factors, the second type of correlation factors and the third type of correlation factors among all the correlation factors as input and the load value as output, the load forecasting model corresponding to the fourth type of correlation factors is trained.
6. The method according to claim 1, characterized in that The method further comprises: For the first target class association factors in each class of association factors whose complexity exceeds the first preset threshold, the load prediction model corresponding to each first target class association factor is split into multiple sub-models, and each sub-model is trained to predict the load through the trained sub-model.
7. The method according to claim 1, characterized in that The method further comprises: For the second target class association factors in each class of association factors whose data distribution uniformity exceeds the second preset threshold, the load prediction model corresponding to each second target class association factor is split into multiple sub-models, and each sub-model is trained to predict the load through the trained sub-model.
8. A load forecasting device for an energy management system, characterized in that: include: An acquisition module is used to acquire all associated factors of the load in the energy management system; A classification module, used for classifying all the associated factors according to the corresponding relationship between the associated factors and the power load data; A construction module is used to construct and train the load forecasting model corresponding to each type of correlation factor; The determination module is used to determine the load prediction value corresponding to each type of correlation factor according to each trained load prediction model, and add the load prediction value corresponding to each type of correlation factor to obtain the total load prediction value of the energy management system.
9. A computer device comprising: It comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the load forecasting method for an energy management system as claimed in any one of claims 1 to 7 are implemented.
10. 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 load forecasting method for an energy management system described in any one of claims 1 to 7 are implemented.