Regional load prediction method, system, equipment and medium

Abstract

The invention discloses a regional load prediction method, system, equipment and medium, and particularly relates to the technical field of power load prediction, and the technical key points are as follows: a plurality of load nodes are clustered in sequence according to weather data, population flow data and urban power consumption characteristics by using a third-order clustering analysis method to obtain a tree-type clustering cluster topological structure; a representative load node is selected from each cluster subset, average correlation coefficients between influence features in weather data of each representative load node and load data are calculated, and the influence features of the first three maximum average correlation coefficients are selected as influence input features of the representative load node; performing modal decomposition on the load data representing the load nodes by using an EMD decomposition method to obtain a plurality of connotation modal components and residual terms; and inputting the influence input feature of each representative load node and the plurality of decomposed connotation mode components into a load prediction model to obtain a load prediction result corresponding to each representative load node.

Description

Technical Field

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

[0002] Power load forecasting is an important part of smart grid construction. Power companies need to understand power consumption patterns and power demand through load forecasting, so as to ensure sufficient power supply and take into account economic benefits. Regional load forecasting can guide power production, planning and scheduling, and its importance is increasing, so higher requirements are placed on the accuracy of regional load forecasting.

[0003] However, in the process of regional power load forecasting, there are too many urban areas in some regions, resulting in excessive load data and influencing characteristics. At the same time, the influencing characteristics of different urban areas in the region vary greatly. If all load data and all influencing characteristics are input, the complexity of the prediction calculation will increase due to the large amount of data; if the average value is directly taken as the influencing characteristic of the region, it will not be representative enough, resulting in reduced prediction accuracy.

[0004] Therefore, the present invention aims to provide a regional load forecasting method, system, device and medium to solve the above-mentioned related problems. Summary of the invention

[0005] The technical problem to be solved by the present invention is that due to the excessive number of urban areas in a region, the amount of load data and the amount of influencing characteristics are too large, which leads to an increase in the complexity of prediction calculations. At the same time, there are differences in the influencing characteristics of different urban areas. The existing technology cannot comprehensively consider the influencing characteristics of different urban areas, resulting in a technical problem of reduced prediction accuracy. The purpose is to provide a regional load prediction method, system, equipment and medium, which uses a third-order clustering analysis method to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics to obtain a tree-type clustering cluster topology structure, and realizes clustering of multiple load nodes by comprehensively considering the similarity of multiple influencing characteristics, and then selects a load node from the multiple cluster subsets at the bottom of the tree-type clustering cluster topology structure to represent the load data pattern of the cluster subset. The load forecast is carried out according to the law and influencing characteristic law, so as to reduce the calculation complexity of the prediction model and comprehensively consider the differences in the influencing characteristics of different urban areas; at the same time, the urban electricity consumption characteristics are added in the clustering process so that cities with similar urban electricity consumption patterns can be clustered together, thereby increasing the regularity of the load data and increasing the representativeness of the load nodes; at the same time, the modal decomposition method is combined to perform modal decomposition on the load data representing the load nodes to reduce the randomness of the load and increase the prediction accuracy; the technical problem that the too many urban areas in the region, the large amount of load data and the amount of influencing characteristics, which leads to the increase of the prediction calculation complexity, and the differences in the influencing characteristics of different urban areas, and the inability of the existing technology to comprehensively consider the influencing characteristics of different urban areas, resulting in reduced prediction accuracy is solved.

[0006] The present invention is achieved through the following technical solutions:

[0007] A regional load forecasting method, the method comprising:

[0008] The load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region are obtained respectively, and the multiple load nodes are clustered in turn according to the weather data, population flow data and urban electricity consumption characteristics using the third-order cluster analysis method to obtain a tree-type clustering cluster topology structure, wherein the bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets;

[0009] Select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset, calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node;

[0010] The EMD decomposition method is used to perform modal decomposition on the load data representing the load nodes to obtain multiple intrinsic modal components and residual terms;

[0011] Input the impact input features of each representative load node and the decomposed multiple intrinsic modal components into the load forecasting model to obtain the load forecasting result corresponding to each representative load node; wherein, the load forecasting model is obtained by training a multi-modal three-dimensional convolutional neural network;

[0012] The load forecast results of each representative load node are averaged and processed to obtain the regional load forecast result.

[0013] Furthermore, the third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics, and a tree-type clustering cluster topology structure is obtained. The bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets, specifically:

[0014] In the region, a K-means algorithm is used to cluster multiple load nodes in the region according to the Pearson correlation coefficient of weather data of two adjacent load nodes to obtain multiple first clusters;

[0015] In each first cluster, a K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of population flow data of two adjacent load nodes to obtain multiple second clusters;

[0016] In each second cluster, the K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of the urban electricity consumption characteristics of two adjacent load nodes to obtain multiple cluster subsets, where the urban electricity consumption characteristics refer to the proportion of industrial, commercial and residential loads within the load node.

[0017] Furthermore, the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data is calculated, and the influencing features with the top three largest average correlation coefficients are selected as the influencing input features of the representative load node, where the influencing features in the weather data include temperature, rainfall, relative humidity, cloud type, wind speed, air pressure and dew point, specifically:

[0018] Calculate the Pearson correlation coefficients of the temperature data, rainfall data, relative humidity data, cloud type, wind speed data, air pressure data and dew point data in the weather data representing each load node and the load data corresponding to the load node, and obtain multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud type, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients and multiple dew point correlation coefficients;

[0019] Respectively averaging multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud layer correlation coefficients, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients, and multiple dew point correlation coefficients to obtain an average temperature correlation coefficient, an average rainfall correlation coefficient, an average relative humidity correlation coefficient, an average cloud layer correlation coefficient, an average wind speed correlation coefficient, an average air pressure correlation coefficient, and an average dew point correlation coefficient;

[0020] Compare the absolute values ​​of the average temperature correlation coefficient, the average precipitation correlation coefficient, the average relative humidity correlation coefficient, the average cloud layer correlation coefficient, the average wind speed correlation coefficient, the average air pressure correlation coefficient and the average dew point correlation coefficient, and obtain the influence characteristics of the first three maximum values ​​as the influence input characteristics of the representative load node.

[0021] Furthermore, the impact input features of each representative load node and the decomposed multiple intrinsic modal components are input into the load forecasting model to obtain the load forecasting result corresponding to each representative load node; wherein, the load forecasting model is obtained by training a multimodal three-dimensional convolutional neural network, specifically:

[0022] The load forecasting model includes an input layer, multiple three-dimensional dense modules and an output layer connection, and a transition layer for transition connection is provided between each three-dimensional dense module; the input layer includes a splicing unit, a first convolution unit and a maximum pooling unit connected in sequence, and each three-dimensional dense module uses multiple second convolution units, and each second convolution unit in each three-dimensional dense module is connected to the second convolution unit of the next layer in a feedforward manner; the output layer includes an average pooling unit and a rectified linear unit connected in sequence; the transition layer includes a third convolution unit and an average pooling unit connected in sequence.

[0023] The present invention further provides a regional load forecasting system, which is used in any one of the regional load forecasting methods described above, and the system comprises:

[0024] The third-order clustering module is used to obtain the load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region respectively, and use the third-order clustering analysis method to cluster multiple load nodes in turn according to the weather data, population flow data and urban electricity consumption characteristics to obtain a tree-type clustering cluster topology structure, wherein the bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets;

[0025] A feature selection module is used to select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset, calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node;

[0026] The modal decomposition module is used to perform modal decomposition on the load data representing the load node using the EMD decomposition method to obtain multiple intrinsic modal components and residual terms;

[0027] The node prediction module is used to input the impact input characteristics of each representative load node and the decomposed multiple intrinsic modal components into the load prediction model to obtain the load prediction result corresponding to each representative load node; wherein the load prediction model is obtained by training a multi-modal three-dimensional convolutional neural network;

[0028] The regional prediction module is used to average the load prediction results of each representative load node to obtain the regional load prediction result.

[0029] Furthermore, the third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics, and a tree-type clustering cluster topology structure is obtained. The bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets, specifically:

[0030] Clustering multiple load nodes in the region according to the Pearson correlation coefficient of weather data of two adjacent load nodes in the region to obtain multiple first clustering clusters;

[0031] In each first cluster, a K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of population flow data of two adjacent load nodes to obtain multiple second clusters;

[0032] In each second cluster, the K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of the urban electricity consumption characteristics of two adjacent load nodes to obtain multiple cluster subsets, where the urban electricity consumption characteristics refer to the proportion of industrial, commercial and residential loads within the load node.

[0033] Furthermore, the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data is calculated, and the influencing features with the top three largest average correlation coefficients are selected as the influencing input features of the representative load node, where the influencing features in the weather data include temperature, rainfall, relative humidity, cloud type, wind speed, air pressure and dew point, specifically:

[0034] Calculate the Pearson correlation coefficients of the temperature data, rainfall data, relative humidity data, cloud type, wind speed data, air pressure data and dew point data in the weather data representing each load node and the load data corresponding to the load node, and obtain multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud type, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients and multiple dew point correlation coefficients;

[0035] Respectively averaging multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud layer correlation coefficients, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients, and multiple dew point correlation coefficients to obtain an average temperature correlation coefficient, an average rainfall correlation coefficient, an average relative humidity correlation coefficient, an average cloud layer correlation coefficient, an average wind speed correlation coefficient, an average air pressure correlation coefficient, and an average dew point correlation coefficient;

[0036] Compare the absolute values ​​of the average temperature correlation coefficient, the average precipitation correlation coefficient, the average relative humidity correlation coefficient, the average cloud layer correlation coefficient, the average wind speed correlation coefficient, the average air pressure correlation coefficient and the average dew point correlation coefficient, and obtain the influence characteristics of the first three maximum values ​​as the influence input characteristics of the representative load node.

[0037] The present invention also provides a computer device, comprising a system memory and a processor, wherein the system memory stores a computer program, and wherein the processor implements the steps of any one of the above-described methods when executing the computer program.

[0038] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of any one of the methods described above when executed by a processor.

[0039] The present invention also provides a computer program product comprising instructions, wherein when the instructions are executed by a computer device cluster, the computer device cluster executes any of the above methods.

[0040] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0041] In the present invention, a third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics to obtain a tree-type clustering cluster topology structure, which realizes clustering of multiple load nodes by comprehensively considering the similarity of multiple influencing characteristics, and then selects a load node from multiple cluster subsets at the bottom of the tree-type clustering cluster topology structure to represent the load data law and influencing characteristic law of the cluster subset for load prediction, thereby reducing the calculation complexity of the prediction model while comprehensively considering the differences in influencing characteristics of different urban areas; at the same time, urban electricity consumption characteristics are added in the clustering process so that cities with similar urban electricity consumption patterns can be clustered together, thereby increasing the regularity of load data and increasing the representativeness of representative load nodes; at the same time, the modal decomposition method is combined to perform modal decomposition on the load data representing the load node to reduce the randomness of the load and increase the prediction accuracy; the technical problem that the complexity of prediction calculation increases due to too many urban areas in the region, the amount of load data and the amount of influencing characteristics are too large, and at the same time, the influencing characteristics of different urban areas are different, and the existing technology cannot comprehensively consider the influencing characteristics of different urban areas, resulting in reduced prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0043] Figure 1 This is a method flow chart of a regional load forecasting method in this embodiment;

[0044] Figure 2 It is a schematic diagram of the structure of a tree-type clustering cluster topology structure obtained by three-order clustering in a regional load forecasting method in this embodiment;

[0045] Figure 3 is a structural schematic diagram of a load prediction model in a regional load prediction method in this embodiment;

[0046] Figure 4 Schematic diagram of the structure of the input layer of the load forecasting model in a regional load forecasting method in this embodiment;

[0047] Figure 5 It is a schematic structural diagram of a three-dimensional dense module of a load prediction model in a regional load prediction method in this embodiment;

[0048] Figure 6Schematic diagram of the structure of a transition layer of a load forecasting model in a regional load forecasting method in this embodiment;

[0049] Figure 7 Schematic diagram of the structure of the output layer of the load forecasting model in a regional load forecasting method in this embodiment;

[0050] Figure 8 This is a schematic diagram of module connections of a regional load forecasting system in this embodiment;

[0051] Fig. 9 It is a structural schematic diagram of a computer device in this embodiment. DETAILED DESCRIPTION

[0052] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0053] In the present disclosure, unless otherwise specified, the use of the terms "first", "second", etc. to describe various elements is not intended to limit the positional relationship, timing relationship, or importance relationship of these elements, and such terms are only used to distinguish one element from another element. In some examples, the first element and the second element may refer to the same instance of the element, and in some cases, based on the description of the context, they may also refer to different instances.

[0054] The terms used in the description of various examples in this disclosure are only for the purpose of describing specific examples and are not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element can be one or more. In addition, the term "and / or" used in this disclosure covers any one of the listed items and all possible combinations.

[0055] Example 1

[0056] See also Figure 1-Figure 2 As shown, this embodiment provides a regional load forecasting method, the method comprising:

[0057] S1: Obtain the load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region respectively, and use the third-order cluster analysis method to cluster multiple load nodes in turn according to the weather data, population flow data and urban electricity consumption characteristics to obtain a tree-type clustering cluster topology structure, wherein the bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets, specifically:

[0058] Obtain load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region respectively;

[0059] It should be noted that, in this embodiment, the region includes multiple urban areas, and each load node represents an urban area, wherein the load data of each urban area consists of the load data of the last two years, and a data point is obtained at an interval of 30 minutes every day, that is, there are 48 data points every day, and the load data of an urban area has 35040 data points; the weather data of each urban area includes temperature data, rainfall data, relative humidity data, cloud type data, wind speed data, air pressure data and dew point data, and each type of weather data is consistent with the load data, consisting of the data of the last two years, and a data point is obtained at an interval of 30 minutes every day.

[0060] In the region, a K-means algorithm is used to cluster multiple load nodes in the region according to the Pearson correlation coefficient of weather data of two adjacent load nodes to obtain multiple first clusters;

[0061] It should be noted that, in this embodiment, since the correlation between weather data between load nodes is strong, the Pearson correlation coefficient is selected as the similarity metric, and the K-means algorithm is used for clustering operation. The clustering process is specifically as follows: first, the number of the first clustering clusters is determined by the elbow rule. In this embodiment, the number is two, and in other embodiments, three, four or five can also be selected, which will not be described in detail here; then, a load node i is randomly selected from the region as a central load node, and the Pearson correlation coefficient of the weather data of each load node i in the region and the central load node is calculated. The Pearson correlation coefficient calculation formula is: Where r represents the Pearson correlation coefficient between two load nodes; X and Y represent the weather data series of the two load nodes; n is the number of data points; Represents the mean of the weather data series of the two load nodes; the load node with the largest difference in Pearson correlation coefficient is taken as the next central load node, and then clustering is performed according to the Pearson correlation coefficient of the weather data of the load node and the two central load nodes, and finally two first clustering clusters are obtained.

[0062] In each first cluster, a K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of population flow data of two adjacent load nodes to obtain multiple second clusters;

[0063] It should be noted that, in this embodiment, since the correlation between the population flow data between the load nodes is low, the Euclidean distance is selected as the similarity metric, and the K-means algorithm is used for clustering operation. The clustering process is specifically as follows: first, the number of the second clustering clusters is determined by the elbow rule. In this embodiment, the number is two, and in other embodiments, three, four or five can also be selected, which will not be described in detail here; a load node is randomly selected from the first clustering cluster as a central load node, and the Euclidean distance of the population flow data between each load node in the area and the central load node is calculated. The Euclidean distance calculation formula is: Where D represents the Euclidean distance between two load nodes; a i represents the i-th load node in the first cluster; u r Represents the selected central load node; the load node with the largest Euclidean distance difference is taken as the next central load node, and then clustering is performed according to the Euclidean distance between the load node and the two central load nodes, and finally two second clusters are obtained under each first cluster.

[0064] In each second cluster, the K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of the urban electricity consumption characteristics of two adjacent load nodes to obtain multiple cluster subsets;

[0065] It should be noted that, in this embodiment, the urban electricity consumption characteristics specifically refer to the proportion of industrial, commercial and residential loads in urban areas, which can make the representative load nodes selected in the next step more representative; at the same time, it should be noted that, similar to the Euclidean distance clustering method based on the population flow data of two adjacent load nodes, two cluster subsets are finally obtained under each second cluster cluster, and finally eight cluster subsets are obtained.

[0066] S2: Select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset, calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node, where the influencing features in the weather data include temperature, rainfall, relative humidity, cloud type, wind speed, air pressure and dew point, specifically:

[0067] Select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset;

[0068] It should be noted that, in this embodiment, the load nodes with the largest load data are respectively selected from the eight cluster subsets as the representative load nodes of the cluster subsets.

[0069] Calculate the Pearson correlation coefficients of the temperature data, rainfall data, relative humidity data, cloud type, wind speed data, air pressure data and dew point data in the weather data representing each load node and the load data corresponding to the load node, and obtain multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud type, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients and multiple dew point correlation coefficients;

[0070] Respectively averaging multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud layer correlation coefficients, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients, and multiple dew point correlation coefficients to obtain an average temperature correlation coefficient, an average rainfall correlation coefficient, an average relative humidity correlation coefficient, an average cloud layer correlation coefficient, an average wind speed correlation coefficient, an average air pressure correlation coefficient, and an average dew point correlation coefficient;

[0071] Compare the absolute values ​​of the average temperature correlation coefficient, the average rainfall correlation coefficient, the average relative humidity correlation coefficient, the average cloud layer correlation coefficient, the average wind speed correlation coefficient, the average air pressure correlation coefficient, and the average dew point correlation coefficient, and obtain the influence characteristics of the first three maximum values ​​as the influence input characteristics of the representative load node;

[0072] S3: Use EMD decomposition method to perform modal decomposition on the load data representing the load node to obtain multiple intrinsic modal components and residual terms;

[0073] It should be noted that, in this embodiment, the EMD decomposition method is used to perform modal decomposition on the load data representing the load node to obtain multiple intrinsic modal components and residual terms, specifically: A = IMF1 + IMF2 + ... + IMF m +Residual; A represents the original load data; IMF m Represents the m intrinsic modal components obtained by decomposition; Residual represents the residual term; the decomposition process is specifically as follows: find the maximum and minimum data points of the load data to form the maximum envelope z1 and the minimum envelope z2, average the maximum envelope z1 and the minimum envelope z2 to obtain z3, subtract z3 from the load data to obtain z4, if z4 meets the requirements of the intrinsic mode function (the number of extreme points and zero-crossing points is equal or differs by at most one, and at any time, the mean of the upper and lower envelopes composed of local maximum points and local minimum points is zero), the decomposition is completed, and z4 is taken as the first intrinsic modal component. If it does not meet the requirements, repeat the above steps until the first intrinsic modal component is obtained; then subtract the first intrinsic modal component from the load data to form new data and repeat the above steps, and finally obtain multiple intrinsic modal components and residual terms.

[0074] S4: Input the impact input features of each representative load node and the decomposed multiple connotation modal components into the load forecasting model to obtain the load forecasting result corresponding to each representative load node; wherein the load forecasting model is obtained by training a multi-modal three-dimensional convolutional neural network;

[0075] It should be noted that, in this embodiment, see Figure 3-Figure 7 As shown, the load forecasting model includes an input layer, four three-dimensional dense modules and an output layer connection, and a transition layer for transition connection is provided between each three-dimensional dense module; the input layer includes a splicing unit, a first convolution unit and a maximum pooling unit connected in sequence, and five second convolution units are used in each three-dimensional dense module. Each second convolution unit in each three-dimensional dense module is connected to the second convolution unit of the next layer in a feedforward manner; the output layer includes an average pooling unit and a rectified linear unit connected in sequence; the transition layer includes a third convolution unit and an average pooling unit connected in sequence;

[0076] At the same time, it should be noted that the first convolution unit adopts a convolution operation with a convolution kernel of 7×7×7 and a step size of 2, the maximum pooling unit adopts a maximum pooling operation with a convolution kernel of 3×3×3 and a step size of 2, and each second convolution unit is also provided with a ReLU activation function and BN batch normalization, and the second convolution unit adopts a convolution operation with a convolution kernel of 3×3×3; the average pooling unit adopts an average pooling operation with a convolution kernel of 2×2×2 and a step size of 2.

[0077] Specifically, in this embodiment, the impact input features of each representative load node, the decomposed multiple intrinsic modal components and the time series are input into the load prediction model obtained by training the multimodal three-dimensional convolutional neural network, and the multi-channel technology of the three-dimensional convolutional neural network is used to stack data of different modes into different data channels, thereby achieving the effect of fusing multimodal data. The early fusion strategy adopted by the multi-channel technology can more closely fuse the information of multiple modes in deeper network layers, so that the data information of different modes can be more effectively integrated.

[0078] S5: Calculate and process the load forecast results of each representative load node on average to obtain the regional load forecast result.

[0079] It should be noted that the load forecast result of each representative load node is averaged and processed to obtain the load forecast result of the region.

[0080] Specifically, in this embodiment, a third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics to obtain a tree-type clustering cluster topology structure, which realizes clustering of multiple load nodes by comprehensively considering the similarity of multiple influencing characteristics, and then selects a load node from multiple cluster subsets at the bottom of the tree-type clustering cluster topology structure to represent the load data law and influencing characteristic law of the cluster subset for load prediction, thereby reducing the calculation complexity of the prediction model while comprehensively considering the differences in influencing characteristics of different urban areas; at the same time, urban electricity consumption characteristics are added in the clustering process so that cities with similar urban electricity consumption patterns can be clustered together, thereby increasing the regularity of load data and increasing the representativeness of representative load nodes; at the same time, the modal decomposition method is combined to perform modal decomposition on the load data representing the load node to reduce the randomness of the load and increase the prediction accuracy; the technical problem that the complexity of prediction calculation increases due to too many urban areas in the region, the amount of load data and the amount of influencing characteristics are too large, and the influencing characteristics of different urban areas are different, and the existing technology cannot comprehensively consider the influencing characteristics of different urban areas, resulting in reduced prediction accuracy.

[0081] Example 2

[0082] See also Figure 3 As shown, the present invention also provides a regional load forecasting system, which is used in any one of the regional load forecasting methods described above, and the system includes:

[0083] The third-order clustering module 100 is used to respectively obtain the load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region, and cluster the multiple load nodes in turn according to the weather data, population flow data and urban electricity consumption characteristics using the third-order clustering analysis method to obtain a tree-type clustering cluster topology structure, wherein the bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets;

[0084] The feature selection module 200 is used to select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset, calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the first three largest average correlation coefficients as the influencing input features of the representative load node;

[0085] The modal decomposition module 300 is used to perform modal decomposition on the load data representing the load node using the EMD decomposition method to obtain multiple intrinsic modal components and residual terms;

[0086] The node prediction module 400 is used to input the impact input characteristics of each representative load node and the decomposed multiple connotation modal components into the load prediction model to obtain the load prediction result corresponding to each representative load node; wherein the load prediction model is obtained by training a multi-modal three-dimensional convolutional neural network;

[0087] The regional prediction module 500 is used to average the load prediction results of each representative load node to obtain a regional load prediction result.

[0088] Furthermore, the third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics, and a tree-type clustering cluster topology structure is obtained. The bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets, specifically:

[0089] In the region, a K-means algorithm is used to cluster multiple load nodes in the region according to the Pearson correlation coefficient of weather data of two adjacent load nodes to obtain multiple first clusters;

[0090] In each first cluster, a K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of population flow data of two adjacent load nodes to obtain multiple second clusters;

[0091] In each second cluster, the K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of the urban electricity consumption characteristics of two adjacent load nodes to obtain multiple cluster subsets, where the urban electricity consumption characteristics refer to the proportion of industrial, commercial and residential loads within the load node.

[0092] Furthermore, the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data is calculated, and the influencing features with the top three largest average correlation coefficients are selected as the influencing input features of the representative load node, where the influencing features in the weather data include temperature, rainfall, relative humidity, cloud type, wind speed, air pressure and dew point, specifically:

[0093] Calculate the Pearson correlation coefficients of the temperature data, rainfall data, relative humidity data, cloud type, wind speed data, air pressure data and dew point data in the weather data representing each load node and the load data corresponding to the load node, and obtain multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud type, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients and multiple dew point correlation coefficients;

[0094] Respectively averaging multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud layer correlation coefficients, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients, and multiple dew point correlation coefficients to obtain an average temperature correlation coefficient, an average rainfall correlation coefficient, an average relative humidity correlation coefficient, an average cloud layer correlation coefficient, an average wind speed correlation coefficient, an average air pressure correlation coefficient, and an average dew point correlation coefficient;

[0095] Compare the absolute values ​​of the average temperature correlation coefficient, the average precipitation correlation coefficient, the average relative humidity correlation coefficient, the average cloud layer correlation coefficient, the average wind speed correlation coefficient, the average air pressure correlation coefficient and the average dew point correlation coefficient, and obtain the influence characteristics of the first three maximum values ​​as the influence input characteristics of the representative load node.

[0096] It should be noted that the modules in the system of Example 2 correspond to the steps in the method of Example 1. The steps in the method of Example 1 have been described in detail in Example 1. In this Example 2, the contents of the modules in the system will not be described in detail.

[0097] Example 3

[0098] See also Figure 4 As shown, this embodiment further provides a computer device, including a system memory 1005 and a processor 1001, wherein the system memory 1005 stores a computer program, and the processor 1001 implements the steps of any one of the above methods when executing the computer program.

[0099] It should be noted that the processor 1001 is used to execute the steps in the above method embodiments according to the instructions in the program code. Alternatively, the processor 1001 implements the functions of each module / unit in the above system / device embodiments when executing the computer program.

[0100] Specifically, in this embodiment, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the system memory 1005, and are executed by the processor 1001 to complete the present application. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0101] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor 1001 and a system memory 1005. Those skilled in the art will appreciate that this does not constitute a limitation on the terminal device, and may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the terminal device may also include an input / output device 1003, a network access device 1002, a bus 1006, etc.

[0102] The processor 1001 may be a central processing unit (CPU), or other general-purpose processors 1001, digital signal processors 1001 (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor 1001 may be a microprocessor or any conventional processor, etc.

[0103] The system memory 1005 may be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. The system memory 1005 may also be a storage device 1004 of the terminal device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the terminal device. Further, the system memory 1005 may also include both the internal storage unit of the terminal device and the storage device 1004. The system memory 1005 is used to store computer programs and other programs and data required by the terminal device. The system memory 1005 may also be used to temporarily store data that has been output or is to be output.

[0104] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0105] Example 4

[0106] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the above methods are implemented.

[0107] Among them, the computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, system or device, or any combination of the above. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk. Random Access Memory (RAM), Read-Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), registers, hard disks, optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above, or any other form of computer-readable storage medium known in the art.

[0108] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application-specific integrated circuit (ASIC). In an embodiment of the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system, or device.

[0109] Example 5

[0110] This embodiment also provides a computer program product including instructions. When the instructions are executed by a computer device cluster, the computer device cluster executes the method described in Embodiment 1.

[0111] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A regional load forecasting method, characterized in that: Methods include: The load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region are obtained respectively, and the multiple load nodes are clustered in turn according to the weather data, population flow data and urban electricity consumption characteristics using the third-order cluster analysis method to obtain a tree-type clustering cluster topology structure, wherein the bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets; Select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset, calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node; The EMD decomposition method is used to perform modal decomposition on the load data representing the load nodes to obtain multiple intrinsic modal components and residual terms; Input the impact input features of each representative load node and the decomposed multiple intrinsic modal components into the load forecasting model to obtain the load forecasting result corresponding to each representative load node; wherein, the load forecasting model is obtained by training a multi-modal three-dimensional convolutional neural network; The load forecast results of each representative load node are averaged and processed to obtain the regional load forecast result.

2. A regional load forecasting method according to claim 1, characterized in that: The third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics, and a tree-type clustering cluster topology structure is obtained. The bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets, specifically: In the region, a K-means algorithm is used to cluster multiple load nodes in the region according to the Pearson correlation coefficient of weather data of two adjacent load nodes to obtain multiple first clusters; In each first cluster, a K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of population flow data of two adjacent load nodes to obtain multiple second clusters; In each second cluster, the K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of the urban electricity consumption characteristics of two adjacent load nodes to obtain multiple cluster subsets, where the urban electricity consumption characteristics refer to the proportion of industrial, commercial and residential loads within the load node.

3. A regional load forecasting method according to claim 1, characterized in that: Calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node. The influencing features in the weather data include temperature, rainfall, relative humidity, cloud type, wind speed, air pressure and dew point, specifically: Calculate the Pearson correlation coefficients of the temperature data, rainfall data, relative humidity data, cloud type, wind speed data, air pressure data and dew point data in the weather data representing each load node and the load data corresponding to the load node, and obtain multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud type, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients and multiple dew point correlation coefficients; Respectively averaging multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud layer correlation coefficients, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients, and multiple dew point correlation coefficients to obtain an average temperature correlation coefficient, an average rainfall correlation coefficient, an average relative humidity correlation coefficient, an average cloud layer correlation coefficient, an average wind speed correlation coefficient, an average air pressure correlation coefficient, and an average dew point correlation coefficient; Compare the absolute values ​​of the average temperature correlation coefficient, the average precipitation correlation coefficient, the average relative humidity correlation coefficient, the average cloud layer correlation coefficient, the average wind speed correlation coefficient, the average air pressure correlation coefficient and the average dew point correlation coefficient, and obtain the influence characteristics of the first three maximum values ​​as the influence input characteristics of the representative load node.

4. A regional load forecasting method according to claim 1, characterized in that: The impact input features of each representative load node and the decomposed multiple intrinsic modal components are input into the load forecasting model to obtain the load forecasting result corresponding to each representative load node; wherein, the load forecasting model is obtained by training a multimodal three-dimensional convolutional neural network, specifically: The load forecasting model includes an input layer, multiple three-dimensional dense modules and an output layer connection, and a transition layer for transition connection is provided between each three-dimensional dense module; the input layer includes a splicing unit, a first convolution unit and a maximum pooling unit connected in sequence, and each three-dimensional dense module uses multiple second convolution units, and each second convolution unit in each three-dimensional dense module is connected to the second convolution unit of the next layer in a feedforward manner; the output layer includes an average pooling unit and a rectified linear unit connected in sequence; the transition layer includes a third convolution unit and an average pooling unit connected in sequence.

5. A regional load forecasting system, characterized in that: The system is used in a regional load forecasting method according to any one of claims 1 to 4, and the system comprises: The third-order clustering module is used to obtain the load data, weather data, population flow data and urban electricity consumption characteristics of multiple load nodes in the region respectively, and use the third-order clustering analysis method to cluster multiple load nodes in turn according to the weather data, population flow data and urban electricity consumption characteristics to obtain a tree-type clustering cluster topology structure, wherein the bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets; A feature selection module is used to select the load node with the largest load data from each cluster subset as the representative load node of the cluster subset, calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node; The modal decomposition module is used to perform modal decomposition on the load data representing the load node using the EMD decomposition method to obtain multiple intrinsic modal components and residual terms; The node prediction module is used to input the impact input characteristics of each representative load node and the decomposed multiple intrinsic modal components into the load prediction model to obtain the load prediction result corresponding to each representative load node; wherein the load prediction model is obtained by training a multi-modal three-dimensional convolutional neural network; The regional prediction module is used to average the load prediction results of each representative load node to obtain the regional load prediction result.

6. A regional load forecasting system according to claim 5, characterized in that: The third-order cluster analysis method is used to cluster multiple load nodes in turn according to weather data, population flow data and urban electricity consumption characteristics, and a tree-type clustering cluster topology structure is obtained. The bottom layer of the tree-type clustering cluster topology structure includes multiple cluster subsets, specifically: In the region, a K-means algorithm is used to cluster multiple load nodes in the region according to the Pearson correlation coefficient of weather data of two adjacent load nodes to obtain multiple first clusters; In each first cluster, a K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of population flow data of two adjacent load nodes to obtain multiple second clusters; In each second cluster, the K-means algorithm is used to cluster multiple load nodes according to the Euclidean distance of the urban electricity consumption characteristics of two adjacent load nodes to obtain multiple cluster subsets, where the urban electricity consumption characteristics refer to the proportion of industrial, commercial and residential loads within the load node.

7. A regional load forecasting system according to claim 5, characterized in that: Calculate the average correlation coefficient between the influencing features in the weather data of each representative load node and the load data, and select the influencing features with the top three largest average correlation coefficients as the influencing input features of the representative load node. The influencing features in the weather data include temperature, rainfall, relative humidity, cloud type, wind speed, air pressure and dew point, specifically: Calculate the Pearson correlation coefficients of the temperature data, rainfall data, relative humidity data, cloud type, wind speed data, air pressure data and dew point data in the weather data representing each load node and the load data corresponding to the load node, and obtain multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud type, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients and multiple dew point correlation coefficients; Respectively averaging multiple temperature correlation coefficients, multiple rainfall correlation coefficients, multiple relative humidity correlation coefficients, multiple cloud layer correlation coefficients, multiple wind speed correlation coefficients, multiple air pressure correlation coefficients, and multiple dew point correlation coefficients to obtain an average temperature correlation coefficient, an average rainfall correlation coefficient, an average relative humidity correlation coefficient, an average cloud layer correlation coefficient, an average wind speed correlation coefficient, an average air pressure correlation coefficient, and an average dew point correlation coefficient; Compare the absolute values ​​of the average temperature correlation coefficient, the average precipitation correlation coefficient, the average relative humidity correlation coefficient, the average cloud layer correlation coefficient, the average wind speed correlation coefficient, the average air pressure correlation coefficient and the average dew point correlation coefficient, and obtain the influence characteristics of the first three maximum values ​​as the influence input characteristics of the representative load node.

8. A computer device comprising a system memory and a processor, wherein the system memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

10. A computer program product comprising instructions, characterized in that When the instructions are executed by a computer device cluster, the computer device cluster executes the method according to any one of claims 1 to 4.

Images