Power demand prediction method and system based on big data analysis

Abstract

The invention discloses an electric power demand prediction method and system based on big data analysis, relates to the technical field of electric power systems, and mainly aims to solve the problem that the existing electric power demand prediction accuracy and effectiveness are poor so as to save the electric power transmission effectiveness. Comprising the steps that load information to be demanded and time period average use data corresponding to the load information are determined, and the time period average use data comprise the average value of power dispatching use in multiple time periods; a corresponding demand prediction model is called based on the load information, the time period average use data is processed based on the demand prediction model, power demand reference data of the target time period is determined, and the demand prediction model is obtained by performing model training on a deep learning network based on multi-time-period use sample data; and determining a reference time period of the target time period, and adjusting the power demand reference data based on the power demand extreme value and the power demand average value of the reference time period to obtain a power demand prediction result.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and particularly to a power demand prediction method and system based on big data analysis. Background Art

[0002] In a power system, power demand is an index for power supply and transmission. Accurate prediction can avoid waste of electric power energy. Currently, the prediction of existing power demand is usually obtained through single calculation based on historical usage. However, historical usage cannot meet the changes in sudden power demand, nor can it meet the flexibility requirements for power demand in different scenarios, thus greatly reducing the prediction accuracy and effectiveness of power demand. Summary of the Invention

[0003] In view of this, the present invention provides a power demand prediction method and system based on big data analysis, mainly aiming to solve the problems of poor prediction accuracy and effectiveness of existing power demand, and to save the effectiveness of power transmission.

[0004] According to one aspect of the present invention, there is provided a power demand prediction method based on big data analysis, including:

[0005] Determine the load information to be demanded, and the period average usage data corresponding to the load information, where the period average usage data includes the average value of power dispatching usage within multiple time periods;

[0006] Retrieve a corresponding demand prediction model based on the load information, and process the period average usage data based on the demand prediction model to determine the power demand reference data for the target period, where the demand prediction model is obtained by training a deep learning network with multi-period usage sample data;

[0007] Determine the reference period for the target period, and adjust the power demand reference data based on the power demand extreme value and the power demand average value of the reference period to obtain the power demand prediction result.

[0008] Further, before retrieving a corresponding demand prediction model based on the load information and processing the period average usage data based on the demand prediction model to determine the power demand reference data for the target period, the method further includes:

[0009] Create a deep learning network including multiple neural network layers, and obtain multi-period usage sample data;

[0010] Extract the matching sample data from the multi-period usage sample data according to different load information, and perform model training on the deep learning network based on the sample data to obtain a demand prediction model that matches different load information and has completed model training. The load information includes load voltage, load current, load power, and load usage duration.

[0011] Further, the extracting the matching sample data from the multi-period usage sample data according to different load information includes:

[0012] Query the corresponding target time periods from the multi-period usage sample data according to multiple extreme values among the load voltage, the load current, the load power, and the load usage duration;

[0013] If the target time periods overlap, extract sample data according to the adjacent time periods of the target time periods;

[0014] If the target time periods do not overlap and the duration between multiple target time periods does not exceed a preset time length, extract sample data according to the target time periods;

[0015] If the target time periods do not overlap and the duration between multiple target time periods exceeds the preset time length, adjust the target time periods according to the preset time length, and extract sample data according to the adjusted target time periods.

[0016] Further, the performing model training on the deep learning network based on the sample data to obtain a demand prediction model that matches different load information and has completed model training includes:

[0017] Extract the sample data according to a first ratio, and perform model training on the deep learning network using the extracted first sample data. The weight values of multiple neural network layers in the deep learning network are configured based on the number of time periods of the extracted first sample data;

[0018] Extract the sample data according to a second ratio, and determine the model loss value of the deep learning network after model training using the extracted second sample data;

[0019] If the model loss value is less than a preset loss threshold, complete the model training of the deep learning network;

[0020] If the model loss value is greater than or equal to the preset loss threshold, re-execute the step of extracting the sample data according to the first ratio and performing model training on the deep learning network using the extracted first sample data.

[0021] Further, determining a reference time period within the target time period and adjusting the power demand reference data based on the extreme value of power demand and the average value of power demand in the reference time period to obtain a power demand prediction result includes:

[0022] In response to a time period adjustment request, determining a reference time period for the target time period according to the time period adjustment demand information carried in the time period adjustment request;

[0023] If the reference extreme value in the power demand reference data is greater than the power demand extreme value, and the reference average value in the reference time period is greater than the power demand average value, then perform a downward adjustment on the power demand reference data;

[0024] If the reference extreme value in the power demand reference data is less than or equal to the power demand extreme value, and the reference average value in the reference time period is less than or equal to the power demand average value, then perform an upward adjustment on the power demand reference data.

[0025] Further, the method further includes:

[0026] Outputting the power demand prediction result;

[0027] In response to the power demand prediction confirmation instruction, driving a power storage device or a power generation device to transmit power resources according to the power demand prediction result.

[0028] Further, after adjusting the power demand reference data based on the extreme value of power demand and the average value of power demand in the reference time period to obtain a power demand prediction result, the method further includes:

[0029] Obtaining power transmission loss information of the power system, and calculating transmission power data of different power transmission paths in the power system based on the power demand prediction result and the power transmission loss information, so as to generate a power transmission instruction based on the transmission power data to request transmission power.

[0030] According to another aspect of the present invention, there is provided a power demand prediction system based on big data analysis, including:

[0031] A determination module, configured to determine load information to be demanded, and average usage data corresponding to the load information, where the average usage data in the time period includes an average value of power dispatching usage in a plurality of time periods;

[0032] A processing module, configured to retrieve a corresponding demand prediction model based on the load information, and process the average usage data of the time period based on the demand prediction model to determine the reference data of the power demand for the target time period, where the demand prediction model is obtained by training a deep learning network based on multi-time period usage sample data;

[0033] An adjustment module, configured to determine a reference time period for the target time period, and adjust the power demand reference data based on the extreme value and average value of the power demand in the reference time period to obtain a power demand prediction result.

[0034] According to another aspect of the present invention, there is provided a storage medium storing at least one executable instruction, and the executable instruction causes a processor to perform operations corresponding to the above-mentioned power demand prediction method based on big data analysis.

[0035] According to still another aspect of the present invention, there is provided a terminal, including: a processor, a memory, a communication interface, and a communication bus, and the processor, the memory, and the communication interface complete communication with each other through the communication bus;

[0036] The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the above-mentioned power demand prediction method based on big data analysis.

[0037] By means of the above technical solutions, the technical solutions provided by the embodiments of the present invention at least have the following advantages:

[0038] The present invention provides a power demand prediction method and system based on big data analysis. Compared with the prior art, the embodiments of the present invention determine the load information to be demanded and the average usage data of the time period corresponding to the load information, where the average usage data of the time period includes the average value of power dispatching usage in multiple time periods; retrieve a corresponding demand prediction model based on the load information, and process the average usage data of the time period based on the demand prediction model to determine the reference data of the power demand for the target time period, where the demand prediction model is obtained by training a deep learning network based on multi-time period usage sample data; determine a reference time period for the target time period, and adjust the power demand reference data based on the extreme value and average value of the power demand in the reference time period to obtain a power demand prediction result, realizing prediction analysis based on the power consumption of different time periods and using the power consumption of multiple time periods as the prediction basis, greatly improving the accuracy of power demand prediction for different loads, thereby ensuring the accuracy and effectiveness of power transmission.

[0039] The above description is only an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention, it can be implemented according to the content of the specification. And in order to make the above and other objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention are specifically exemplified below. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0041] Figure 1 shows a flowchart of a power demand prediction method based on big data analysis provided by an embodiment of the present invention;

[0042] Figure 2 shows a schematic diagram of a deep learning network structure provided by an embodiment of the present invention;

[0043] Figure 3 shows a schematic diagram of a power demand display interface provided by an embodiment of the present invention;

[0044] Figure 4 shows a block diagram of a power demand prediction system based on big data analysis provided by an embodiment of the present invention;

[0045] Figure 5 shows a schematic diagram of the structure of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0047] An embodiment of the present invention provides a power demand prediction method based on big data analysis, as Figure 1 shown, the method includes:

[0048] 101. Determine the load information to be demanded and the average usage data of the corresponding time period of the load information.

[0049] In the embodiments of the present invention, the current execution entity, as the execution end for predicting power demand, can be the processing server of the power system or the processor terminal of the power system, so as to obtain the load information of the to-be-demanded in real time. At this time, the load information is the specific information of the terminal device of the demanded power quantity. The terminal device includes, but is not limited to, different forms of electrical equipment, etc. For example, residential electrical equipment, industrial electrical equipment, etc. The load information can be the time of residential power consumption, voltage, current, and total power quantity, etc. For example, the load information includes load voltage, load current, load power, and load usage duration. The embodiments of the present application do not make specific limitations. At the same time, the current execution end obtains the period average usage data corresponding to the load information. At this time, the period average usage data includes the average value of power dispatching usage within multiple time periods, that is, after each terminal as a load is divided according to historical time periods, the average value obtained by dividing according to time periods. For example, the average power consumption used on the 16th is calculated every 2 hours, that is, 12 average power consumption values are obtained as the period average usage data on the 16th.

[0050] It should be noted that when obtaining the load information, it can be input by power operators based on the power system, or obtained through the default value calculated in the power system. The embodiments of the present invention do not make specific limitations.

[0051] 102. Retrieve the corresponding demand prediction model based on the load information, and process the period average usage data based on the demand prediction model to determine the power demand reference data for the target period.

[0052] In the embodiments of the present invention, different demand prediction models used by different load terminals are pre-trained in the current execution end to achieve the purpose of accurate prediction. Therefore, the current execution end can retrieve the corresponding demand prediction model according to the load information, and process the period average usage data based on this demand prediction model to determine the power demand reference data for the target period. Among them, the demand prediction model is obtained by training a deep learning network based on multi-period usage sample data, so as to use the deep learning network to learn the reference demand situation of each period from the multi-period usage sample data and predict the power demand data for the target period.

[0053] 103. Determine the reference period for the target period, and adjust the power demand reference data based on the power demand extreme value and power demand average value of the reference period to obtain the power demand prediction result.

[0054] In an embodiment of the present invention, after the current execution end obtains the power demand reference data for the target time period, a reference time period is determined based on the target time period. At this time, the parameter time period is the basis time period for adjusting the power demand reference data. For example, the target time period can be from 1:00 to 2:00 am every day, and the corresponding reference time period can be from 0:00 to 3:00 am on a certain day, or from 2:00 to 3:00 am every day, etc. The embodiments of the present invention do not make specific limitations. Furthermore, the power demand extreme value and the power demand average value of this reference time period are retrieved to adjust the power demand reference data, and the final power demand prediction result is obtained. At this time, the power demand extreme value is the maximum and minimum values of the power transmission recorded at the reference time period, and the power demand average value is calculated from the completed power transmission data recorded at the reference time period, which can be retrieved from the historical database in the power system. The embodiments of the present invention do not make specific limitations.

[0055] It should be noted that in the embodiments of the present invention, in order to make the power demand prediction more accurate, when adjusting the power demand reference data using the power demand extreme value and the power demand skin keyword, the reference data can be increased or decreased according to the extreme value, and the reference data can also be increased or decreased according to the average value. The embodiments of the present invention do not make specific limitations.

[0056] In another embodiment of the present invention, for further limitation and explanation, before the step of retrieving the corresponding demand prediction model based on the load information and processing the time period average usage data based on the demand prediction model to determine the power demand reference data for the target time period, the method further includes:

[0057] Create a deep learning network including multiple neural network layers and obtain multi-time period usage sample data;

[0058] Extract matching sample data from the multi-time period usage sample data according to different load information, and train the deep learning network based on the sample data to obtain a demand prediction model that matches different load information and has completed model training.

[0059] In order to effectively predict the power demand based on artificial intelligence, the current execution end pre-creates a deep learning network that needs to be learned to learn based on the multi-time period usage sample data. Among them, when creating the deep learning network, it is constructed based on multiple neural network layers, specifically a 3-layer neural network, such as Figure 2The deep learning network shown, where the input can be the electricity consumption extracted from multiple target time periods. Correspondingly, the output is the electricity demand for the expected time period (the expected time period can be an entire time period at this time), and w1, w2, and w3 are the weights of each network layer. At the same time, the current execution end obtains the electricity consumption data of different time periods recorded in the power system as multi-time period usage sample data. For example, the electricity consumption (which can also be understood as the electricity transmitted by the power system) from 1:00 to 2:00, 2:00 to 3:00, 3:00 to 4:00, etc. on the 1st is a, and the embodiments of the present invention do not make specific limitations. Furthermore, in order to adapt to different load terminals, the current execution end first extracts the matching sample data from the multi-time period usage sample data based on different load information, and uses this sample data to train the model of the deep learning network to obtain a demand prediction model that matches different load information.

[0060] It should be noted that since the load information is the electricity consumption situation recorded by different load terminals, the load information includes load voltage, load current, load power, and load usage duration. The load voltage is the voltage value when the load terminal uses electricity, the load current is the current value when the load terminal uses electricity, the load power is the rated power when the load terminal uses electricity, and the load usage duration is the total electricity usage duration of the load terminal, which can be obtained from the historical records in the power system, and the embodiments of the present invention do not make specific limitations.

[0061] In another embodiment of the present invention, for further limitation and explanation, the steps of extracting the matching sample data from the multi-time period usage sample data according to different load information include:

[0062] Query the corresponding target time period from the multi-time period usage sample data according to multiple extreme values of the load voltage, the load current, the load power, and the load usage duration;

[0063] If the target time periods overlap, extract the sample data according to the adjacent time periods of the target time periods;

[0064] If the target time periods do not overlap and the duration between multiple target time periods does not exceed the preset time length, extract the sample data according to the target time periods;

[0065] If the target time periods do not overlap and the duration between multiple target time periods exceeds the preset time length, adjust the target time periods according to the preset time length and extract the sample data according to the adjusted target time periods.

[0066] For the purpose of extracting samples from multi-segment usage sample data to improve the learning accuracy of artificial intelligence algorithms, thereby enhancing the accuracy of predicting power demand, the current execution end first determines multiple extreme values among the load voltage, load current, load power, and load usage duration in the load information, and searches for the corresponding target time from the multi-period usage sample data according to the multiple extreme values. At this time, the target period is the period corresponding to the determined extreme value, which can be one or multiple. For example, if the extreme value of the load voltage appears at 2:10 and the extreme value of the load current appears at 3:25, the corresponding target periods are 2, which can be from 2 o'clock to 3 o'clock and from 3 o'clock to 4 o'clock. The embodiments of the present invention do not make specific limitations.

[0067] In some embodiments, since a period has a time length, after querying multiple target periods from the multi-period usage sample data, first determine whether there is an overlap among the multiple target periods. If there is an overlap among the target periods, extract the sample data according to the target periods and the corresponding adjacent periods, that is, not only extract the sample data from the target periods, but also extract from the adjacent periods. At this time, the adjacent periods include the previous period and the next period of the target period. For example, if the target period is from 4 o'clock to 5 o'clock, the adjacent periods include the periods from 2 o'clock to 3 o'clock and from 5 o'clock to 6 o'clock. The embodiments of the present invention do not make specific limitations. If there is no overlap among the multiple target periods, at the same time, determine whether the duration among the multiple target periods exceeds a preset time length. If the duration among the multiple target periods does not exceed the preset time length, extract the sample data according to the target periods. At this time, the preset time length is used to limit the period for extracting samples to avoid too long a period. Therefore, it can be set to 1 day, 2 days, etc. The embodiments of the present invention do not make specific limitations. In addition, if there is no overlap among the multiple target periods and the duration among the multiple target periods exceeds the preset time length, in order to avoid redundancy in sample extraction, adjust the target periods according to the preset time length and extract the sample data according to the adjusted target periods. At this time, when adjusting the target periods using the preset time length, the multiple target periods can be directly intercepted according to the preset time length to obtain the periods for extracting sample data. For example, if the preset time length is 1 day, the target period with a duration of 1 day can be intercepted from the starting first target period as the final target period for extracting sample data. The embodiments of the present invention do not make specific limitations.

[0068] It should be noted that the length of the period in the embodiments of the present invention can be configured based on different prediction scenarios for power demand, including but not limited to 1 hour, 2 hours, or 1 day, 2 days, etc. The embodiments of the present invention do not make specific limitations. In addition, when extracting sample data from the target periods, it can be randomly selected or selected according to a ratio. The embodiments of the present invention do not make specific limitations.

[0069] In another embodiment of the present invention, for further limitation and illustration, the step of performing model training on the deep learning network based on the sample data to obtain a demand prediction model that matches different load information and has completed model training includes:

[0070] Extract the sample data according to a first ratio, and use the extracted first sample data to perform model training on the deep learning network. The weight values of multiple neural network layers in the deep learning network are configured based on the number of time periods of the extracted first sample data;

[0071] Extract the sample data according to a second ratio, and use the extracted second sample data to determine the model loss value of the deep learning network after model training;

[0072] If the model loss value is less than a preset loss threshold, the model training of the deep learning network is completed;

[0073] If the model loss value is greater than or equal to the preset loss threshold, re-execute the step of extracting the sample data according to the first ratio and using the extracted first sample data to perform model training on the deep learning network.

[0074] In order to achieve precise training of the deep learning network and thus adapt to power demand prediction scenarios with different loads, when the current execution end performs model training, it extracts the sample data according to the first ratio and uses the extracted first sample data as the training sample to perform model training on the deep learning network. At the same time, it extracts the sample data according to the second ratio and uses the extracted second sample data as the validation sample to determine the model loss value of the deep learning network after model training. At this time, the loss value can be calculated based on the loss function, and the loss function includes but is not limited to L1 loss, mean square error, cross entropy, etc., which are not specifically limited in the embodiments of the present invention. Furthermore, if the calculated model loss value is less than the preset loss threshold, the model training of this deep learning network is completed. If the model loss value is greater than or equal to the preset loss threshold, re-execute the step of extracting the sample data according to the first ratio and using the extracted first sample data to perform model training on the deep learning network for the next iteration training. In addition, the weight values of multiple neural network layers in the deep learning network in the embodiments of the present invention are configured based on the number of time periods of the extracted first sample data. Specifically, first, determine the number of time periods for the target time period of extracting the sample data. For example, there are 5 target time periods that can all be used as the time periods for extracting the sample data, so the number of time periods is 5. When configuring the weights for the deep learning network constructed by a three-layer neural network, they are respectively configured as 1 / 5, 3 / 5, 1 / 5. In the embodiments of the present invention, the weight values of the first layer and the last layer are set to 1 / a, where a is the number of time periods.

[0075] In another embodiment of the present invention, for further limitation and illustration, the steps of determining the reference time period within the target time period and adjusting the power demand reference data based on the extreme value and average value of the power demand within the reference time period to obtain the power demand prediction result include:

[0076] In response to the time period adjustment request, determine the reference time period of the target time period according to the time period adjustment demand information carried in the time period adjustment request;

[0077] If the reference extreme value in the power demand reference data is greater than the power demand extreme value, and the reference average value of the reference time period is greater than the power demand average value, then perform a downward adjustment on the power demand reference data;

[0078] If the reference extreme value in the power demand reference data is less than or equal to the power demand extreme value, and the reference average value of the reference time period is less than or equal to the power demand average value, then perform an upward adjustment on the power demand reference data.

[0079] To meet the flexible requirements for the effectiveness of power demand forecasting, when the current execution end determines the reference time period and adjusts the power demand reference data, specifically, the operator first triggers a time period adjustment request through the power system. After the current execution end receives the time period adjustment request, the reference time period of the target time period is determined based on the time period adjustment requirement information carried in the time period adjustment request. At this time, the time period adjustment requirement information is used to represent the specific time period for which the target time period needs to be adjusted. For example, if the target time period is from 2 o'clock to 3 o'clock, the carried time period adjustment requirement information can be from 1 o'clock to 2 o'clock every day. Therefore, from 1 o'clock to 2 o'clock every day is used as the reference time period for the target time period from 2 o'clock to 3 o'clock, and the reference average value of this reference time period is obtained. Furthermore, the current execution end uses the reference extreme value in the power demand reference data to judge with the power demand extreme value. At the same time, the reference average value of the reference time period and the power demand average value are compared to ensure that the power demand value will not be too large or too small during the prediction process to meet the purpose of accurate prediction. Correspondingly, if the reference extreme value in the power demand reference data is greater than the power demand extreme value, and the reference average value of the reference time period is greater than the power demand average value, it indicates that the power demand reference data is too large. Therefore, the power demand reference data is adjusted downward. At this time, the downward adjustment can be made according to a percentage, that is, a threshold of 10% is preset as the limit of the downward adjustment. After a 10% reduction, the reference extreme value and the power demand extreme value in the power demand reference data are compared again. If the reference extreme value in the power demand reference data is less than or equal to the power demand extreme value, and the reference average value of the reference time period is less than or equal to the power demand average value, then the power demand reference data is adjusted upward. At this time, the upward adjustment can also be made according to a percentage, that is, a threshold of 10% is preset as the limit of the upward adjustment. After a 10% increase, the reference extreme value and the power demand extreme value in the power demand reference data are compared again.

[0080] In another embodiment of the present invention, for further limitation and explanation, the steps further include:

[0081] Output the power demand forecasting result;

[0082] In response to the power demand forecasting confirmation instruction, drive the power storage device or the power generation device to transmit power resources according to the power demand forecasting result.

[0083] To meet the requirements for the security and accuracy of different power transmissions, after the current execution end obtains the power demand forecasting result, the current execution end outputs the power demand forecasting result on the front-end interface of the power system so that the operator can view it. Since the power demand forecasting result is for the target time period, when outputting the power demand forecasting result, such as Figure 3As shown, the corresponding power demand prediction results are output according to different time periods and presented in the form of a bar chart, which greatly improves the visualization effect of the power demand prediction results. Correspondingly, a confirmation button is configured on the current execution end based on the front-end interface of the power system for the operator to confirm and trigger after verifying the power demand prediction. Therefore, after the operator views it, the current execution end can receive the triggered power demand prediction confirmation instruction, so that the current execution end can drive the storage device or power generation device to transmit power resources according to the power demand prediction result.

[0084] In some embodiments, the current execution end serves as the server of the power system. After determining to transmit the power according to the power demand prediction result, it can transmit power to the load terminal through the connected power storage device, such as industrial batteries or capacitors, etc., and can also transmit power to the load terminal through the connected power generation device. The embodiments of the present invention do not make specific limitations.

[0085] In another embodiment of the present invention, for further limitation and explanation, after adjusting the power demand reference data based on the extreme value and average value of the power demand in the reference time period to obtain the power demand prediction result, the method further includes:

[0086] Obtain the power transmission loss information of the power system, and calculate the transmission power data of different power transmission paths in the power system based on the power demand prediction result and the power transmission loss information, so as to generate a power transmission instruction based on the transmission power data to request the transmission of power.

[0087] To avoid the situation that the power loss generated during power transmission affects the reduction of the power finally received by the load terminal and improve the prediction accuracy of the power demand, after obtaining the power demand prediction result, the current execution end first obtains the power transmission loss information of the power system. Among them, the power transmission loss information is the line loss power of different transmission routes in the power system, that is, the material of different lines and the transmission distance will both cause a certain amount of power loss. Therefore, the specific power loss generated between the power transmission end and the load end can be calculated, and then the transmission power data of different power transmission paths in the power system can be calculated based on the power demand prediction result and this power transmission loss information. In an embodiment scenario, for the calculation of the transmission power data, it can be calculated based on the transmission power formula, and the transmission power formula is expressed as: where δ is the power transmission loss value of the i-th transmission path, q i is the power demand prediction result of the i-th transmission path, and h is all transmission paths.

[0088] An embodiment of the present invention provides a power demand forecasting method based on big data analysis. In the embodiment of the present invention, the load information to be demanded and the corresponding period average usage data are determined, and the period average usage data includes the average value of power dispatching usage in multiple time periods; based on the load information, the corresponding demand forecasting model is retrieved, and the period average usage data is processed based on the demand forecasting model to determine the power demand reference data for the target period. The demand forecasting model is obtained by training a deep learning network based on multi-period usage sample data; the reference period for the target period is determined, and the power demand reference data is adjusted based on the power demand extreme value and the power demand average value of the reference period to obtain the power demand forecasting result, realizing forecasting analysis based on the power consumption in different periods and using the multi-period power consumption as the forecasting basis, greatly improving the accuracy of power demand forecasting for different loads, thereby ensuring the accuracy and effectiveness of power transmission.

[0089] Further, for the implementation of the method described above Figure 1 An embodiment of the present invention provides a power demand forecasting system based on big data analysis, as Figure 4 shown. The system includes:

[0090] A determination module 21, configured to determine the load information to be demanded and the corresponding period average usage data, where the period average usage data includes the average value of power dispatching usage in multiple time periods;

[0091] A processing module 22, configured to retrieve the corresponding demand forecasting model based on the load information, and process the period average usage data based on the demand forecasting model to determine the power demand reference data for the target period. The demand forecasting model is obtained by training a deep learning network based on multi-period usage sample data;

[0092] An adjustment module 23, configured to determine the reference period for the target period, and adjust the power demand reference data based on the power demand extreme value and the power demand average value of the reference period to obtain the power demand forecasting result.

[0093] An embodiment of the present invention provides a power demand forecasting system based on big data analysis. In the embodiment of the present invention, the load information to be demanded and the average usage data corresponding to the load information are determined, and the average usage data for the time period includes the average value of power dispatching usage in multiple time periods; based on the load information, the corresponding demand forecasting model is retrieved, and the average usage data for the time period is processed based on the demand forecasting model to determine the power demand reference data for the target time period, where the demand forecasting model is obtained by training a deep learning network based on multi-time period usage sample data; the reference time period for the target time period is determined, and the power demand reference data is adjusted based on the extreme value and average value of the power demand in the reference time period to obtain the power demand forecasting result, realizing forecasting and analysis based on the power consumption in different time periods, and using the power consumption in multiple time periods as the forecasting basis, greatly improving the accuracy of power demand forecasting for different loads, thereby ensuring the accuracy and effectiveness of power transmission.

[0094] According to an embodiment of the present invention, a storage medium is provided. The storage medium stores at least one executable instruction, and the computer executable instruction can execute the power demand forecasting method based on big data analysis in any of the above method embodiments.

[0095] Figure 5 The structure diagram of a terminal provided according to an embodiment of the present invention is shown. The specific implementation of the terminal is not limited in the specific embodiment of the present invention.

[0096] As Figure 5 shown, the terminal may include: a processor 302, a communication interface 304, a memory 306, and a communication bus 308.

[0097] Among them: the processor 302, the communication interface 304, and the memory 306 communicate with each other through the communication bus 308.

[0098] The communication interface 304 is used to communicate with network elements of other devices such as clients or other servers.

[0099] The processor 302 is used to execute the program 310, and specifically can execute the relevant steps in the above embodiment of the power demand forecasting method based on big data analysis.

[0100] Specifically, the program 310 may include program code, and the program code includes computer operation instructions.

[0101] The processor 302 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. One or more processors included in the terminal may be of the same type, such as one or more CPUs; or may be of different types, such as one or more CPUs and one or more ASICs.

[0102] The memory 306 is used to store the program 310. The memory 306 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk memory.

[0103] The program 310 is specifically configured to cause the processor 302 to perform the following operations:

[0104] Determine the load information to be demanded and the period average usage data corresponding to the load information, where the period average usage data includes the average value of power dispatching usage within multiple time periods;

[0105] Retrieve the corresponding demand prediction model based on the load information, and process the period average usage data based on the demand prediction model to determine the power demand reference data for the target period. The demand prediction model is obtained by training a deep learning network based on multi-period usage sample data;

[0106] Determine the reference period for the target period, and adjust the power demand reference data based on the extreme value and average value of the power demand during the reference period to obtain the power demand prediction result.

[0107] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general computing system. They can be concentrated on a single computing system or distributed on a network composed of multiple computing systems. Optionally, they can be implemented by program codes executable by the computing system, so that they can be stored in the storage system and executed by the computing system. And in some cases, the steps shown or described herein can be executed in a different order, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. In this way, the present invention is not limited to any specific combination of hardware and software.

[0108] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A power demand forecasting method based on big data analysis, characterized in that, Including: Determine the load information to be demanded, and the period average usage data corresponding to the load information, where the period average usage data includes the average value of power dispatching usage within multiple time periods; Retrieve the corresponding demand prediction model based on the load information, and process the period average usage data based on the demand prediction model to determine the power demand reference data for the target period. The demand prediction model is obtained by training a deep learning network based on multi-period usage sample data; Determine the reference period for the target period, and adjust the power demand reference data based on the power demand extreme value and the average power demand of the reference period to obtain the power demand prediction result.

2. The method according to claim 1, wherein Before the step of retrieving the corresponding demand prediction model based on the load information and processing the period average usage data based on the demand prediction model to determine the power demand reference data for the target period, the method further includes: Create a deep learning network including multiple neural network layers, and obtain multi-period usage sample data; Extract matching sample data from the multi-period usage sample data according to different load information, and train the deep learning network based on the sample data to obtain demand prediction models that match different load information and have completed model training. The load information includes load voltage, load current, load power, and load usage duration.

3. The method according to claim 2, wherein The step of extracting matching sample data from the multi-period usage sample data according to different load information includes: Query the corresponding target period from the multi-period usage sample data according to multiple extreme values of the load voltage, the load current, the load power, and the load usage duration; If the target periods overlap, extract sample data according to the adjacent periods of the target periods; If the target periods do not overlap and the duration between multiple target periods does not exceed a preset time length, extract sample data according to the target periods; If the target periods do not overlap and the duration between multiple target periods exceeds the preset time length, adjust the target periods according to the preset time length, and extract sample data according to the adjusted target periods.

4. The method according to claim 2, wherein The step of training the deep learning network based on the sample data to obtain demand prediction models that match different load information and have completed model training includes: Extract the sample data according to the first ratio, and use the extracted first sample data to train the deep learning network. The weight values of multiple neural network layers in the deep learning network are configured based on the number of time periods of the extracted first sample data; Extract the sample data according to the second ratio, and use the extracted second sample data to determine the model loss value of the deep learning network after model training; If the model loss value is less than the preset loss threshold, complete the model training of the deep learning network. If the model loss value is greater than or equal to a preset loss threshold, the steps of extracting the sample data according to a first ratio and training the deep learning network with the extracted first sample data are re-executed.

5. The method according to claim 1, characterized in that, Determining the reference time period corresponding to the target time period, and adjusting the power demand reference data based on the extreme value of power demand and the average value of power demand in the reference time period to obtain the power demand prediction result includes: In response to a time period adjustment request, determining the reference time period of the target time period according to the time period adjustment demand information carried in the time period adjustment request; If the reference extreme value in the power demand reference data is greater than the power demand extreme value and the reference average value in the reference time period is greater than the power demand average value, the power demand reference data is adjusted downward; If the reference extreme value in the power demand reference data is less than or equal to the power demand extreme value and the reference average value in the reference time period is less than or equal to the power demand average value, the power demand reference data is adjusted upward.

6. The method according to claim 1, wherein The method further includes: Outputting the power demand prediction result; In response to the power demand prediction confirmation instruction, driving a power storage device or a power generation device to transmit power resources according to the power demand prediction result.

7. The method according to any one of claims 1-6, characterized in that After adjusting the power demand reference data based on the extreme value of power demand and the average value of power demand in the reference time period to obtain the power demand prediction result, the method further includes: Obtaining the power transmission loss information of the power system, and calculating the transmission power quantity data of different power transmission paths in the power system based on the power demand prediction result and the power transmission loss information, so as to generate a power transmission instruction based on the transmission power quantity data to request the transmission power quantity.

8. A power demand forecasting system based on big data analysis, characterized in that, Including: A determination module, configured to determine the load information to be demanded and the corresponding time period average usage data of the load information, where the time period average usage data includes the average value of power dispatching usage in multiple time periods; A processing module, configured to retrieve a corresponding demand prediction model based on the load information, and process the time period average usage data based on the demand prediction model to determine the power demand reference data of the target time period, where the demand prediction model is obtained by training a deep learning network with multi-time period usage sample data; An adjustment module, configured to determine the reference time period corresponding to the target time period, and adjust the power demand reference data based on the extreme value of power demand and the average value of power demand in the reference time period to obtain the power demand prediction result.

9. A storage medium, in which at least one executable instruction is stored, and the executable instruction causes a processor to execute the operations corresponding to the power demand prediction method based on big data analysis according to any one of claims 1-7.

10. A terminal, comprising: A processor, a memory, a communication interface and a communication bus, and the processor, the memory and the communication interface complete communication with each other through the communication bus; The memory is used to store at least one executable instruction, and the executable instruction causes the processor to perform operations corresponding to the power demand prediction method based on big data analysis according to any one of claims 1-7.

Images