A method and system for regional power distribution load scheduling

Abstract

The present invention discloses a method and system for regional power distribution load scheduling, which comprises: performing image recognition on a remote sensing image of a target area to obtain vegetation coverage information of the target area; predicting a vegetation irrigation quantity sequence of the target area according to the vegetation coverage information and weather data of a time period to be predicted; screening out historical energy load data that matches the vegetation irrigation quantity sequence in a time dimension from a historical energy load database of the target area; obtaining predicted energy load data through an energy load prediction model according to the vegetation irrigation quantity sequence and the historical energy load data; obtaining predicted power grid load data through a power grid load prediction model according to the predicted energy load data; and performing power distribution load scheduling on the target area according to the predicted power grid load data within the time period to be predicted; the present invention fully considers the influence of regional vegetation on regional power load, can effectively improve the accuracy of regional power grid load data prediction, and thus improves the regional power distribution scheduling effect.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution dispatching, and in particular to a method and system for regional power distribution load dispatching. Background Art

[0002] With the development of urbanization and urbanization, the sources and amount of carbon dioxide emissions have increased year by year, and the environmental problems in cities have become increasingly serious. Green plants can release oxygen, fix carbon, and purify the air. Large-scale deployment of garden greening devices and increasing urban vegetation area are beneficial to solving urban environmental problems. Regional vegetation is closely related to the energy load of the region. For example, the size of the vegetation area affects the temperature of the region, thereby affecting the use of equipment such as air conditioners, resulting in changes in regional energy load. For example, with the intelligent management of vegetation, regional vegetation requires large-scale deployment of garden greening devices for 24-hour monitoring and management, resulting in changes in regional energy load. Vegetation irrigation of regional vegetation is the main link in the entire vegetation monitoring and management, and it is also the link with the highest energy consumption, which affects the energy load of the region. However, the current regional power grid load dispatching generally only considers factors such as population density and industrial activities in the region, and does not consider the impact of regional vegetation on the energy load of the region. It is impossible to accurately estimate the regional electricity consumption data, thereby affecting the regional power distribution dispatching effect. Summary of the invention

[0003] In view of the problems existing in the prior art, the embodiments of the present invention provide a regional power distribution load scheduling method and system, which fully considers the impact of regional vegetation on regional power load, can effectively improve the accuracy of regional power grid load data prediction, and thus improve the regional power distribution scheduling effect.

[0004] In a first aspect, an embodiment of the present invention provides a method for regional power distribution load scheduling, including:

[0005] Acquire remote sensing images of the target area;

[0006] Performing image recognition on the remote sensing image to obtain vegetation coverage information of the target area;

[0007] Predicting a vegetation irrigation amount sequence for the target area based on the vegetation coverage information and weather data for the time period to be predicted;

[0008] Filtering out historical energy load data that matches the vegetation irrigation amount sequence in the time dimension from a historical energy load database of the target area;

[0009] According to the vegetation irrigation amount sequence and the historical energy load data, using an energy load prediction model, obtaining predicted energy load data;

[0010] According to the predicted energy load data, using a power grid load prediction model, predicted power grid load data is obtained;

[0011] During the time period to be predicted, distribution load is dispatched for the target area according to the predicted power grid load data.

[0012] As an improvement of the above solution, performing image recognition on the remote sensing image to obtain vegetation coverage information of the target area includes:

[0013] Performing denoising processing on the remote sensing image;

[0014] Perform single-band extraction on the denoised remote sensing image to obtain infrared band images and near-infrared band images;

[0015] Generating a vegetation coverage map of the target area by normalizing vegetation index according to the infrared band image and the near-infrared band image;

[0016] Performing texture recognition on the denoised remote sensing image using a pre-built image texture recognition model to obtain a vegetation texture map of at least one vegetation type;

[0017] Calculating the vegetation coverage degree of the corresponding vegetation type according to the vegetation coverage map and each of the vegetation texture maps;

[0018] The vegetation coverage information is obtained according to the vegetation types and the vegetation coverage degrees of each of the vegetation types.

[0019] As an improvement of the above solution, the step of generating a vegetation coverage map of the target area by normalizing vegetation index according to the infrared band image and the near-infrared band image includes:

[0020] Performing color correction on the infrared band image and the near infrared band image;

[0021] Calculating a first reflectivity of the infrared band image according to the pixel value of the corrected infrared band image and the pixel value of the standard grayscale image;

[0022] Calculating a second reflectivity of the near-infrared band image according to the pixel value of the corrected near-infrared band image and the pixel value of the standard grayscale image;

[0023] Calculating the normalized vegetation index of each pixel pair according to the first reflectivity of the infrared band image and the second reflectivity of the near-infrared band image; wherein each pixel pair includes pixels located at the same position in the infrared band image and the near-infrared band image;

[0024] A vegetation coverage map of the target area is generated according to the vegetation indexes of all the pixel pairs.

[0025] As an improvement of the above solution, the step of calculating the vegetation coverage degree of the corresponding vegetation type according to the vegetation coverage map and each of the vegetation texture maps includes:

[0026] For each of the vegetation texture maps, projecting the vegetation texture map onto the vegetation coverage map;

[0027] The pixel ratio of the vegetation texture in the vegetation texture map in the vegetation coverage map is calculated as the vegetation coverage degree of the corresponding vegetation type.

[0028] As an improvement of the above solution, predicting the vegetation irrigation amount sequence of the target area according to the vegetation coverage information and the weather data of the time period to be predicted includes:

[0029] Querying the basic transpiration rate of each vegetation type in the vegetation coverage information;

[0030] Calculating the actual coverage area of ​​each vegetation type according to the vegetation coverage degree of each vegetation type and the area of ​​the target area;

[0031] Calculate multiple temperature influencing factors according to the temperature data in the weather data at a preset time granularity;

[0032] Calculate multiple light intensity influencing factors according to the light intensity data in the weather data and at the time granularity;

[0033] Calculating multiple wind speed influencing factors according to the wind speed data in the weather data and at the time granularity;

[0034] Calculating a plurality of humidity influencing factors according to the time granularity based on the air humidity data in the weather data;

[0035] For each of the vegetation types, according to the basic transpiration rate of the vegetation type, the actual coverage area, the multiple temperature influencing factors, the multiple light intensity influencing factors, the multiple wind speed influencing factors, and the multiple humidity influencing factors, the transpiration rate sequence of the corresponding vegetation type is calculated;

[0036] Generate a rainfall sequence according to the time granularity based on the rainfall data in the weather data;

[0037] The vegetation irrigation amount sequence of the target area is calculated based on the transpiration rate sequence of all the vegetation types and the rainfall sequence.

[0038] As an improvement of the above solution, the predicted energy load data is obtained through an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data, including:

[0039] Calculating a first Z score value of each vegetation irrigation amount in the vegetation irrigation amount sequence, and performing a data outlier check on each of the first Z score values;

[0040] Calculating a second Z score value of each historical energy load in the historical energy load data, and performing a data outlier check on each of the second Z score values;

[0041] Normalizing the first Z score values ​​that pass the verification, and sorting the normalized first Z score values ​​in chronological order to generate a first standard data sequence;

[0042] Normalizing the verified second Z score values, and sorting the normalized second Z score values ​​in chronological order to generate a second standard data sequence;

[0043] The first standard data sequence and the second standard data sequence are input into an energy load prediction model constructed based on a gradient boosting decision tree to obtain predicted energy load data; wherein the predicted energy load data includes an energy load time series indicating the energy load expected to be required for the target area during the time period to be predicted.

[0044] As an improvement of the above solution, the method further includes:

[0045] Filter out historical power grid load data that matches the historical energy load data in the time dimension from a historical power grid load database of the target area;

[0046] The historical energy load data is sampled according to a preset time granularity to obtain a historical energy load time series; the historical energy load time series includes a plurality of energy load values ​​sorted by time;

[0047] Aligning the historical power grid load data with the historical energy load time series according to the time granularity to generate a historical power grid load time series; the historical power grid load time series includes a plurality of power grid load values ​​sorted by time;

[0048] Constructing a training data set according to the historical energy load time series and the historical power grid load time series; wherein the training data set includes a plurality of data pairs, each of which includes an energy load value and a power grid load value corresponding to the same time point;

[0049] The training data set is used to train a power grid load forecasting model constructed based on a long short-term memory network to obtain a trained power grid load forecasting model.

[0050] As an improvement of the above solution, the predicted power grid load data is obtained through a power grid load prediction model according to the predicted energy load data, including:

[0051] The predicted energy load data is sampled according to the time granularity to obtain a predicted energy load time series; the predicted energy load time series includes a plurality of predicted energy load values ​​sorted by time;

[0052] Inputting a plurality of the predicted energy load values ​​in the predicted energy load time series into a trained power grid load prediction model to obtain a plurality of predicted power grid load values;

[0053] Sorting the plurality of predicted power grid load values ​​in time to generate a predicted power grid load time series;

[0054] According to the historical energy load time series, the predicted power grid load time series is subjected to drift verification, and the predicted power grid load time series that passes the verification is used as the predicted power grid load data.

[0055] As an improvement of the above scheme, the predicted power grid load time series is subjected to drift verification according to the historical energy load time series, and the predicted power grid load time series that passes the verification is used as the predicted power grid load data, including:

[0056] Using a cubic spline function to perform curve fitting on the predicted energy load value in the predicted energy load time series and the predicted power grid load value in the predicted power grid load time series to obtain a first load change curve between the energy load and the power grid load;

[0057] Using a cubic spline function to perform curve fitting on the energy load values ​​in the historical energy load time series and the grid load values ​​in the historical grid load time series to obtain a second load change curve between the energy load and the grid load;

[0058] determining whether a deviation rate between the first load change curve and the second load change curve is less than a preset drift threshold;

[0059] If so, output the predicted power grid load time series as the predicted power grid load data;

[0060] if not, modifying the first load variation curve according to the second load variation curve;

[0061] A plurality of new predicted grid load values ​​are extracted from the modified first load change curve according to the time granularity, and the plurality of new predicted grid load values ​​are sorted according to time to generate a new predicted grid load time series as the predicted grid load data.

[0062] In a second aspect, an embodiment of the present invention provides a regional power distribution load dispatching system, including:

[0063] An image acquisition module, used to acquire remote sensing images of a target area;

[0064] An image recognition module is used to perform image recognition on the remote sensing image to obtain vegetation coverage information of the target area;

[0065] A vegetation irrigation amount prediction module, used to predict the vegetation irrigation amount sequence of the target area according to the vegetation coverage information and the weather data of the time period to be predicted;

[0066] A time granularity matching module is used to match the time granularity of the vegetation irrigation amount sequence with the time granularity of each historical energy load data in a pre-built historical energy load database to obtain historical energy load data with matching time granularity;

[0067] An energy load prediction module, used to obtain predicted energy load data through an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data;

[0068] A power grid load prediction module, used to obtain predicted power grid load data through a power grid load prediction model according to the predicted energy load data;

[0069] The power distribution load scheduling module is used to perform power distribution load scheduling on the target area according to the predicted power grid load data within the predicted time period.

[0070] Compared with the prior art, a regional power distribution load scheduling method and system according to an embodiment of the present invention obtains a remote sensing image of a target area; performs image recognition on the remote sensing image to obtain vegetation coverage information of the target area; predicts a vegetation irrigation quantity sequence of the target area based on the vegetation coverage information and weather data of a time period to be predicted; screens out historical energy load data that matches the vegetation irrigation quantity sequence in the time dimension from a historical energy load database of the target area; obtains predicted energy load data through an energy load prediction model based on the vegetation irrigation quantity sequence and the historical energy load data; obtains predicted power grid load data through a power grid load prediction model based on the predicted energy load data; and performs distribution load scheduling for the target area according to the predicted power grid load data within the time period to be predicted; the embodiment of the present invention fully considers the impact of regional vegetation on regional power load, and can effectively improve the accuracy of regional power grid load data prediction, thereby improving the regional power distribution scheduling effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the technical solution of the present invention, the drawings used in the implementation mode will be briefly introduced below. Obviously, the drawings described below are only some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0072] Figure 1 is a flow chart of a regional power distribution load scheduling method provided by an embodiment of the present invention;

[0073] Figure 2 is a schematic diagram of a process of image recognition of remote sensing images provided by an embodiment of the present invention;

[0074] Figure 3 It is a schematic diagram of the prediction process of the vegetation irrigation amount sequence provided by an embodiment of the present invention;

[0075] Figure 4 It is a schematic diagram of the process of energy load forecasting provided by an embodiment of the present invention;

[0076] Figure 5 It is a structural block diagram of a regional power distribution load dispatching system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] It is understood that the various numbers involved in the embodiments of the present invention are only for the convenience of description and are not intended to limit the scope of the present application. The size of the sequence number of each process does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic.

[0079] In embodiments of the present invention, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "include", "comprises", or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or device. In the absence of further restrictions, the elements defined by the statement "includes..." do not exclude the presence of other identical elements in the process, method, article, or device including the elements.

[0080] See also Figure 1 , Figure 1 1 is a flow chart of a regional power distribution load scheduling method provided by an embodiment of the present invention. The regional power distribution load scheduling method specifically includes:

[0081] S11: Acquire remote sensing images of the target area;

[0082] S12: performing image recognition on the remote sensing image to obtain vegetation coverage information of the target area;

[0083] S13: predicting a vegetation irrigation amount sequence for the target area according to the vegetation coverage information and weather data for the time period to be predicted;

[0084] S14: Filtering out historical energy load data that matches the vegetation irrigation amount sequence in the time dimension from the historical energy load database of the target area;

[0085] S15: obtaining predicted energy load data through an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data;

[0086] S16: Obtain predicted power grid load data through a power grid load prediction model according to the predicted energy load data;

[0087] S17: Performing distribution load scheduling for the target area according to the predicted power grid load data within the time period to be predicted.

[0088] It should be noted that the regional power distribution load scheduling method described in the embodiment of the present invention can be implemented by a server. In the embodiment of the present invention, the remote sensing image of the target area can be captured by terminal devices such as drones and low-orbit satellites, and the captured remote sensing image can be uploaded to the cloud for storage. At the same time, the cloud also stores the historical energy load data and historical power grid load data of the target area. Energy load includes but is not limited to the load of energy such as thermal energy, wind energy, electric energy, and hydropower. The power grid load refers to the load of electric energy, which is mainly supplied by the distribution network. Considering that the main factors for the maintenance and management of vegetation are irrigation and weather, and the irrigation of vegetation is mainly water resources and electric energy resources (power supply of smart watering equipment), in the example of the present invention, the remote sensing image of the target area is first obtained from the cloud for image recognition, and the vegetation coverage information of the target area is identified, and the weather data of the target area in the time period to be predicted is obtained from a third-party website (such as a meteorological monitoring platform / website), including but not limited to light intensity, wind speed, temperature, air humidity, etc. Then, the vegetation coverage information of the target area and the weather data in the time period to be predicted are combined to predict the vegetation irrigation amount sequence in the time period to be predicted; the vegetation irrigation amount sequence records multiple vegetation irrigation amounts and the timestamp of each vegetation irrigation amount, wherein the time granularity set between two adjacent vegetation irrigation amounts. It should be noted that the embodiment of the present invention does not specifically limit the value of the time granularity, for example, it can be 4 hours, 8 hours, 1 day, 1 week, 1 month, etc. Then, the historical energy load data of the target area that matches the vegetation irrigation quantity sequence in the time dimension is screened out from the historical energy load database in the cloud, and based on the vegetation irrigation quantity sequence and the historical energy load data that match in the time dimension, the energy load prediction model is used to perform energy load prediction, and the predicted energy load data of the target area in the time period to be predicted is obtained; finally, based on the predicted energy load data, the power grid load prediction model is used to perform power grid load prediction, and the predicted power grid load data of the target area in the time period to be predicted is obtained; within the time period to be predicted, the target area is distributed with the predicted power grid load data predicted above, so that the distribution load scheduling of the target area fully considers the impact of regional vegetation on regional power load, which can effectively improve the accuracy of regional power grid load data prediction, thereby improving the regional distribution scheduling effect.

[0089] like Figure 2 As shown, step S12: performing image recognition on the remote sensing image to obtain vegetation coverage information of the target area includes:

[0090] S121: performing denoising processing on the remote sensing image;

[0091] For example, after acquiring a remote sensing image, the remote sensing image is first subjected to denoising processing to remove noise pollution in the remote sensing image. The embodiment of the present invention does not specifically limit the denoising method of the remote sensing image. For example, the remote sensing image can be first subjected to Fourier transform to obtain a spectrum diagram of the remote sensing image, and then the spectrum diagram is subjected to Gaussian filtering, and then the spectrum diagram after Gaussian filtering is subjected to inverse Fourier transform to obtain a denoised remote sensing image. For another example, a sliding window with an adaptive median is used to perform denoising processing on the remote sensing image. The denoising processing is mainly performed by assigning the average value of a pixel and all pixels in its neighborhood to the corresponding pixel in the remote sensing image, so that a sliding window slides on the remote sensing image. The adaptive median of the sliding window (i.e., the center position of the window) is the average value of the grayscale values ​​of each pixel in the sliding window. Then, the grayscale values ​​of each pixel in the neighborhood are sorted, and the middle value is taken as the new value of the grayscale of the central pixel. The neighborhood here is the sliding window. When the sliding window moves up, down, left, and right in the remote sensing image, the median filtering algorithm can be used to smooth the image well and eliminate isolated noise points.

[0092] S122: performing single band extraction on the denoised remote sensing image to obtain an infrared band image and a near infrared band image;

[0093] Since the near-infrared band has the characteristic of high reflection and the red light band has the characteristic of strong absorption, the vegetation area on the remote sensing image usually shows high contrast in the red band and the near-red band. Based on this, the embodiment of the present invention performs multi-spectral analysis on the denoised remote sensing image, and extracts the infrared band information and the near-infrared band information to obtain the infrared band image and the near-infrared band image.

[0094] S123: generating a vegetation coverage map of the target area by normalizing vegetation index according to the infrared band image and the near-infrared band image;

[0095] In an optional embodiment, the normalized vegetation index may be calculated based on the pixel values ​​of the infrared band image and the near infrared band image, for example, the normalized vegetation index (NDVI) of the infrared band image and the near infrared band image may be calculated by the following formula (1);

[0096] (1);

[0097] Among them, NIR i Represents the pixel value of the i-th pixel in the near-infrared band image, RED i Represents the pixel value of the i-th pixel in the infrared band image, NDVI i1 The normalized difference vegetation index of the i-th pixel is used to measure vegetation density. i1 >1, indicating that the ground is covered with vegetation and NDVI i1The larger the value, the denser the vegetation density, and vice versa. i1 ≤1, indicating that the ground is not covered by vegetation. At this time, the ground may be covered by clouds, water, and other visible highly reflective materials.

[0098] In another optional embodiment, the normalized vegetation index may also be calculated in the following manner, and the specific process is as follows:

[0099] Step A: performing color correction on the infrared band image and the near infrared band image;

[0100] Step B: calculating the first reflectivity of the infrared band image according to the pixel value of the corrected infrared band image and the pixel value of the standard grayscale image;

[0101] Step C: calculating the second reflectivity of the near-infrared band image according to the pixel value of the corrected near-infrared band image and the pixel value of the standard grayscale image;

[0102] Step D: Calculating the normalized vegetation index of each pixel pair according to the first reflectivity of the infrared band image and the second reflectivity of the near-infrared band image; wherein each pixel pair includes pixels located at the same position in the infrared band image and the near-infrared band image;

[0103] Step E: Generate a vegetation coverage map of the target area according to the vegetation indexes of all the pixel pairs.

[0104] For example, a standard grayscale image can be used to perform color correction on the infrared band image and the near-infrared band image, wherein the standard grayscale image can be an image obtained by photographing a 32-degree standard gray card (i.e., a color card made by dividing the grayscale value of 0-256 into 32 intervals and taking the grayscale average value of each interval as the standard grayscale of the interval) under preset standard conditions, such as preset reference temperature, reference light intensity, etc. Specifically, the infrared band image and the near-infrared band image can be color corrected using the following formula;

[0105] For near-infrared band images:

[0106] (2);

[0107] For infrared images:

[0108] (3);

[0109] Among them, DN NIRi2 DN REDi2 Respectively represent the grayscale value of the i-th pixel in the corrected near-infrared band image and infrared band image, DNNIRi1 DN REDi1 Respectively represent the gray value of the i-th pixel in the near-infrared band image and the infrared band image before correction, DN g18N DN g18R They represent the grayscale average values ​​of all pixels in the near-infrared band and infrared band of the standard grayscale image respectively.

[0110] For the first reflectivity r of the i-th pixel in the infrared band image REDi :

[0111] (4);

[0112] For the second reflectivity r of the i-th pixel in the near-infrared band image NIRi :

[0113] (5);

[0114] Among them, DN g18 Represents the grayscale average value of all pixels in the standard grayscale image; then the normalized vegetation index of the ith pixel (i.e., the ith pixel pair) is:

[0115] (6).

[0116] Afterwards, the normalized vegetation index of each pixel is used to update the respective pixel value, so as to generate the vegetation coverage map.

[0117] S124: performing texture recognition on the denoised remote sensing image using a pre-built image texture recognition model to obtain a vegetation texture map of at least one vegetation type;

[0118] Exemplarily, the denoised remote sensing image is input into an image texture recognition model based on a convolutional neural network for texture information recognition, and multiple vegetation types in the remote sensing image and texture information of the vegetation type, namely, a vegetation texture map, can be obtained.

[0119] It should be understood that the image texture recognition model can be trained by vegetation texture samples in a pre-constructed vegetation texture database, wherein the vegetation texture database stores image samples of different vegetation types and texture labels of corresponding vegetation types. The image samples are used as the input of the image texture recognition model, and the texture labels are used as the output of the image texture recognition model. The image texture recognition model is trained until a preset model accuracy is reached or a preset number of training times is reached, and a trained image texture recognition model can be obtained.

[0120] S125: Calculating the vegetation coverage degree of the corresponding vegetation type according to the vegetation coverage map and each of the vegetation texture maps;

[0121] S126: Obtain the vegetation coverage information according to the vegetation types and the vegetation coverage degrees of the vegetation types.

[0122] Specifically, for each of the vegetation texture maps, the vegetation texture map is projected onto the vegetation coverage map; then the pixel ratio of the vegetation texture in the vegetation texture map to the vegetation coverage map is calculated as the vegetation coverage degree of the corresponding vegetation type. For example, for each vegetation type, the vegetation texture map is projected onto the corresponding pixel position of the vegetation coverage map according to the one-to-one correspondence between pixel positions, and then the pixel ratio of the texture pixels projected onto the vegetation coverage map to the vegetation coverage map is calculated to obtain the vegetation coverage degree of the corresponding vegetation type. Finally, the vegetation types and coverage degrees of each vegetation type obtained by the above calculations are used as the vegetation coverage information of the target area.

[0123] like Figure 3 As shown, step S13: predicting the vegetation irrigation amount sequence of the target area according to the vegetation coverage information and the weather data of the time period to be predicted, including:

[0124] S131: querying the basic transpiration rate of each vegetation type in the vegetation coverage information;

[0125] For example, the basic transpiration rate of various vegetation types under standard conditions can be queried from the cloud, that is, the amount of water lost by transpiration per unit area of ​​vegetation per unit time.

[0126] S132: Calculating the actual coverage area of ​​each vegetation type according to the vegetation coverage degree of each vegetation type and the area of ​​the target area;

[0127] In the embodiment of the present invention, according to the vegetation coverage degree (i.e., pixel ratio) of each vegetation type in the remote sensing image and the area of ​​the target area, the actual coverage area of ​​the corresponding vegetation type in the target area can be calculated, for example, by s j =(z j / I)×S, where z j / I represents the pixel proportion of the jth vegetation type, z j represents the number of pixels of the jth vegetation type in the vegetation coverage map, I represents the total number of pixels in the vegetation coverage map, S represents the area of ​​the target area, and s j Represents the actual coverage area of ​​the jth vegetation type in the target area.

[0128] S133: Calculating multiple temperature influencing factors according to the temperature data in the weather data according to a preset time granularity;

[0129] Among them, the function expression of temperature influencing factor is:

[0130] (7);

[0131] represents the temperature impact factor at time x, T0 represents the base temperature impact factor, which is a fixed value, k represents the attenuation coefficient, g represents the time granularity, i.e., the time interval, T x Represents the temperature at time x, obtained from weather data.

[0132] S134: Calculating a plurality of light intensity influencing factors according to the light intensity data in the weather data and at the time granularity;

[0133] Among them, the function expression of the light intensity influencing factor is:

[0134] (8);

[0135] It represents the light intensity influence factor at time x, I0 represents the reference light intensity influence factor, which is a fixed value. represents the light intensity growth rate, I x Indicates the wind speed at time x, obtained from weather data.

[0136] S135: Calculating a plurality of wind speed influencing factors according to the time granularity based on the wind speed data in the weather data;

[0137] Among them, the function expression of wind speed influence factor is:

[0138] (9);

[0139] represents the wind speed influence factor at time x, V0 represents the reference wind speed influence factor, which is a fixed value. Indicates the wind speed adjustment factor, V x Indicates the wind speed at time x, obtained from weather data.

[0140] S136: Calculating a plurality of humidity influencing factors according to the time granularity based on the air humidity data in the weather data;

[0141] Among them, the function expression of humidity influence factor is:

[0142] (10);

[0143] represents the humidity impact factor at time x, H0 represents the base humidity impact factor, which is a fixed value. Represents the humidity adjustment factor, H xRepresents the air humidity at time x, obtained from weather data.

[0144] S137: For each of the vegetation types, according to the basic transpiration rate of the vegetation type, the actual coverage area, the multiple temperature influencing factors, the multiple light intensity influencing factors, the multiple wind speed influencing factors, and the multiple humidity influencing factors, a transpiration rate sequence of the corresponding vegetation type is calculated;

[0145] Among them, the functional expression of the transpiration rate of the jth vegetation type at the xth time is:

[0146] (11);

[0147] E jx represents the transpiration rate of the jth vegetation type at time x, E j0 represents the basic transpiration rate of the jth vegetation type, s j represents the actual coverage area of ​​the jth vegetation type, a, b, c, and d represent the weights of the temperature influencing factor, light intensity influencing factor, wind speed influencing factor, and humidity influencing factor, respectively, and a+b+c+d=1.

[0148] Based on the above formula, the transpiration rates at multiple times can be obtained for each vegetation type. By sorting them in chronological order, the transpiration rate sequence of the vegetation type can be generated.

[0149] S138: generating a rainfall sequence according to the rainfall data in the weather data and the time granularity;

[0150] The rainfall in the target area during the forecast period is sampled according to the preset time granularity, and the sampled rainfall is sorted in chronological order to generate a rainfall sequence, wherein the rainfall sequence is time-aligned with the above transpiration rate sequence.

[0151] S139: Calculate the vegetation irrigation amount sequence of the target area based on the transpiration rate sequences of all the vegetation types and the rainfall sequence.

[0152] In an embodiment of the present invention, the amount of vegetation irrigation mainly considers the influence of the transpiration rate and rainfall of the vegetation. For example, the difference between the transpiration amount and the precipitation of the vegetation is considered as the irrigation water demand (i.e., the amount of vegetation irrigation). At this time, the amount of vegetation irrigation = (transpiration rate of vegetation × vegetation area) - (rainfall × vegetation area). For the rainfall sequence and the transpiration rate sequence of each vegetation type, the above formula is used to calculate the vegetation irrigation amount of the vegetation type at each time, and the vegetation irrigation amounts of all vegetation types at the same time are summed to obtain the total vegetation irrigation amount of the target area at that time; finally, the multiple total vegetation irrigation amounts calculated are sorted in chronological order to generate the vegetation irrigation amount sequence of the target area.

[0153] like Figure 4 As shown, step S15: obtaining predicted energy load data through an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data, including:

[0154] S151: Calculate a first Z score value of each vegetation irrigation amount in the vegetation irrigation amount sequence, and perform a data outlier check on each of the first Z score values;

[0155] S152: Calculate a second Z score value of each historical energy load in the historical energy load data, and perform a data outlier check on each of the second Z score values;

[0156] S153: normalizing the first Z score values ​​that pass the verification, and sorting the normalized first Z score values ​​in chronological order to generate a first standard data sequence;

[0157] S154: normalizing the verified second Z score values, and sorting the normalized second Z score values ​​in chronological order to generate a second standard data sequence;

[0158] S155: Input the first standard data sequence and the second standard data sequence into an energy load prediction model constructed based on a gradient boosting decision tree to obtain predicted energy load data; wherein the predicted energy load data includes an energy load time series indicating the energy load expected to be required for the target area during the time period to be predicted.

[0159] In an embodiment of the present invention, for historical energy load data, the same time granularity as the vegetation irrigation amount sequence can be sampled, and the sampled historical energy loads can be sorted in chronological order to obtain a historical energy load sequence; then, for each vegetation irrigation amount in the vegetation irrigation amount sequence and each historical energy load in the historical energy load sequence, Z scores are calculated respectively; and for the first Z score value corresponding to the calculated vegetation irrigation amount and the second Z score value corresponding to the historical energy load, if the absolute value of the first Z score value / the second Z score value exceeds a preset outlier threshold (such as 3), it means that the first Z score value / the second Z score value is an abnormal value, and the value is eliminated; if the first Z score value / the second Z score value does not exceed the preset outlier threshold, it means that the first Z score value / the second Z score value is a normal value, and the value is retained. Based on the retained first Z score value, the first standard data sequence can be generated by sorting in chronological order, and based on the retained second Z score value, the second standard data sequence can be generated by sorting in chronological order.

[0160] Afterwards, the first Z score value and the second Z score value in the generated first standard data sequence and the second standard data sequence at the same time are taken as data pairs and input into the energy load prediction model constructed based on the gradient boosting decision tree in sequence for energy load prediction, so as to obtain the energy load time series of the energy load expected to be required in the target area during the forecast time period.

[0161] Furthermore, the method further comprises:

[0162] Filter out historical power grid load data that matches the historical energy load data in the time dimension from a historical power grid load database of the target area;

[0163] The historical energy load data is sampled according to a preset time granularity to obtain a historical energy load time series; the historical energy load time series includes a plurality of energy load values ​​sorted by time;

[0164] Aligning the historical power grid load data with the historical energy load time series according to the time granularity to generate a historical power grid load time series; the historical power grid load time series includes a plurality of power grid load values ​​sorted by time;

[0165] Constructing a training data set according to the historical energy load time series and the historical power grid load time series; wherein the training data set includes a plurality of data pairs, each of which includes an energy load value and a power grid load value corresponding to the same time point;

[0166] The training data set is used to train a power grid load forecasting model constructed based on a long short-term memory network to obtain a trained power grid load forecasting model.

[0167] Specifically, step S16: obtaining predicted power grid load data through a power grid load prediction model according to the predicted energy load data, including:

[0168] The predicted energy load data is sampled according to the time granularity to obtain a predicted energy load time series; the predicted energy load time series includes a plurality of predicted energy load values ​​sorted by time;

[0169] Inputting a plurality of the predicted energy load values ​​in the predicted energy load time series into a trained power grid load prediction model to obtain a plurality of predicted power grid load values;

[0170] Sorting the plurality of predicted power grid load values ​​in time to generate a predicted power grid load time series;

[0171] According to the historical energy load time series, the predicted power grid load time series is subjected to drift verification, and the predicted power grid load time series that passes the verification is used as the predicted power grid load data.

[0172] The step of performing drift verification on the predicted power grid load time series according to the historical energy load time series, and using the predicted power grid load time series that passes the verification as the predicted power grid load data, includes:

[0173] Using a cubic spline function to perform curve fitting on the predicted energy load value in the predicted energy load time series and the predicted power grid load value in the predicted power grid load time series to obtain a first load change curve between the energy load and the power grid load;

[0174] Using a cubic spline function to perform curve fitting on the energy load values ​​in the historical energy load time series and the grid load values ​​in the historical grid load time series to obtain a second load change curve between the energy load and the grid load;

[0175] determining whether a deviation rate between the first load change curve and the second load change curve is less than a preset drift threshold;

[0176] If so, outputting the predicted power grid load time series as the predicted power grid load data;

[0177] if not, modifying the first load variation curve according to the second load variation curve;

[0178] A plurality of new predicted grid load values ​​are extracted from the modified first load change curve according to the time granularity, and the plurality of new predicted grid load values ​​are sorted according to time to generate a new predicted grid load time series as the predicted grid load data.

[0179] The embodiment of the present invention obtains a first load change curve between energy load and grid load, i.e., a predicted change law, by curve fitting the predicted energy load value in the predicted energy load time series and the predicted grid load value in the predicted grid load time series; and obtains a second load change curve between energy load and grid load, i.e., a historical change law, by curve fitting the energy load value in the historical energy load time series and the grid load value in the historical grid load time series; then the first load change curve is corrected by using the second load change curve, for example, if the absolute value of the difference between the predicted grid load value at a certain time in the first load change curve and the historical grid load value at the same time in the second load change curve exceeds a preset drift threshold, it means that the predicted grid load value at the time is an abnormal value, and the historical grid load value at the same time in the second load change curve is used to update the predicted grid load value of the first load change curve, otherwise the predicted grid load value is retained; then the cubic spline function is re-adopted to perform curve fitting according to the updated predicted grid load value and the predicted grid load value finally retained to obtain the corrected first load change curve.

[0180] It should be noted that the embodiment of the present invention does not impose any specific limitation on the value of the drift threshold, and the value may be limited according to actual needs.

[0181] Finally, multiple predicted power grid load values ​​are extracted from the corrected first load change curve according to the preset time granularity, and sorted by time to generate the final predicted power grid load time series. Subsequently, the distribution load of the target area is dispatched at the corresponding time according to the predicted power grid load value at each time in the predicted power grid load time series, which can realize accurate dispatch of regional distribution load, make regional distribution load dispatch more in line with the actual situation of the area, and improve the regional distribution dispatch effect.

[0182] See also Figure 5 , Figure 5 The embodiment of the present invention provides a structural block diagram of a regional power distribution load dispatching system, wherein the regional power distribution load dispatching system comprises:

[0183] An image acquisition module 11 is used to acquire a remote sensing image of a target area;

[0184] An image recognition module 12 is used to perform image recognition on the remote sensing image to obtain vegetation coverage information of the target area;

[0185] A vegetation irrigation amount prediction module 13 is used to predict the vegetation irrigation amount sequence of the target area according to the vegetation coverage information and the weather data of the time period to be predicted;

[0186] A time granularity matching module 14 is used to match the time granularity of the vegetation irrigation amount sequence with the time granularity of each historical energy load data in a pre-built historical energy load database to obtain historical energy load data with matching time granularity;

[0187] An energy load prediction module 15 is used to obtain predicted energy load data through an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data;

[0188] A power grid load prediction module 16, configured to obtain predicted power grid load data through a power grid load prediction model according to the predicted energy load data;

[0189] The power distribution load scheduling module 17 is used to perform power distribution load scheduling for the target area according to the predicted power grid load data within the time period to be predicted.

[0190] In an optional embodiment, the image recognition module 12 includes:

[0191] A denoising unit, used for performing denoising processing on the remote sensing image;

[0192] A single-band extraction unit is used to perform single-band extraction on the denoised remote sensing image to obtain an infrared band image and a near-infrared band image;

[0193] A normalized vegetation index calculation unit, used to generate a vegetation coverage map of the target area by normalizing the vegetation index according to the infrared band image and the near-infrared band image;

[0194] A texture recognition unit, used to perform texture recognition on the denoised remote sensing image through a pre-built image texture recognition model to obtain a vegetation texture map of at least one vegetation type;

[0195] A vegetation coverage degree calculation unit, used to calculate the vegetation coverage degree of the corresponding vegetation type according to the vegetation coverage map and each of the vegetation texture maps;

[0196] The vegetation coverage information obtaining unit is used to obtain the vegetation coverage information according to the vegetation type and the vegetation coverage degree of each vegetation type.

[0197] In an optional embodiment, the normalized vegetation index calculation unit includes:

[0198] A color correction subunit, used for performing color correction on the infrared band image and the near infrared band image;

[0199] A first reflectivity calculation subunit, used for calculating a first reflectivity of the infrared band image according to the pixel values ​​of the corrected infrared band image and the pixel values ​​of the standard grayscale image;

[0200] A second reflectivity calculation subunit, used for calculating a second reflectivity of the near-infrared band image according to the pixel value of the corrected near-infrared band image and the pixel value of the standard grayscale image;

[0201] A vegetation index calculation subunit, used to calculate the normalized vegetation index of each pixel pair according to the first reflectivity of the infrared band image and the second reflectivity of the near-infrared band image; wherein each pixel pair includes pixels located at the same position in the infrared band image and the near-infrared band image;

[0202] The vegetation coverage degree calculation subunit is used to generate a vegetation coverage map of the target area according to the vegetation indexes of all the pixel pairs.

[0203] In an optional embodiment, the vegetation coverage degree calculation unit includes:

[0204] A projection subunit, configured to project each of the vegetation texture maps onto the vegetation coverage map;

[0205] The pixel ratio calculation subunit is used to calculate the pixel ratio of the vegetation texture in the vegetation texture map in the vegetation coverage map as the vegetation coverage degree of the corresponding vegetation type.

[0206] In an optional embodiment, the vegetation irrigation amount prediction module 13 includes:

[0207] A basic transpiration rate query unit, used to query the basic transpiration rate of each vegetation type in the vegetation coverage information;

[0208] A coverage area calculation unit, used for calculating the actual coverage area of ​​each vegetation type according to the vegetation coverage degree of each vegetation type and the area of ​​the target area;

[0209] A temperature influence factor calculation unit, used to calculate a plurality of temperature influence factors according to a preset time granularity based on the temperature data in the weather data;

[0210] A light intensity impact factor calculation unit, configured to calculate a plurality of light intensity impact factors according to the light intensity data in the weather data and at the time granularity;

[0211] A wind speed influence factor calculation unit, used to calculate a plurality of wind speed influence factors according to the wind speed data in the weather data and at the time granularity;

[0212] A humidity impact factor calculation unit, used to calculate a plurality of humidity impact factors according to the air humidity data in the weather data and the time granularity;

[0213] A transpiration rate sequence generating unit is used to calculate the transpiration rate sequence of the corresponding vegetation type for each vegetation type according to the basic transpiration rate of the vegetation type, the actual coverage area, the multiple temperature influencing factors, the multiple light intensity influencing factors, the multiple wind speed influencing factors, and the multiple humidity influencing factors;

[0214] A rainfall sequence generating unit, configured to generate a rainfall sequence according to the rainfall data in the weather data and the time granularity;

[0215] The vegetation irrigation amount sequence calculation unit is used to calculate the vegetation irrigation amount sequence of the target area according to the transpiration rate sequence of all the vegetation types and the rainfall sequence.

[0216] In an optional embodiment, the energy load prediction module 15 includes:

[0217] a first Z score calculation unit, configured to calculate a first Z score value of each vegetation irrigation amount in the vegetation irrigation amount sequence, and perform a data outlier check on each of the first Z score values;

[0218] A second Z score calculation unit, used to calculate a second Z score value of each historical energy load in the historical energy load data, and perform a data outlier check on each of the second Z score values;

[0219] A second standard data sequence generating unit is used to normalize the first Z score values ​​that pass the verification, and sort the normalized first Z score values ​​in chronological order to generate a first standard data sequence;

[0220] A second standard data sequence generating unit, configured to perform normalization processing on the verified second Z score values, and sort the normalized second Z score values ​​in chronological order to generate a second standard data sequence;

[0221] An energy load data prediction unit is used to input the first standard data sequence and the second standard data sequence into an energy load prediction model constructed based on a gradient boosting decision tree to obtain predicted energy load data; wherein the predicted energy load data includes an energy load time series indicating the energy load expected to be required for the target area during the time period to be predicted.

[0222] In an optional embodiment, the system further includes:

[0223] A historical power grid load data acquisition module, used to filter out historical power grid load data that matches the historical energy load data in the time dimension from the historical power grid load database of the target area;

[0224] A historical energy load sampling module is used to sample the historical energy load data according to a preset time granularity to obtain a historical energy load time series; the historical energy load time series includes multiple energy load values ​​sorted by time;

[0225] A time alignment module, used to align the historical power grid load data with the historical energy load time series according to the time granularity to generate a historical power grid load time series; the historical power grid load time series includes a plurality of power grid load values ​​sorted by time;

[0226] A training data set construction module, used to construct a training data set according to the historical energy load time series and the historical power grid load time series; wherein the training data set includes a plurality of data pairs, each of which includes an energy load value and a power grid load value corresponding to the same time point;

[0227] The model training module is used to use the training data set to train the power grid load prediction model built based on the long short-term memory network to obtain a trained power grid load prediction model.

[0228] In an optional embodiment, the power grid load prediction module 16 includes:

[0229] An energy load sampling unit, used for sampling the predicted energy load data according to the time granularity to obtain a predicted energy load time series; the predicted energy load time series includes a plurality of predicted energy load values ​​sorted by time;

[0230] An energy load value prediction unit, used for inputting a plurality of the predicted energy load values ​​in the predicted energy load time series into a trained power grid load prediction model to obtain a plurality of predicted power grid load values;

[0231] A predicted power grid load time series generating unit, used for sorting the plurality of predicted power grid load values ​​in time order to generate a predicted power grid load time series;

[0232] The drift verification unit is used to perform drift verification on the predicted power grid load time series according to the historical energy load time series, and use the predicted power grid load time series that passes the verification as the predicted power grid load data.

[0233] In an optional embodiment, the drift verification unit includes:

[0234] A first curve fitting subunit is used to use a cubic spline function to perform curve fitting on the predicted energy load value in the predicted energy load time series and the predicted grid load value in the predicted grid load time series to obtain a first load change curve between the energy load and the grid load;

[0235] A second curve fitting subunit is used to use a cubic spline function to perform curve fitting on the energy load value in the historical energy load time series and the grid load value in the historical grid load time series to obtain a second load change curve between the energy load and the grid load;

[0236] a judgment subunit, configured to judge whether the offset rate between the first load change curve and the second load change curve is less than a preset drift threshold; if so, output the predicted power grid load time series as the predicted power grid load data; if not, correct the first load change curve according to the second load change curve;

[0237] The curve correction subunit is used to extract multiple new predicted power grid load values ​​from the corrected first load change curve according to the time granularity, and sort the multiple new predicted power grid load values ​​according to time to generate a new predicted power grid load time series as the predicted power grid load data.

[0238] It should be noted that the working process of each module in the regional power distribution load dispatching system described in the embodiment of the present invention can refer to the working process of the regional power distribution load dispatching method described in the above embodiment, and the technical effect achieved is the same as the regional power distribution load dispatching method described in the above embodiment, which will not be repeated here.

[0239] It should be noted that the system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiment provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0240] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, many improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for regional power distribution load scheduling, characterized in that: include: Acquire remote sensing images of the target area; Performing image recognition on the remote sensing image to obtain vegetation coverage information of the target area; Predicting a vegetation irrigation amount sequence for the target area based on the vegetation coverage information and weather data for the time period to be predicted; Filtering out historical energy load data that matches the vegetation irrigation amount sequence in the time dimension from a historical energy load database of the target area; According to the vegetation irrigation amount sequence and the historical energy load data, using an energy load prediction model, obtaining predicted energy load data; According to the predicted energy load data, using a power grid load prediction model, predicted power grid load data is obtained; Performing distribution load scheduling for the target area according to the predicted power grid load data within the time period to be predicted; The method further comprises: Filter out historical power grid load data that matches the historical energy load data in the time dimension from a historical power grid load database of the target area; The historical energy load data is sampled according to a preset time granularity to obtain a historical energy load time series; the historical energy load time series includes a plurality of energy load values ​​sorted by time; Aligning the historical power grid load data with the historical energy load time series according to the time granularity to generate a historical power grid load time series; the historical power grid load time series includes a plurality of power grid load values ​​sorted by time; Constructing a training data set according to the historical energy load time series and the historical power grid load time series; wherein the training data set includes a plurality of data pairs, each of which includes an energy load value and a power grid load value corresponding to the same time point; Using the training data set, training a power grid load forecasting model built based on a long short-term memory network to obtain a trained power grid load forecasting model; The step of obtaining predicted energy load data by using an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data includes: Calculating a first Z score value of each vegetation irrigation amount in the vegetation irrigation amount sequence, and performing a data outlier check on each of the first Z score values; Calculating a second Z score value of each historical energy load in the historical energy load data, and performing a data outlier check on each of the second Z score values; Normalizing the first Z score values ​​that pass the verification, and sorting the normalized first Z score values ​​in chronological order to generate a first standard data sequence; Normalizing the verified second Z score values, and sorting the normalized second Z score values ​​in chronological order to generate a second standard data sequence; Inputting the first standard data sequence and the second standard data sequence into an energy load forecasting model constructed based on a gradient boosting decision tree to obtain forecasted energy load data; wherein the forecasted energy load data includes an energy load time series indicating the energy load expected to be required for the target area within the forecasting time period; The step of obtaining predicted power grid load data by using a power grid load prediction model according to the predicted energy load data includes: The predicted energy load data is sampled according to the time granularity to obtain a predicted energy load time series; the predicted energy load time series includes a plurality of predicted energy load values ​​sorted by time; Inputting a plurality of the predicted energy load values ​​in the predicted energy load time series into a trained power grid load prediction model to obtain a plurality of predicted power grid load values; Sorting the plurality of predicted power grid load values ​​in time to generate a predicted power grid load time series; According to the historical energy load time series, the predicted power grid load time series is subjected to drift verification, and the predicted power grid load time series that passes the verification is used as the predicted power grid load data.

2. The regional power distribution load dispatching method according to claim 1, characterized in that: The performing image recognition on the remote sensing image to obtain vegetation coverage information of the target area includes: Performing denoising processing on the remote sensing image; Perform single-band extraction on the denoised remote sensing image to obtain infrared band images and near-infrared band images; Generating a vegetation coverage map of the target area by normalizing vegetation index according to the infrared band image and the near-infrared band image; Performing texture recognition on the denoised remote sensing image using a pre-built image texture recognition model to obtain a vegetation texture map of at least one vegetation type; Calculating the vegetation coverage degree of the corresponding vegetation type according to the vegetation coverage map and each of the vegetation texture maps; The vegetation coverage information is obtained according to the vegetation types and the vegetation coverage degrees of each of the vegetation types.

3. The regional power distribution load dispatching method according to claim 2, characterized in that: The step of generating a vegetation coverage map of the target area by normalizing vegetation index according to the infrared band image and the near infrared band image includes: Performing color correction on the infrared band image and the near infrared band image; Calculating a first reflectivity of the infrared band image according to the pixel value of the corrected infrared band image and the pixel value of the standard grayscale image; Calculating a second reflectivity of the near-infrared band image according to the pixel value of the corrected near-infrared band image and the pixel value of the standard grayscale image; Calculating the normalized vegetation index of each pixel pair according to the first reflectivity of the infrared band image and the second reflectivity of the near-infrared band image; wherein each pixel pair includes pixels located at the same position in the infrared band image and the near-infrared band image; A vegetation coverage map of the target area is generated according to the vegetation indexes of all the pixel pairs.

4. The regional power distribution load dispatching method according to claim 2, characterized in that: The calculating the vegetation coverage degree of the corresponding vegetation type according to the vegetation coverage map and each of the vegetation texture maps includes: For each of the vegetation texture maps, projecting the vegetation texture map onto the vegetation coverage map; The pixel ratio of the vegetation texture in the vegetation texture map in the vegetation coverage map is calculated as the vegetation coverage degree of the corresponding vegetation type.

5. The regional power distribution load dispatching method according to claim 2, characterized in that: The step of predicting the vegetation irrigation amount sequence of the target area according to the vegetation coverage information and the weather data of the time period to be predicted includes: Querying the basic transpiration rate of each vegetation type in the vegetation coverage information; Calculating the actual coverage area of ​​each vegetation type according to the vegetation coverage degree of each vegetation type and the area of ​​the target area; Calculate multiple temperature influencing factors according to the temperature data in the weather data at a preset time granularity; Calculate multiple light intensity influencing factors according to the light intensity data in the weather data and at the time granularity; Calculating multiple wind speed influencing factors according to the time granularity based on the wind speed data in the weather data; Calculating a plurality of humidity influencing factors according to the time granularity based on the air humidity data in the weather data; For each of the vegetation types, according to the basic transpiration rate of the vegetation type, the actual coverage area, the multiple temperature influencing factors, the multiple light intensity influencing factors, the multiple wind speed influencing factors, and the multiple humidity influencing factors, the transpiration rate sequence of the corresponding vegetation type is calculated; Generate a rainfall sequence according to the time granularity based on the rainfall data in the weather data; The vegetation irrigation amount sequence of the target area is calculated based on the transpiration rate sequence of all the vegetation types and the rainfall sequence.

6. The regional power distribution load dispatching method according to claim 1, characterized in that: The method of performing drift verification on the predicted power grid load time series according to the historical energy load time series, and using the predicted power grid load time series that passes the verification as the predicted power grid load data, includes: Using a cubic spline function to perform curve fitting on the predicted energy load value in the predicted energy load time series and the predicted power grid load value in the predicted power grid load time series to obtain a first load change curve between the energy load and the power grid load; Using a cubic spline function to perform curve fitting on the energy load values ​​in the historical energy load time series and the grid load values ​​in the historical grid load time series to obtain a second load change curve between the energy load and the grid load; determining whether a deviation rate between the first load change curve and the second load change curve is less than a preset drift threshold; If so, output the predicted power grid load time series as the predicted power grid load data; if not, modifying the first load variation curve according to the second load variation curve; A plurality of new predicted grid load values ​​are extracted from the modified first load change curve according to the time granularity, and the plurality of new predicted grid load values ​​are sorted according to time to generate a new predicted grid load time series as the predicted grid load data.

7. A regional power distribution load dispatching system, characterized in that: include: An image acquisition module, used to acquire remote sensing images of a target area; An image recognition module is used to perform image recognition on the remote sensing image to obtain vegetation coverage information of the target area; A vegetation irrigation amount prediction module, used to predict the vegetation irrigation amount sequence of the target area according to the vegetation coverage information and the weather data of the time period to be predicted; A time granularity matching module is used to match the time granularity of the vegetation irrigation amount sequence with the time granularity of each historical energy load data in a pre-built historical energy load database to obtain historical energy load data with matching time granularity; An energy load prediction module, used to obtain predicted energy load data through an energy load prediction model according to the vegetation irrigation amount sequence and the historical energy load data; A power grid load prediction module, used to obtain predicted power grid load data through a power grid load prediction model according to the predicted energy load data; A distribution load scheduling module, used for performing distribution load scheduling for the target area according to the predicted power grid load data within the predicted time period; A historical power grid load data acquisition module, used to filter out historical power grid load data that matches the historical energy load data in the time dimension from the historical power grid load database of the target area; A historical energy load sampling module is used to sample the historical energy load data according to a preset time granularity to obtain a historical energy load time series; the historical energy load time series includes multiple energy load values ​​sorted by time; A time alignment module, used to align the historical power grid load data with the historical energy load time series according to the time granularity to generate a historical power grid load time series; the historical power grid load time series includes a plurality of power grid load values ​​sorted by time; A training data set construction module, used to construct a training data set according to the historical energy load time series and the historical power grid load time series; wherein the training data set includes a plurality of data pairs, each of which includes an energy load value and a power grid load value corresponding to the same time point; A model training module is used to train a power grid load forecasting model constructed based on a long short-term memory network using the training data set to obtain a trained power grid load forecasting model; The energy load forecasting module includes: a first Z score calculation unit, configured to calculate a first Z score value of each vegetation irrigation amount in the vegetation irrigation amount sequence, and perform a data outlier check on each of the first Z score values; A second Z score calculation unit, used to calculate a second Z score value of each historical energy load in the historical energy load data, and perform a data outlier check on each of the second Z score values; A second standard data sequence generating unit is used to normalize the first Z score values ​​that pass the verification, and sort the normalized first Z score values ​​in chronological order to generate a first standard data sequence; A second standard data sequence generating unit, configured to perform normalization processing on the verified second Z score values, and sort the normalized second Z score values ​​in chronological order to generate a second standard data sequence; An energy load data prediction unit, configured to input the first standard data sequence and the second standard data sequence into an energy load prediction model constructed based on a gradient boosting decision tree to obtain predicted energy load data; wherein the predicted energy load data includes an energy load time series indicating the energy load expected to be used in the target area within the time period to be predicted; The power grid load forecasting module includes: An energy load sampling unit, used for sampling the predicted energy load data according to the time granularity to obtain a predicted energy load time series; the predicted energy load time series includes a plurality of predicted energy load values ​​sorted by time; An energy load value prediction unit, used for inputting a plurality of the predicted energy load values ​​in the predicted energy load time series into a trained power grid load prediction model to obtain a plurality of predicted power grid load values; A predicted power grid load time series generating unit, used for sorting the plurality of predicted power grid load values ​​in time order to generate a predicted power grid load time series; The drift verification unit is used to perform drift verification on the predicted power grid load time series according to the historical energy load time series, and use the predicted power grid load time series that passes the verification as the predicted power grid load data.

Images