Regional electrical load prediction method and device based on load density and data time sequence

Abstract

The invention discloses a regional electrical load prediction method and device based on load density and a data time sequence, and the method comprises the steps: dividing a research region into a plurality of set regions, and collecting and obtaining parameter data of the set regions; according to the parameter data, electric load density is obtained through calculation; carrying out feature screening by adopting XGBoost to obtain a feature factor which has the most influence on the electric load density; performing data enhancement through a conditional generative adversarial network, and generating identically distributed electrical load density and characteristic variable data; combining the real data of the electric load density and the characteristic variable of the set area with the electric load density and the characteristic variable data of the same distribution generated by enhancement to obtain a new data set; constructing an electric load density time sequence prediction model, and predicting future electric load density; and calculating to obtain an electric load prediction value of the set area, and summarizing and summing to obtain an electric load total amount prediction value of the research area. According to the method, the contribution degree of regional land use planning information and each time position of the characteristic variable can be considered, and the accuracy of electric load prediction is improved.

Description

Technical Field

[0001] The present invention relates to the field of new energy and energy-saving technologies, and in particular to a method and device for predicting regional electric load based on load density and data time series. Background Art

[0002] Accurate regional electricity load forecasting can reduce regional energy waste, optimize regional power supply, and improve the operating efficiency of regional power grids. At the same time, electricity load forecasting is also one of the key tools for power grid modernization and sustainable development, which helps to better integrate regional renewable energy, promote energy efficiency, and achieve smarter and more flexible regional power and energy management and planning.

[0003] At present, with the application of machine learning, deep learning and artificial intelligence technology, the accuracy and efficiency of power load forecasting have been improved. These technologies can process large data sets, learn from historical patterns and adapt to new trends, and improve the accuracy of forecasts. However, these methods mainly model and predict the total power load based on historical influencing factors such as population and economy, without considering regional land use planning information, and cannot obtain the prediction results of the spatial distribution of power load, which has little guiding significance for regional power development.

[0004] In addition, the model built for electric load forecasting focuses on the selection of characteristic variables, and almost ignores the temporal influence of the historical data of the selected characteristic variables, and does not consider the different contributions of the characteristic variables at different time positions, resulting in a lack of accuracy in electric load forecasting.

[0005] Therefore, how to invent a regional electric load forecasting method that takes into account the different contributions of characteristic variables at different time positions while considering regional land use planning information and improves the accuracy of electric load forecasting has become an urgent problem to be solved. Summary of the invention

[0006] To this end, the present invention provides a regional electric load prediction method and device based on load density and data time series, which can take into account the different contributions of characteristic variables at different time positions while considering regional land use planning information, thereby improving the accuracy of electric load prediction.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a regional electric load prediction method based on load density and data time series, comprising:

[0008] Dividing the study area into a number of set areas according to the land use type; collecting data on the set areas to obtain parameter data of the set areas;

[0009] According to the obtained parameter data of the set area, the electric load density of the set area is calculated; XGBoost is used to perform feature screening on factors affecting the electric load density of the set area to obtain the most influential feature factors on the electric load density of the set area;

[0010] Performing data enhancement on the real data of the electric load density and characteristic variables of the set area through a conditional generative adversarial network to generate the electric load density and characteristic variable data of the set area with the same distribution;

[0011] Combining the real data of the electric load density and characteristic variables of the set area with the enhanced and generated electric load density and characteristic variable data of the set area with the same distribution to obtain a new data set;

[0012] A time series prediction model for electric load density is constructed according to an improved autoregressive distributed lag algorithm; the future electric load density of the set area is predicted by solving the contribution of time and position characteristic variables to the time series prediction model for electric load density;

[0013] The predicted electric load value of the set area is calculated based on the future electric load density of the set area and the parameter data of the set area; the predicted electric load values ​​of several set areas are summarized and summed to obtain the predicted total electric load value of the study area.

[0014] As a preferred solution of the regional electric load prediction method based on load density and data time series, in the process of dividing the study area into several set areas according to land use types, the set areas include: agricultural land, industrial land, residential land, commercial land and hospital land.

[0015] As a preferred solution of the regional electric load prediction method based on load density and data time series, in the process of collecting data for the set area and obtaining parameter data of the set area, the parameter data of the set area include: the electric load, area, factors affecting the electric load and future land use planning data of the set area.

[0016] As a preferred solution of the regional electric load prediction method based on load density and data time series, in the process of using XGBoost to perform feature screening on the factors affecting the electric load density of the set area, the important features affecting the electric load density are selected by XGBoost, and the contribution of each important feature to the electric load density can be obtained; the electric load density model is:

[0017] Y t =αA t +βB t +γC t +ε t

[0018] Where Y t is the electric load density; α, β, γ are all weight parameters; A t is the population characteristics; B t Economic characteristics; C t is the temperature characteristic; t is the data year; ε t is white noise.

[0019] As a preferred solution of the regional electric load prediction method based on load density and data time series, in the process of data enhancement of the electric load density and characteristic variable real data of the set area by the conditional generative adversarial network, the objective function expression of the conditional generative adversarial network is:

[0020]

[0021] Where D(x|y) is the probability of classifying x as real data rather than generated data under conditional label y; is the probability of classifying real data as real data; D(G(z|y)) is the probability of classifying generated data as real data; 1-D(G(z|y)) is the probability of classifying generated data as generated data; is the probability of classifying the generated data into the generated data.

[0022] As a preferred solution of the regional electric load prediction method based on load density and data time series, in the process of predicting the future electric load density of the set area by solving the contribution of the time position characteristic variable of the electric load density time series prediction model, the expression of the future electric load density of the set area is:

[0023]

[0024] In the formula, w i , σ i , i , are all weight parameters; p is the autoregressive order of Y; q, r, s are the lag orders of A, B, and C respectively;

[0025] Obtain the total power load forecast value T of the study area e The formula is:

[0026] E i =Y i *M i

[0027]

[0028] Where Y iis the predicted electric load density of each area; M i Land use planning data for each district; i is the predicted value of electricity load in each district; i is the specific land type divided in the study area; n is the specific number of land divided in the study area.

[0029] The present invention also provides a regional electric load prediction device based on load density and data time series, based on the above regional electric load prediction method based on load density and data time series, including:

[0030] The area division and parameter data acquisition module is used to divide the study area into several set areas according to the land use type; collect data for the set areas to obtain parameter data for the set areas;

[0031] A characteristic factor screening module is used to calculate the electric load density of the set area according to the obtained parameter data of the set area; use XGBoost to perform characteristic screening on the factors affecting the electric load density of the set area to obtain the characteristic factors that have the greatest influence on the electric load density of the set area;

[0032] A data enhancement module, used to enhance the real data of the electric load density and characteristic variables of the set area through a conditional generative adversarial network, and generate the electric load density and characteristic variable data of the set area with the same distribution;

[0033] A new data set acquisition module, used for combining the real data of the electric load density and characteristic variables of the set area with the enhanced and generated data of the electric load density and characteristic variables of the set area with the same distribution, to obtain a new data set;

[0034] A future electric load density prediction module is used to construct an electric load density time series prediction model according to an improved autoregressive distributed lag algorithm; by solving the contribution of time and position characteristic variables to the electric load density time series prediction model, the future electric load density of the set area is predicted;

[0035] The total electric load forecast value acquisition module is used to calculate the electric load forecast value of the set area according to the future electric load density of the set area and the parameter data of the set area; and to summarize and sum the electric load forecast values ​​of several set areas to obtain the total electric load forecast value of the study area.

[0036] As a preferred solution of the regional electric load prediction device based on load density and data timing, in the regional division and parameter data acquisition module, in the process of dividing the study area into a number of set areas according to land use types, the set areas include: agricultural land, industrial land, residential land, commercial land and hospital land.

[0037] As a preferred solution for the regional electric load prediction device based on load density and data timing, in the regional division and parameter data acquisition module, in the process of collecting data for the set area and obtaining the parameter data of the set area, the parameter data of the set area include: the electric load, area, factors affecting the electric load and future land use planning data of the set area.

[0038] As a preferred solution of the regional electric load prediction device based on load density and data time series, in the characteristic factor screening module, in the process of using XGBoost to perform feature screening on the factors affecting the electric load density of the set area, the important features affecting the electric load density are selected by XGBoost, and the contribution of each important feature to the electric load density can be obtained; the electric load density model is:

[0039] Y t =αA t +βB t +γC t +ε t

[0040] Where Y t is the electric load density; α, β, γ are all weight parameters; A t is the population characteristic; B t Economic characteristics; C t is the temperature characteristic; t is the data year; ε t is white noise.

[0041] As a preferred solution of the regional electric load prediction device based on load density and data time series, in the data enhancement module, in the process of data enhancement of the electric load density and characteristic variable real data of the set area by the conditional generative adversarial network, the objective function expression of the conditional generative adversarial network is:

[0042]

[0043] Where D(x|y) is the probability of classifying x as real data rather than generated data under conditional label y; is the probability of classifying real data as real data; D(G(z|y)) is the probability of classifying generated data as real data; 1-D(G(z|y)) is the probability of classifying generated data as generated data; is the probability of classifying the generated data as generated data.

[0044] As a preferred solution of the regional electric load prediction device based on load density and data time series, in the future electric load density prediction module, in the process of predicting the future electric load density of the set area by solving the contribution of the time position characteristic variable of the electric load density time series prediction model, the expression of the future electric load density of the set area is:

[0045]

[0046] In the formula, w i , σ i , i , are all weight parameters; p is the autoregressive order of Y; q, r, s are the lag orders of A, B, and C respectively;

[0047] In the total electric load forecast value acquisition module, the total electric load forecast value T of the study area is obtained. e The formula is:

[0048] E i =Y i *M i

[0049]

[0050] Where Y i is the predicted electric load density of each area; M i Land use planning data for each district; i is the predicted value of electricity load in each district; i is the specific land type divided in the study area; n is the specific number of land divided in the study area.

[0051] The present invention has the following advantages: dividing the study area into several set areas according to the land use type; collecting data on the set areas to obtain parameter data of the set areas; calculating and obtaining the electric load density of the set areas based on the obtained parameter data of the set areas; using XGBoost to perform feature screening on the factors affecting the electric load density of the set areas to obtain the most influential characteristic factors on the electric load density of the set areas; performing data enhancement on the real data of the electric load density and characteristic variables of the set areas through a conditional generative adversarial network to generate the electric load density and characteristic variable data of the set areas with the same distribution; The real data of the electric load density and characteristic variables of the region are combined with the electric load density and characteristic variable data of the set region with the same distribution generated by enhancement to obtain a new data set; the electric load density time series prediction model is constructed according to the improved autoregressive distributed lag algorithm; the future electric load density of the set region is predicted by solving the contribution of the time position characteristic variables of the electric load density time series prediction model; the electric load prediction value of the set region is calculated according to the future electric load density of the set region and the parameter data of the set region; the electric load prediction values ​​of several set regions are summarized and summed to obtain the total electric load prediction value of the study area. The present invention takes into account the future land use planning information of the region. Incorporating this information into the prediction model can not only improve the accuracy of the electric load prediction, but also predict the electric load change trend of the regions with different land use types, thereby more effectively supporting the power and energy development planning. The present invention also takes into account a variety of variables that affect the electric load density, which can ensure that the prediction results will not be biased due to ignoring certain important variables, and at the same time increases the mining of the measured electric load density historical data, which can identify the trend of electric load density changes and support future trend prediction. The present invention uses a data enhancement model to increase the amount of data, helping the model to better capture the changing patterns and trends in the data, thereby improving the model's prediction accuracy and enhancing the model's performance. The present invention takes into account the influence of each variable's time series data. The closer the variable data is to the predicted year, the greater the weight, and the farther away, the smaller the weight. Increasing the weight of recent data helps reduce the deviation caused by historical data, and can ensure that the model uses the latest data information for prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0053] The structures, proportions, sizes, etc. illustrated in this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with the technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantial technical significance. Any structural modification, change in proportion or adjustment of size shall still fall within the scope of the technical contents disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention.

[0054] Figure 1 A schematic flow chart of a method for regional electric load prediction based on load density and data time series provided in Example 1 of the present invention;

[0055] Figure 2 A schematic diagram of a conditional generative adversarial network structure in a method for regional electric load prediction based on load density and data time series provided in Example 1 of the present invention;

[0056] Figure 3 This is a schematic diagram of the architecture of a regional electric load prediction device based on load density and data timing provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0057] The following is a description of the implementation of the present invention by specific embodiments. People familiar with the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are 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.

[0058] Example 1

[0059] See also Figure 1 Embodiment 1 of the present invention provides a method for regional electric load prediction based on load density and data time series, comprising the following steps:

[0060] S1. Divide the study area into several set areas according to the land use type; collect data for the set areas to obtain parameter data for the set areas;

[0061] S2. Calculate the electric load density of the set area according to the obtained parameter data of the set area; use XGBoost to perform feature screening on factors affecting the electric load density of the set area to obtain the most influential feature factors on the electric load density of the set area;

[0062] S3, performing data enhancement on the real data of the electric load density and characteristic variables of the set area through a conditional generative adversarial network to generate the electric load density and characteristic variable data of the set area with the same distribution;

[0063] S4, combining the real data of the electric load density and characteristic variables of the set area with the enhanced and generated data of the electric load density and characteristic variables of the set area with the same distribution to obtain a new data set;

[0064] S5. constructing an electric load density time series prediction model according to an improved autoregressive distributed lag algorithm; and predicting the future electric load density of the set area by solving the contribution of the time position characteristic variables to the electric load density time series prediction model;

[0065] S6. Calculate and obtain a predicted electric load value for the set area based on the future electric load density of the set area and the parameter data of the set area; and aggregate and sum the predicted electric load values ​​of several set areas to obtain a predicted total electric load value for the study area.

[0066] In this embodiment, in step S1, the study area is divided into a number of set areas according to the land use type; data is collected for the set areas to obtain parameter data of the set areas;

[0067] Specifically, there is a big difference in electricity load among different land use types, so the region is divided into agricultural land, industrial land, residential land, commercial land, hospital land, etc. according to different land use types; data collection is carried out in each district, including the electricity load of each district, the area of ​​each district, the factors affecting the electricity load of each district (population, economy, temperature, precipitation, etc.) and the future land use planning data of the region.

[0068] In this embodiment, in step S2, the electric load density of the set area is calculated based on the obtained parameter data of the set area; XGBoost is used to perform feature screening on factors affecting the electric load density of the set area to obtain the most influential feature factors on the electric load density of the set area;

[0069] Specifically, the electric load density of each area is calculated based on the electric load and area of ​​each area. There are many factors that affect the electric load density. When considering the influence of multiple factors, we can avoid the disaster of data dimension, improve the performance of the model, and use XGBoost to realize the screening of important features of the electric load density. XGBoost is a machine learning algorithm based on a tree model. It combines multiple weak learners (usually decision trees) to build a powerful model that can effectively process high-dimensional sparse data and automatically select features. The goal of training XGBoost is to minimize the objective function. The expression of the objective function is as follows:

[0070] Objective function = loss function + regularization term

[0071] By using XGBoost to select important features that affect the electric load density Y (for example, population A, economy B, temperature C), we can get the contribution of each important feature to the electric load density (weight parameters α, β, γ). The electric load density Y model is as follows:

[0072] Y t =αA t +βB t +γC t +ε t

[0073] Where t represents the data year; ε t is white noise.

[0074] In this embodiment, in step S3, the real data of the electric load density and characteristic variables of the set area are enhanced by a conditional generative adversarial network to generate the electric load density and characteristic variable data of the set area with the same distribution;

[0075] Specifically, a small amount of data may lead to inaccurate model evaluation because there is not enough data to divide the training set and the test set. To this end, a conditional generative adversarial network (CGAN) is used to build a data augmentation model for feature variables A, B, and C to obtain augmented variables

[0076] The conditional generative adversarial network (CGAN) consists of a generator (G) and a discriminator (D). The task of the generator is to generate new data samples based on given conditional labels and random noise. The task of the discriminator is to determine whether the input data is a real sample and give the corresponding probability value. During the training process, the generator and the discriminator will be trained adversarially, making it difficult for the discriminator to distinguish between the generated fake samples and the real samples. The generator uses the A, B, C conditional labels (y) and random noise (z) to generate (G(z|y)), and the discriminator uses the conditional label to judge the generated Whether it is similar to the real A, B, C data, and give the judgment result (D(x|y)). The objective function of the conditional generative adversarial network (CGAN) is:

[0077]

[0078] Where D(x|y) is the probability of classifying x as real data rather than generated data under conditional label y; is the probability of classifying real data as real data; D(G(z|y)) is the probability of classifying generated data as real data; 1-D(G(z|y)) is the probability of classifying generated data as generated data; is the probability of classifying the generated data as generated data.

[0079] The objective function of the conditional generative adversarial network (CGAN) is the ability to classify real data into real data and generated data into generated data. For the discriminator, the higher this ability is, the better the discriminator effect is, so the parameter optimization process of the discriminator is a process of maximizing the objective function; after optimizing the discriminator, it is necessary to fix the parameters of the discriminator and then optimize the parameters of the generator. When the discriminator parameters are fixed, the first term of the objective function becomes a constant, and only the second term changes when adjusting the generator parameters. The second term represents the ability to classify generated data into generated data. For the generator, the smaller the ability is, the better, so that the generator can deceive the discriminator, so the objective function of the conditional generative adversarial network is the minimax game between the generator G and the discriminator D. In the training loop, the parameters of the generator and the discriminator are updated alternately. In each iteration, the generator is first fixed, and the discriminator is trained to distinguish between real A, B, C and generated data; then fix the discriminator and train the generator to generate data that can deceive the discriminator Data. The specific structure of the conditional generation adversarial network is as follows Figure 2 shown.

[0080] Through the generator and discriminator of the conditional generative adversarial network (CGAN), it is possible to generate data similar to the real data A, B, and C. data.

[0081]

[0082] Based on enhanced variables Obtaining enhanced target variables through feature analysis models

[0083] In this embodiment, in step S4, the real data of the electric load density and characteristic variables of the set area are combined with the enhanced and generated electric load density and characteristic variable data of the set area with the same distribution to obtain a new data set;

[0084] Specifically, the collected electric load density Y of each district, the real data of characteristic variables A, B, C and the generated characteristic variables of each district are compared. Electric load density Combine to obtain a new data set.

[0085] In this embodiment, in step S5, an electric load density time series prediction model is constructed according to an improved autoregressive distributed lag algorithm; the future electric load density of the set area is predicted by solving the contribution of the time position characteristic variables to the electric load density time series prediction model;

[0086] Specifically, the improved autoregressive distributed lag algorithm (ARDL) is used to establish the electric load density time series prediction model. Based on the characteristic analysis model parameters α, β, and γ, the data A, B, C, Y and enhanced data Further solve the data contribution of each time position (weight parameter w i , σ i , i ,φ i ), automatically learning the temporal relationship between features.

[0087] The expression of the future electric load density of the set area is:

[0088]

[0089] In the formula, w i , σ i , i , are all weight parameters; p is the autoregressive order of Y; q, r, s are the lag orders of A, B, and C respectively.

[0090] The electric load density time series prediction model based on the improved autoregressive distributed lag algorithm (ARDL) is used to predict the electric load density in each area.

[0091] In this embodiment, in step S6, the electric load forecast value of the set area is calculated based on the future electric load density of the set area and the parameter data of the set area; the electric load forecast values ​​of several set areas are summarized and summed to obtain the total electric load forecast value of the study area.

[0092] Specifically, based on the predicted electric load density Y of each area i , combined with the land use planning data of each district M i , calculate the predicted value of electric load E in each area i , add up the predicted values ​​of electric load in each area to get the predicted value of total electric load in the study area T e :

[0093] E i =Y i *M i

[0094]

[0095] Where i is the specific land use type (e.g., agricultural land) into which the study area is divided; n is the specific number of land uses into which the study area is divided.

[0096] In summary, the present invention divides the study area into several set areas according to the land use type; collects data for the set areas to obtain parameter data for the set areas; calculates and obtains the electric load density of the set areas based on the obtained parameter data for the set areas; uses XGBoost to perform feature screening on the factors affecting the electric load density of the set areas to obtain the most influential feature factors on the electric load density of the set areas; performs data enhancement on the real data of the electric load density and feature variables of the set areas through a conditional generative adversarial network to generate the electric load density and feature variable data of the set areas with the same distribution; The real data of the electric load density and characteristic variables of the domain are combined with the electric load density and characteristic variable data of the set area with the same distribution generated by enhancement to obtain a new data set; the electric load density time series prediction model is constructed according to the improved autoregressive distributed lag algorithm; the future electric load density of the set area is predicted by solving the contribution of the time position characteristic variables of the electric load density time series prediction model; the electric load prediction value of the set area is calculated according to the future electric load density of the set area and the parameter data of the set area; the electric load prediction values ​​of several set areas are summarized and summed to obtain the total electric load prediction value of the study area. The present invention takes into account the future land use planning information of the region. Incorporating this information into the prediction model can not only improve the accuracy of the electric load prediction, but also predict the electric load change trend of the areas with different land use types, thereby more effectively supporting the power and energy development planning. The present invention also takes into account a variety of variables that affect the electric load density, which can ensure that the prediction results will not be biased due to ignoring certain important variables, and at the same time increases the mining of the measured electric load density historical data, which can identify the trend of electric load density changes and support future trend prediction. The present invention uses a data enhancement model to increase the amount of data, helping the model to better capture the changing patterns and trends in the data, thereby improving the model's prediction accuracy and enhancing the model's performance. The present invention takes into account the influence of each variable's time series data. The closer the variable data is to the predicted year, the greater the weight, and the farther away, the smaller the weight. Increasing the weight of recent data helps reduce the deviation caused by historical data, and can ensure that the model uses the latest data information for prediction.

[0097] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will interact with each other to complete the described method.

[0098] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0099] Example 2

[0100] See also Figure 3 Embodiment 2 of the present invention also provides a regional electric load prediction device based on load density and data time series, including:

[0101] The area division and parameter data acquisition module 001 is used to divide the study area into a number of set areas according to the land use type; collect data for the set areas to obtain parameter data for the set areas;

[0102] The characteristic factor screening module 002 is used to calculate the electric load density of the set area according to the obtained parameter data of the set area; use XGBoost to perform characteristic screening on the factors affecting the electric load density of the set area to obtain the characteristic factors that have the greatest influence on the electric load density of the set area;

[0103] The data enhancement module 003 is used to enhance the real data of the electric load density and characteristic variables of the set area through a conditional generative adversarial network to generate the electric load density and characteristic variable data of the set area with the same distribution;

[0104] A new data set acquisition module 004 is used to combine the real data of the electric load density and characteristic variables of the set area with the enhanced and generated electric load density and characteristic variable data of the set area with the same distribution to obtain a new data set;

[0105] The future electric load density prediction module 005 is used to construct an electric load density time series prediction model according to the improved autoregressive distributed lag algorithm; by solving the contribution of the time position characteristic variables of the electric load density time series prediction model, the future electric load density of the set area is predicted;

[0106] The total electric load forecast value acquisition module 006 is used to calculate the electric load forecast value of the set area according to the future electric load density of the set area and the parameter data of the set area; and to summarize and sum the electric load forecast values ​​of several set areas to obtain the total electric load forecast value of the study area.

[0107] In this embodiment, in the region division and parameter data acquisition module 001, in the process of dividing the study area into a plurality of set areas according to land use types, the set areas include: agricultural land, industrial land, residential land, commercial land and hospital land;

[0108] In the process of collecting data on the set area and obtaining parameter data of the set area, the parameter data of the set area includes: the electric load, area, factors affecting the electric load and future land use planning data of the set area.

[0109] In this embodiment, in the characteristic factor screening module 002, in the process of using XGBoost to perform characteristic screening on the factors affecting the electric load density of the set area, the important features affecting the electric load density are selected by XGBoost, and the contribution of each important feature to the electric load density can be obtained; the electric load density model is:

[0110] Y t =αA t +βB t +γC t +ε t

[0111] Where Y t is the electric load density; α, β, γ are all weight parameters; A t is the population characteristic; B t Economic characteristics; C t is the temperature characteristic; t is the data year; ε t is white noise.

[0112] In this embodiment, in the data enhancement module 003, in the process of performing data enhancement on the real data of the electric load density and characteristic variables of the set area through the conditional generative adversarial network, the objective function expression of the conditional generative adversarial network is:

[0113]

[0114] Where D(x|y) is the probability of classifying x as real data rather than generated data under conditional label y; is the probability of classifying real data as real data; D(G(z|y)) is the probability of classifying generated data as real data; 1-D(G(z|y)) is the probability of classifying generated data as generated data; is the probability of classifying the generated data as generated data.

[0115] In this embodiment, in the future electric load density prediction module 005, in the process of predicting the future electric load density of the set area by solving the contribution of the time position characteristic variable to the electric load density time series prediction model, the expression of the future electric load density of the set area is:

[0116]

[0117] In the formula, w i , σ i , i , are all weight parameters; p is the autoregressive order of Y; q, r, s are the lag orders of A, B, and C respectively;

[0118] In the total electric load forecast value acquisition module 006, the total electric load forecast value T of the study area is obtained. e The formula is:

[0119] E i =Y i *M i

[0120]

[0121] Where Y i is the predicted electric load density of each area; M i Land use planning data for each district; i is the predicted value of electricity load in each district; i is the specific land type divided in the study area; n is the specific number of land divided in the study area.

[0122] It should be noted that the information interaction, execution process and other contents between the modules of the above-mentioned system are based on the same concept as the method embodiment in Example 1 of the present application, and the technical effects they bring are the same as those of the method embodiment of the present application. For specific contents, please refer to the description in the method embodiment shown above in the present application, and will not be repeated here.

[0123] Example 3

[0124] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium, in which a program code of a regional electric load prediction method based on load density and data timing is stored, and the program code includes instructions for executing the regional electric load prediction method based on load density and data timing of embodiment 1 or any possible implementation thereof.

[0125] The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0126] Example 4

[0127] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor;

[0128] The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the regional electric load prediction method based on load density and data timing of Example 1 or any possible implementation thereof.

[0129] Specifically, the processor can be implemented by hardware or by software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor implemented by reading software codes stored in a memory. The memory can be integrated in the processor or can be located outside the processor and exist independently.

[0130] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems. The computer instructions can be stored in a computer-readable storage medium, or transmitted from a computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center.

[0131] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing system, they can be concentrated on a single computing system, or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by a program code executable by a computing system, so that they can be stored in a storage system and executed by the computing system, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0132] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A regional electric load forecasting method based on load density and data time series, characterized in that: include: The study area is divided into several set areas according to land use types; Collecting data for the set area to obtain parameter data for the set area; According to the obtained parameter data of the set area, the electric load density of the set area is calculated; XGBoost is used to perform feature screening on factors affecting the electric load density of the set area to obtain the most influential feature factors on the electric load density of the set area; Performing data enhancement on the real data of the electric load density and characteristic variables of the set area through a conditional generative adversarial network to generate the electric load density and characteristic variable data of the set area with the same distribution; Combining the real data of the electric load density and characteristic variables of the set area with the enhanced and generated electric load density and characteristic variable data of the set area with the same distribution to obtain a new data set; A time series prediction model for electric load density is constructed according to an improved autoregressive distributed lag algorithm; the future electric load density of the set area is predicted by solving the contribution of time and position characteristic variables to the time series prediction model for electric load density; Calculating and obtaining a predicted value of the electric load of the set area according to the future electric load density of the set area and the parameter data of the set area; The predicted electric load values ​​of several set areas are aggregated and summed to obtain the predicted total electric load value of the study area.

2. The method for regional electric load forecasting based on load density and data time series according to claim 1, characterized in that: In the process of dividing the study area into a plurality of the set areas according to land use types, the set areas include: agricultural land, industrial land, residential land, commercial land and hospital land.

3. The method for regional electric load forecasting based on load density and data time series according to claim 2 is characterized in that: In the process of collecting data on the set area and obtaining parameter data of the set area, the parameter data of the set area includes: the electric load, area, factors affecting the electric load and future land use planning data of the set area.

4. The method for regional electric load forecasting based on load density and data time series according to claim 3 is characterized in that: In the process of using XGBoost to perform feature screening on the factors affecting the electric load density of the set area, the important features affecting the electric load density are selected by XGBoost, and the contribution of each important feature to the electric load density can be obtained; the electric load density model is: Y t =αA t +βB t +γC t +e t Where Y t is the electric load density; α, β, γ are all weight parameters; A t is the population characteristic; B t Economic characteristics; C t is the temperature characteristic; t is the data year; ε t is white noise.

5. The method for regional electric load forecasting based on load density and data time series according to claim 4 is characterized in that: In the process of performing data enhancement on the real data of the electric load density and characteristic variables of the set area through the conditional generative adversarial network, the objective function expression of the conditional generative adversarial network is: Where D(x|y) is the probability of classifying x as real data rather than generated data under conditional label y; is the probability of classifying real data as real data; D(G(z|y)) is the probability of classifying generated data as real data; 1-D(G(z|y)) is the probability of classifying generated data as generated data; is the probability of classifying the generated data into the generated data.

6. The method for regional electric load forecasting based on load density and data time series according to claim 5, characterized in that: In the process of predicting the future electric load density of the set area by solving the contribution of the time position characteristic variable to the electric load density time series prediction model, the expression of the future electric load density of the set area is: In the formula, w i , σ i , i , are all weight parameters; p is the autoregressive order of Y; q, r, s are the lag orders of A, B, and C respectively; Obtain the total power load forecast value T of the study area e The formula is: E i =Y i *M i Where Y i is the predicted electric load density in each area; M i Land use planning data for each district; i is the predicted value of electricity load in each district; i is the specific land type divided in the study area; n is the specific number of land divided in the study area.

7. A regional electric load prediction device based on load density and data time series, adopting the regional electric load prediction method based on load density and data time series according to any one of claims 1 to 6, characterized in that: include: The regional division and parameter data acquisition module is used to divide the study area into several set areas according to the land use type; Collecting data for the set area to obtain parameter data for the set area; A characteristic factor screening module is used to calculate the electric load density of the set area according to the obtained parameter data of the set area; use XGBoost to perform characteristic screening on the factors affecting the electric load density of the set area to obtain the characteristic factors that have the greatest influence on the electric load density of the set area; A data enhancement module, used to perform data enhancement on the real data of the electric load density and characteristic variables of the set area through a conditional generative adversarial network, and generate the electric load density and characteristic variable data of the set area with the same distribution; A new data set acquisition module, used for combining the real data of the electric load density and characteristic variables of the set area with the enhanced and generated data of the electric load density and characteristic variables of the set area with the same distribution, to obtain a new data set; A future electric load density prediction module is used to construct an electric load density time series prediction model according to an improved autoregressive distributed lag algorithm; by solving the contribution of time and position characteristic variables to the electric load density time series prediction model, the future electric load density of the set area is predicted; A total electric load forecast value acquisition module, used to calculate and obtain the electric load forecast value of the set area according to the future electric load density of the set area and the parameter data of the set area; The predicted electric load values ​​of several set areas are aggregated and summed to obtain the predicted total electric load value of the study area.

8. The regional electric load prediction device based on load density and data time series according to claim 7 is characterized in that: In the characteristic factor screening module, in the process of using XGBoost to perform characteristic screening on the factors affecting the electric load density of the set area, the important features affecting the electric load density are selected by XGBoost, and the contribution of each important feature to the electric load density can be obtained; the electric load density model is: Y t =αA t +βB t +γC t +e t Where Y t is the electric load density; α, β, γ are all weight parameters; A t is the population characteristic; B t Economic characteristics; C t is the temperature characteristic; t is the data year; ε t is white noise.

9. The regional electric load prediction device based on load density and data time series according to claim 8, characterized in that: In the data enhancement module, in the process of performing data enhancement on the real data of the electric load density and characteristic variables of the set area through the conditional generative adversarial network, the objective function expression of the conditional generative adversarial network is: Where D(x|y) is the probability of classifying x as real data rather than generated data under conditional label y; is the probability of classifying real data as real data; D(G(z|y)) is the probability of classifying generated data as real data; 1-D(G(z|y)) is the probability of classifying generated data as generated data; is the probability of classifying the generated data into the generated data.

10. The regional electric load prediction device based on load density and data time series according to claim 9, characterized in that: In the future electric load density prediction module, in the process of predicting the future electric load density of the set area by solving the contribution of the time position characteristic variable of the electric load density time series prediction model, the expression of the future electric load density of the set area is: In the formula, w i , σ i , i , are all weight parameters; p is the autoregressive order of Y; q, r, s are the lag orders of A, B, and C respectively; In the total electric load forecast value acquisition module, the total electric load forecast value T of the study area is obtained. e The formula is: E i =Y i *M i Where Y i is the predicted electric load density in each area; M i Land use planning data for each district; i is the predicted value of electricity load in each district; i is the specific land type divided in the study area; n is the specific number of land divided in the study area.

Images