Load potential correlation factor identification method, system and device and storage medium

Abstract

The invention relates to the field of data analysis, in particular to a load potential correlation factor identification method, system and device and a storage medium. The method comprises the following steps: acquiring industrial daily load data in a historical time period and data of a plurality of potential correlation factors corresponding to the industrial daily load data, and preprocessing; performing kernel density estimation on the preprocessed industrial daily load data and the preprocessed data of each potential correlation factor in the historical time period by adopting an index kernel function, and then performing integration on the data to obtain a marginal probability distribution function of the industrial daily load and a marginal probability distribution function of each potential correlation factor; coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor and obtain parameters of the joint probability distribution function, and solving correlation coefficients of the industrial daily load and the potential correlation factors. According to the invention, the relation between the industrial daily load and the plurality of potential correlation factors can be accurately quantified.

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and in particular to a method, system, device and storage medium for identifying potential load correlation factors. Background Art

[0002] At present, my country's industrial energy consumption accounts for about 65% of the total social energy consumption. The industrial load capacity base is huge, and there are a large number of electric loads with transferable characteristics, which have good power regulation potential. During peak load periods, power grid companies will encourage industrial enterprises to reduce electricity consumption during specific periods through demand response mechanisms. This requires industrial enterprises to have a certain degree of flexibility so that they can adjust production plans when necessary, thereby affecting the demand for active power.

[0003] The power grid company obtains the electricity consumption data of various industrial enterprises, establishes a load forecasting model, and uses time series models such as ARIMA and exponential smoothing to predict the electricity demand under different production conditions, thereby identifying the adjustable resources and potential load response capabilities of industrial enterprises, and reasonably dispatching various power generation resources and adjusting power generation plans based on real-time load and forecast data. By analyzing the electricity consumption data of enterprises, the power grid company can identify which industrial enterprises have adjustable resources.

[0004] However, the current load forecasting model mainly focuses on electrical quantity data factors, lacking in-depth exploration of the relationship between potential correlation factors and industrial loads. Therefore, it limits the comprehensive evaluation of the active power regulation capability of industrial enterprises by power grid companies, affects the prediction accuracy of load forecasting models, and causes errors in resource allocation of power grid companies in active power regulation, affecting the efficiency and reliability of the overall power system. The lack of in-depth exploration of the relationship between potential correlation factors and industrial loads is mainly because the relationship between these potential factors is complex and dynamic. Various potential factors often do not exist in isolation, but influence each other and are intertwined. In addition, the relationship between potential factors and industrial loads is usually not linear. The change in load may not be proportional to the change in a single factor, but is affected by the combined effect of multiple factors. Over time, the influence and correlation patterns of potential factors may also change. These problems make it extremely challenging to quantify their relationship with industrial loads. Summary of the invention

[0005] To this end, the technical problem to be solved by the present invention is to overcome the defect that the relationship between the existing potential correlation factors and the industrial load is difficult to quantify, resulting in poor prediction accuracy of the current load forecasting model.

[0006] In order to solve the above technical problems, the present invention provides a method for identifying potential correlation factors of typical industrial loads, comprising the following steps:

[0007] Obtain the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, and perform preprocessing;

[0008] After using the exponential kernel function to estimate the kernel density of the industrial daily load data in the preprocessed historical time period, the marginal probability distribution function of the industrial daily load is obtained by integrating it.

[0009] After using the exponential kernel function to estimate the kernel density of each potential correlation factor after preprocessing, the marginal probability distribution function of each potential correlation factor is obtained by integrating the kernel density.

[0010] The marginal probability distribution function of industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of industrial daily load and each potential correlation factor.

[0011] The parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are obtained, and the correlation coefficient between the industrial daily load and each potential correlation factor is solved based on the joint probability distribution function between the industrial daily load and each potential correlation factor.

[0012] Preferably, the potential correlation factors include: at least one of meteorological data, calendar factors, random interference factors and production data of industrial enterprises;

[0013] Among them, meteorological data include: at least one of the highest temperature, the lowest temperature, the air quality index and precipitation; calendar factors include: holidays; random interference factors include: at least one of new orders added on the day and policy intervention; production data of industrial enterprises include: at least one of generated output, raw material consumption and production process switching.

[0014] Preferably, the industrial daily load data within the historical time period and the corresponding data of multiple potential correlation factors are preprocessed, including:

[0015] When the industrial daily load data in the historical time period is missing, the missing data is processed. For the missing industrial daily load data of a single date, the average industrial daily load of the two dates before and after the missing data date is taken as the industrial daily load data of the date of the missing data;

[0016] For missing industrial daily load data for multiple consecutive dates, taking each date of missing data as the center, within the preset date range, select the industrial daily load data whose similarity with the industrial daily load curve of the date of missing data reaches a preset threshold or above, as the industrial daily load data of the date of missing data;

[0017] After processing the missing data, if there are still missing data, the dates containing the missing data will be directly eliminated;

[0018] The data of industrial daily load and its corresponding potential correlation factors in the historical time period after missing data processing are normalized, and the Kalman filtering algorithm is used to smooth the industrial daily load data in the historical time period after normalization to obtain the preprocessed industrial daily load data in the historical time period and its corresponding multiple potential correlation factors.

[0019] Preferably, an exponential kernel function is used to perform kernel density estimation on the preprocessed industrial daily load data in the historical time period, and the formula is:

[0020]

[0021] Among them, F(X) is the kernel density estimation of industrial daily load, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, X is the target position of the kernel density estimation, and X i is the i-th industrial daily load data after preprocessing, σ is the standard deviation of the industrial daily load data,

[0022] Preferably, an exponential kernel function is used to perform kernel density estimation on the preprocessed data of each potential correlation factor, and the formula is:

[0023]

[0024] Among them, F(x l ) is the kernel density estimate of the lth potential correlation factor, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, x l is the target position of the kernel density estimation of the lth potential correlation factor, is the i-th data of the l-th potential correlation factor after preprocessing, σ l is the standard deviation of the lth potential correlation factor,

[0025] Preferably, the Copula function is any one or more of a Gumbel Copula function, a Clayton Copula function, a Frank Copula function, and a Gaussian Copula.

[0026] Preferably, the marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Gumbel Copula function, as the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Gumbel Copula function, the formula is:

[0027] Y1(μ(X),μ(x l ))=C1(μ(X),μ(x l ))

[0028]

[0029] Among them, Y1(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, C1(.) is the Gumbel Copula function, exp{.} is the exponential function with the natural constant as the base, θ1 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0030] Preferably, the marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Clayton Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Clayton Copula function, and the formula is:

[0031] Y2(μ(X),μ(x l ))=C2(μ(X),μ(x l ))

[0032]

[0033] Among them, Y2(μ(X),μ(x l)) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, C2(.) is the Clayton Copula function, θ2 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0034] Preferably, the step of obtaining the parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor includes:

[0035] Based on the pre-processed industrial daily load data in the historical time period and the corresponding data of multiple potential correlation factors, the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor is calculated;

[0036] The calculation formula is:

[0037]

[0038] Among them, τ l is the Kendall rank correlation coefficient between industrial daily load and the lth potential correlation factor, is the number of combinations of selecting 2 data from n data, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, sign{.} is the sign function, X i is the preprocessed daily load data of the ith industry, X j is the j-th industrial day load data after preprocessing, is the i-th data of the l-th potential correlation factor after preprocessing, is the jth data of the lth potential correlation factor after preprocessing;

[0039] According to the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor, the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are calculated respectively.

[0040] The present invention also provides a load potential correlation factor identification system, comprising:

[0041] The data processing module is used to obtain the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, and perform preprocessing;

[0042] The industrial daily load distribution estimation module is used to use the exponential kernel function to perform kernel density estimation on the industrial daily load data in the preprocessed historical time period, and then integrate the kernel density to obtain the marginal probability distribution function of the industrial daily load;

[0043] A potential correlation factor distribution estimation module is used to use an exponential kernel function to perform kernel density estimation on the preprocessed data of each potential correlation factor, and then integrate the kernel density to obtain the marginal probability distribution function of each potential correlation factor;

[0044] The joint probability distribution acquisition module couples the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor;

[0045] The correlation coefficient solving module is used to obtain the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor, and solve the correlation coefficient between the industrial daily load and each potential correlation factor based on the joint probability distribution function between the industrial daily load and each potential correlation factor.

[0046] Preferably, the potential correlation factors include: at least one of meteorological data, calendar factors, random interference factors and production data of industrial enterprises;

[0047] Among them, meteorological data include: at least one of the highest temperature, the lowest temperature, the air quality index and precipitation; calendar factors include: holidays; random interference factors include: at least one of new orders added on the day and policy intervention; production data of industrial enterprises include: at least one of generated output, raw material consumption and production process switching.

[0048] Preferably, the industrial daily load data within the historical time period and the corresponding data of multiple potential correlation factors are preprocessed, including:

[0049] When the industrial daily load data in the historical time period is missing, the missing data is processed. For the missing industrial daily load data of a single date, the average industrial daily load of the two dates before and after the missing data date is taken as the industrial daily load data of the date of the missing data;

[0050] For missing industrial daily load data for multiple consecutive dates, taking each date of missing data as the center, within the preset date range, select the industrial daily load data whose similarity with the industrial daily load curve of the date of missing data reaches a preset threshold or above, as the industrial daily load data of the date of missing data;

[0051] After processing the missing data, if there are still missing data, the dates containing the missing data will be directly eliminated;

[0052] The data of industrial daily load and its corresponding potential correlation factors in the historical time period after missing data processing are normalized, and the Kalman filtering algorithm is used to smooth the industrial daily load data in the historical time period after normalization to obtain the preprocessed industrial daily load data in the historical time period and its corresponding multiple potential correlation factors.

[0053] Preferably, an exponential kernel function is used to perform kernel density estimation on the preprocessed industrial daily load data in the historical time period, and the formula is:

[0054]

[0055] Among them, F(X) is the kernel density estimation of industrial daily load, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, X is the target position of the kernel density estimation, and X i is the i-th industrial daily load data after preprocessing, σ is the standard deviation of the industrial daily load data,

[0056] Preferably, an exponential kernel function is used to perform kernel density estimation on the preprocessed data of each potential correlation factor, and the formula is:

[0057]

[0058] Among them, F(x l ) is the kernel density estimate of the lth potential correlation factor, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, x l is the target position of the kernel density estimation of the lth potential correlation factor, is the i-th data of the l-th potential correlation factor after preprocessing, σ l is the standard deviation of the lth potential correlation factor,

[0059] Preferably, the Copula function is any one or more of a Gumbel Copula function, a Clayton Copula function, a Frank Copula function, and a Gaussian Copula.

[0060] Preferably, the marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Gumbel Copula function, as the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Gumbel Copula function, the formula is:

[0061] Y1(μ(X),μ(x l ))=C1(μ(X),μ(x l ))

[0062]

[0063] Among them, Y1(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, C1(.) is the Gumbel Copula function, exp{.} is the exponential function with the natural constant as the base, θ1 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0064] Preferably, the marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Clayton Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Clayton Copula function, and the formula is:

[0065] Y2(μ(X),μ(x l ))=C2(μ(X),μ(x l ))

[0066]

[0067] Among them, Y2(μ(X),μ(x l)) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, C2(.) is the Clayton Copula function, θ2 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0068] Preferably, the step of obtaining the parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor includes:

[0069] Based on the pre-processed industrial daily load data in the historical time period and the corresponding data of multiple potential correlation factors, the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor is calculated;

[0070] The calculation formula is:

[0071]

[0072] Among them, τ l is the Kendall rank correlation coefficient between industrial daily load and the lth potential correlation factor, is the number of combinations of selecting 2 data from n data, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, sign{.} is the sign function, X i is the preprocessed daily load data of the ith industry, X j is the j-th industrial day load data after preprocessing, is the i-th data of the l-th potential correlation factor after preprocessing, is the jth data of the lth potential correlation factor after preprocessing;

[0073] According to the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor, the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are calculated respectively.

[0074] The present invention also provides a load potential correlation factor identification device, comprising:

[0075] The memory is used to store computer programs; the processor is used to implement the steps of the above-mentioned method for identifying potential load correlation factors when executing the computer program.

[0076] The present invention also 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 the above-mentioned method for identifying potential load correlation factors are implemented.

[0077] The above technical solution of the present invention has the following beneficial effects compared with the prior art:

[0078] The present invention discloses a method, system, device and storage medium for identifying potential load correlation factors. Since industrial load data often presents a skewed distribution, the load value usually maintains a low level for most of the time, but rises rapidly in a specific time period (such as peak production period, when energy consumption increases sharply), the present invention uses an exponential kernel function to perform kernel density estimation on the industrial daily load data in the pre-processed historical time period and the data of each potential correlation factor. The exponential kernel function has the characteristic of rapid decay, which enables it to well simulate the concentration of load data in the high-load time period, and quickly decay to the low-load area. In addition, the exponential kernel function is relatively simple to calculate and is suitable for processing large-scale data. The Copula function is used to combine the industrial daily load and its multiple potential correlation factors, which can effectively capture the complex nonlinear relationship between variables, overcome the dependence between complex potential correlation factors and industrial load, avoid the limitations of traditional linear models, and provide more accurate quantification of the relationship between industrial daily load and multiple potential correlation factors, thereby providing strong data support for decision-making in industrial scheduling and related fields, improving the ability to predict electricity demand under different production conditions, and enabling power grid companies to accurately identify the adjustable resources and potential load response capabilities of industrial enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein:

[0080] Figure 1 It is a flow chart of the steps of a method for identifying potential correlation factors of loads of the present invention.

[0081] Figure 2 It is a schematic diagram of the enterprise electricity consumption curve in Example 2.

[0082] Figure 3 It is a schematic diagram of the enterprise power curve in the second embodiment.

[0083] Figure 4 is a schematic diagram of the nonlinear correlation between the industrial load of an enterprise and potential correlation factors in Example 2, Figure 4 (a) in the figure represents a schematic diagram of function parameters of the Gumbel Copula of the industrial load of an enterprise and potential correlation factors in the second embodiment. Figure 4 (b) in the figure is a schematic diagram of the function parameters of the Clyton Copula of the industrial load of the enterprise and the potential correlation factors in the second embodiment. DETAILED DESCRIPTION

[0084] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0085] Reference Figure 1 As shown, the first embodiment of the present invention provides a method for identifying potential load correlation factors, including the following steps:

[0086] Step S1: Obtain the industrial daily load data within the historical time period and the data of multiple potential correlation factors corresponding thereto, and perform preprocessing to obtain the industrial daily load data X within the historical time period after preprocessing. i , i = 1, 2, ... n and its corresponding data of multiple potential correlation factors and potential correlation data l=1,2,…m, where is the ith data of the lth potential correlation factor, m is the total number of potential correlation factors, l is the potential correlation factor index, and n is the total number of data of industrial daily load or the total number of data of each potential correlation factor.

[0087] Specifically, the industrial daily load is a time series data, and the maximum load, average load or other typical load within a day can be selected.

[0088] In this embodiment, preferably, the potential correlation factors include: at least one of meteorological data, calendar factors, random interference factors and production data of industrial enterprises;

[0089] The meteorological data includes: at least one of the highest temperature, the lowest temperature, the air quality index and the precipitation; the calendar factors include: holidays; the random interference factors include: at least one of the newly added orders and policy interventions on the day; the production data of industrial enterprises include: at least one of the generated output, the raw material consumption and the production process switching. In this embodiment, preferably, the industrial daily load data within the historical time period and the corresponding data of multiple potential correlation factors are preprocessed, including:

[0090] When the industrial daily load data in the historical time period is missing, the missing data is processed. For the missing industrial daily load data of a single date, the average industrial daily load of the two dates before and after the missing data date is taken as the industrial daily load data of the date of the missing data;

[0091] For missing industrial daily load data for multiple consecutive dates, taking each date of missing data as the center, within the preset date range, select the industrial daily load data whose similarity with the industrial daily load curve of the date of missing data reaches a preset threshold or above, as the industrial daily load data of the date of missing data;

[0092] After processing the missing data, if there are still missing data, the date containing the missing data will be directly eliminated; processing the missing industrial daily load data helps to maintain the continuity and integrity of the data and avoid inaccurate analysis results due to missing data. For the missing data of a single date, the average value of the industrial daily load of the two dates before and after the missing data date can be used to fill the missing data to smooth the load fluctuation and reduce the prediction error; and for the missing data of industrial daily load on multiple consecutive dates, the load trend can be restored more accurately by selecting the load curve similar to the missing date for matching and filling. This method not only ensures the integrity and rationality of the data, but also improves the stability and accuracy of the model training, avoids excessive data loss, and thus improves the overall prediction accuracy and data quality.

[0093] The data of industrial daily load and its corresponding potential correlation factors in the historical time period after missing data processing are normalized, and the Kalman filtering algorithm is used to smooth the industrial daily load data in the historical time period after normalization to obtain the preprocessed industrial daily load data in the historical time period and its corresponding multiple potential correlation factors.

[0094] The data of different dimensions are normalized to the same comparison level through normalization operation to eliminate the inconsistency of dimensions. The normalization formula is as follows:

[0095]

[0096] Among them, P ′ is the normalized value, P is the value to be normalized, P min is the minimum value in the data set where the current value is located, P max The maximum value in the data set where the current value is located.

[0097] The Kalman filter algorithm is used to smooth the industrial daily load data in the normalized historical time period. Specifically, the Kalman filter algorithm is used to smooth the CSV data using the pandas and numpy libraries in Python. The smoother filtered data is obtained by iterative calculation prediction and updated state estimation. Industrial daily load data is often affected by various random fluctuations, sensor errors or external environmental factors, resulting in noise in the data. Kalman filtering can effectively suppress these noises, making the load data smoother and reflecting the real trends and changes.

[0098] The present invention assumes that the industrial daily load and each potential correlation factor obey the P-III distribution.

[0099] Step S2: After performing kernel density estimation on the industrial daily load data in the preprocessed historical time period using the exponential kernel function, the marginal probability distribution function μ(X) of the industrial daily load is obtained by integrating the data;

[0100] The exponential kernel function is used to estimate the kernel density of the industrial daily load data in the preprocessed historical time period. The formula is:

[0101]

[0102] Among them, F(X) is the kernel density estimation of industrial daily load, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, μ is the bandwidth parameter, K{.} is the exponential kernel function, X is the target position of the kernel density estimation, and X i is the i-th industrial daily load data after preprocessing, σ is the standard deviation of the industrial daily load data,

[0103] Step S3: After using the exponential kernel function to perform kernel density estimation on the data of each potential correlation factor after preprocessing, integrate it to obtain the marginal probability distribution function μ(x l );

[0104] The exponential kernel function is used to estimate the kernel density of each potential correlation factor after preprocessing. The formula is:

[0105]

[0106] Among them, F(x l ) is the kernel density estimate of the lth potential correlation factor, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, x l is the target position of the kernel density estimation of the lth potential correlation factor, is the i-th data of the l-th potential correlation factor after preprocessing, σ l is the standard deviation of the lth potential correlation factor,

[0107] Step S4: respectively coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor;

[0108] The Copula function determines the conversion bridge from marginal distribution to joint distribution and from joint distribution to marginal distribution. The decomposition of the joint distribution function is the separation of randomness and coupling. The randomness is characterized by the corresponding marginal distribution, and the coupling is described by the Copula function. Different coupling methods correspond to different joint distributions, but this coupling does not change the corresponding relationship between the coupling body and the marginal distribution. The solution to the joint distribution function can be converted into the solution of the Copula function.

[0109] In this embodiment, specifically, the Copula function is any one or more of a Gumbel Copula function, a Clayton Copula function, a Frank Copula function, and a Gaussian Copula.

[0110] In this embodiment, preferably, the marginal probability distribution function of the industrial daily load and the marginal probability distribution function of each potential correlation factor are coupled through the Gumbel Copula function and the Clayton Copula function respectively, and the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the two Copula functions is obtained, and the formula is:

[0111] Y1(μ(X),μ(x l ))=C1(μ(X),μ(x l ))

[0112]

[0113] Among them, Y1(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, C1(.) is the Gumbel Copula function, exp{.} is the exponential function with the natural constant as the base, θ1 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0114] Y2(μ(X),μ(x l ))=C2(μ(X),μ(x l ))

[0115]

[0116] Among them, Y2(μ(X),μ(x l)) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, C2(.) is the Clayton Copula function, θ2 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0117] Using Gumbel Copula and Clayton Copula functions to analyze industrial daily load and its potential correlation factors can effectively capture the nonlinear and asymmetric dependencies between industrial daily load and potential correlation factors. Gumbel Copula is mainly used to describe positive tail dependencies and is suitable for analyzing the relationship between high-load periods and external factors (such as temperature, production peaks, etc.). For example, in the case of extreme fluctuations in industrial load, Gumbel Copula can accurately model the relationship between load demand and factors such as climate change, helping to optimize power scheduling and energy distribution. Clayton Copula is suitable for low-load period analysis, which can reveal the relationship between low-demand periods and factors such as equipment idleness and production stagnation, thereby providing more accurate demand forecasts for low-load periods.

[0118] The joint application of these two Copula functions can handle both high-tail and low-tail dependencies, providing a more comprehensive analysis. By combining Gumbel and Clayton Copula, the dependencies of industrial daily loads in different load ranges can be fully quantified. Gumbel Copula can help analyze extreme demand fluctuations during high-load periods, while Clayton Copula can accurately capture the changing characteristics of low-load periods. The joint use of the two makes it possible to fully consider high-tail and low-tail dependencies when analyzing industrial daily loads, thereby providing more accurate quantitative analysis for the optimal scheduling of power systems, helping to improve the accuracy of power load forecasting and ensure the efficient use of power resources.

[0119] Step S5: Obtain the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor, and solve the correlation coefficient between the industrial daily load and each potential correlation factor based on the joint probability distribution function between the industrial daily load and each potential correlation factor.

[0120] In this embodiment, the method for obtaining the parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor is any one of maximum likelihood estimation, Pearson correlation method, regression method, and non-parametric estimation method.

[0121] In this embodiment, preferably, the step of obtaining the parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor includes:

[0122] Based on the pre-processed industrial daily load data in the historical time period and the corresponding data of multiple potential correlation factors, the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor is calculated;

[0123] The calculation formula is:

[0124]

[0125] Among them, τ l is the Kendall rank correlation coefficient between industrial daily load and the lth potential correlation factor, is the number of combinations of selecting 2 data from n data, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, sign{.} is the sign function, X i is the preprocessed daily load data of the ith industry, X j is the j-th industrial day load data after preprocessing, is the i-th data of the l-th potential correlation factor after preprocessing, is the jth data of the lth potential correlation factor after preprocessing;

[0126] According to the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor, the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are calculated respectively.

[0127] Kendall'sτ is a non-parametric rank correlation measurement method that does not rely on the specific distribution assumptions of the data and is suitable for complex data such as industrial load that may not conform to conventional distribution. Secondly, it can effectively capture the nonlinear relationship and sequential characteristics between industrial daily load and potential factors (such as temperature, equipment status, production process, etc.), especially when there is an obvious rank or sequential relationship in the data, providing a more accurate correlation assessment. At the same time, Kendall'sτ is highly robust to outliers and can reduce the interference of abnormal data fluctuations on the analysis results. Kendall'sτ provides precise dependency parameters for the Copula function, which can accurately describe the joint distribution relationship between multiple variables.

[0128] The corresponding relationship between the Gumbel Copula function parameter value θ1, the Clayton Copula function parameter value θ2 and the Kendall rank correlation coefficient τ is shown in Table 1.

[0129] Table 1

[0130]

[0131] The correlation coefficients of industrial daily load and each potential correlation factor, Gumbel Copula function parameters and Clyton Copula function parameters, are solved. If the Gumbel Copula function parameters are farther away from 1 and the Clyton Copula function parameters are farther away from 0, the Kendall rank correlation coefficient |τ l The closer | are to 1, the stronger the correlation between the industrial daily load to be analyzed and the lth potential correlation factor.

[0132] In this second embodiment, a nonferrous metal processing industry is selected as a typical industrial load correlation analysis sample to analyze its data types with high potential correlation. The experiment selects the enterprise's electricity consumption data in August 2023 and other potential interference data in the region as potential analysis basic data.

[0133] The industrial load data includes the daily power consumption, maximum power, minimum power, and average power of the enterprise in August 2023. Figure 2 , Figure 3 As shown, Figure 2 is a schematic diagram of the enterprise electricity consumption curve in the second embodiment, Figure 3 It is a schematic diagram of the enterprise power curve in the second embodiment.

[0134] The potential correlation factors include the company's daily production output, raw material consumption and other data. The specific data are shown in Table 2, which is a data set of potential correlation factors.

[0135] Table 2

[0136]

[0137] Through the method for identifying potential correlation factors of load proposed by the present invention, the function parameters of Gumbel Copula and Clyton Copula of the daily power consumption, maximum power, minimum power, average power and potential correlation factors of the enterprise in the industrial load are calculated respectively, as shown in the following figure: Figure 4 As shown, Figure 4 is a schematic diagram of the nonlinear correlation between the industrial load of an enterprise and potential correlation factors in Example 2, Figure 4 (a) in the figure represents a schematic diagram of function parameters of the Gumbel Copula of the industrial load of an enterprise and potential correlation factors in the second embodiment. Figure 4 (b) in the figure is a schematic diagram of the function parameters of the Clyton Copula of the industrial load of the enterprise and the potential correlation factors in the second embodiment.

[0138] Depend on Figure 4It can be seen that production output, raw material consumption, production process switching, policy intervention on the day, and administrative penalties are highly correlated with the load trend of the industrial enterprise. Among them, government policy intervention has the greatest impact on enterprise production, and production output and production process switching are second in correlation with the industrial load of the industrial enterprise. The correlation between meteorological data such as temperature and precipitation and load is not reflected in this example.

[0139] The method for identifying potential correlation factors of load proposed in the present invention provides a method for quantifying the relationship between industrial daily load and potential correlation factors, which can accurately identify the power consumption characteristics of industrial load and provide basic conditions for helping professionals to screen adjustable industrial loads in the later stage.

[0140] Embodiment 3 of the present invention provides a system for identifying potential load correlation factors, including:

[0141] The data processing module is used to obtain the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, and perform preprocessing;

[0142] The industrial daily load distribution estimation module is used to use the exponential kernel function to perform kernel density estimation on the industrial daily load data in the preprocessed historical time period, and then integrate the kernel density to obtain the marginal probability distribution function of the industrial daily load;

[0143] A potential correlation factor distribution estimation module is used to use an exponential kernel function to perform kernel density estimation on the preprocessed data of each potential correlation factor, and then integrate the kernel density to obtain the marginal probability distribution function of each potential correlation factor;

[0144] The joint probability distribution acquisition module couples the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor;

[0145] The correlation coefficient solving module is used to obtain the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor, and solve the correlation coefficient between the industrial daily load and each potential correlation factor based on the joint probability distribution function between the industrial daily load and each potential correlation factor.

[0146] In this embodiment, specifically, the potential correlation factors include: at least one of meteorological data, calendar factors, random interference factors, and production data of industrial enterprises;

[0147] Among them, meteorological data include: at least one of the highest temperature, the lowest temperature, the air quality index and precipitation; calendar factors include: holidays; random interference factors include: at least one of new orders added on the day and policy intervention; production data of industrial enterprises include: at least one of generated output, raw material consumption and production process switching.

[0148] In this embodiment, preferably, the industrial daily load data within the historical time period and the corresponding data of multiple potential correlation factors are preprocessed, including:

[0149] When the industrial daily load data in the historical time period is missing, the missing data is processed. For the missing industrial daily load data of a single date, the average industrial daily load of the two dates before and after the missing data date is taken as the industrial daily load data of the date of the missing data;

[0150] For missing industrial daily load data for multiple consecutive dates, taking each date of missing data as the center, within the preset date range, select the industrial daily load data whose similarity with the industrial daily load curve of the date of missing data reaches a preset threshold or above, as the industrial daily load data of the date of missing data;

[0151] After processing the missing data, if there are still missing data, the dates containing the missing data will be directly eliminated;

[0152] The data of industrial daily load and its corresponding potential correlation factors in the historical time period after missing data processing are normalized, and the Kalman filtering algorithm is used to smooth the industrial daily load data in the historical time period after normalization to obtain the preprocessed industrial daily load data in the historical time period and its corresponding multiple potential correlation factors.

[0153] In this embodiment, an exponential kernel function is used to perform kernel density estimation on the preprocessed industrial daily load data in the historical time period, and the formula is:

[0154]

[0155] Among them, F(X) is the kernel density estimation of industrial daily load, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, X is the target position of the kernel density estimation, and X i is the i-th industrial daily load data after preprocessing, σ is the standard deviation of the industrial daily load data,

[0156] In this embodiment, an exponential kernel function is used to perform kernel density estimation on the preprocessed data of each potential correlation factor, and the formula is:

[0157]

[0158] Among them, F(x l ) is the kernel density estimate of the lth potential correlation factor, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, x l is the target position of the kernel density estimation of the lth potential correlation factor, is the i-th data of the l-th potential correlation factor after preprocessing, σ l is the standard deviation of the lth potential correlation factor,

[0159] In this embodiment, specifically, the Copula function is any one or more of a Gumbel Copula function, a Clayton Copula function, a Frank Copula function, and a Gaussian Copula.

[0160] In this embodiment, preferably, the marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Gumbel Copula function, as the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Gumbel Copula function, the formula is:

[0161] Y1(μ(X),μ(x l ))=C1(μ(X),μ(x l ))

[0162]

[0163] Among them, Y1(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, C1(.) is the Gumbel Copula function, exp{.} is the exponential function with the natural constant as the base, θ1 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0164] In this embodiment, preferably, the marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through a Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through a Clayton Copula function to obtain a joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Clayton Copula function, and the formula is:

[0165] Y2(μ(X),μ(x l ))=C2(μ(X),μ(x l ))

[0166]

[0167] Among them, Y2(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, C2(.) is the Clayton Copula function, θ2 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

[0168] In this embodiment, preferably, the step of obtaining the parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor includes:

[0169] Based on the pre-processed industrial daily load data in the historical time period and the corresponding data of multiple potential correlation factors, the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor is calculated;

[0170] The calculation formula is:

[0171]

[0172]

[0173] Among them, τ l is the Kendall rank correlation coefficient between industrial daily load and the lth potential correlation factor, is the number of combinations of selecting 2 data from n data, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, sign{.} is the sign function, X i is the preprocessed daily load data of the ith industry, Xj is the j-th industrial day load data after preprocessing, is the i-th data of the l-th potential correlation factor after preprocessing, is the jth data of the lth potential correlation factor after preprocessing;

[0174] According to the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor, the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are calculated respectively.

[0175] Embodiment 4 of the present invention provides a device for identifying potential load correlation factors, including:

[0176] The memory is used to store computer programs; the processor is used to implement the steps of the above-mentioned method for identifying potential load correlation factors when executing the computer program.

[0177] A fifth embodiment of the present invention 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 the above-mentioned method for identifying potential load correlation factors are implemented.

[0178] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0179] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0180] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0181] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0182] Obviously, the above embodiments are merely examples for the purpose of clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A method for identifying potential load correlation factors, characterized in that: The following steps are involved: Obtain the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, and perform preprocessing; After using the exponential kernel function to estimate the kernel density of the industrial daily load data in the preprocessed historical time period, the marginal probability distribution function of the industrial daily load is obtained by integrating it. After using the exponential kernel function to estimate the kernel density of each potential correlation factor after preprocessing, the marginal probability distribution function of each potential correlation factor is obtained by integrating the kernel density. The marginal probability distribution function of industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of industrial daily load and each potential correlation factor. The parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are obtained, and the correlation coefficient between the industrial daily load and each potential correlation factor is solved based on the joint probability distribution function between the industrial daily load and each potential correlation factor.

2. A method for identifying potential load correlation factors according to claim 1, characterized in that: Potential correlation factors include: at least one of meteorological data, calendar factors, random interference factors and production data of industrial enterprises; Among them, meteorological data include: at least one of the highest temperature, the lowest temperature, the air quality index and precipitation; calendar factors include: holidays; random interference factors include: at least one of new orders added on the day and policy intervention; production data of industrial enterprises include: at least one of generated output, raw material consumption and production process switching.

3. A method for identifying potential load correlation factors according to claim 1, characterized in that: Preprocess the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, including: When the industrial daily load data in the historical time period is missing, the missing data is processed. For the missing industrial daily load data of a single date, the average industrial daily load of the two dates before and after the missing data date is taken as the industrial daily load data of the date of the missing data; For missing industrial daily load data for multiple consecutive dates, taking each date of missing data as the center, within the preset date range, select the industrial daily load data whose similarity with the industrial daily load curve of the date of missing data reaches a preset threshold or above, as the industrial daily load data of the date of missing data; After processing the missing data, if there are still missing data, the dates containing the missing data will be directly eliminated; The data of industrial daily load and its corresponding potential correlation factors in the historical time period after missing data processing are normalized, and the Kalman filtering algorithm is used to smooth the industrial daily load data in the historical time period after normalization to obtain the preprocessed industrial daily load data in the historical time period and its corresponding multiple potential correlation factors.

4. A method for identifying potential load correlation factors according to claim 1, characterized in that: The exponential kernel function is used to estimate the kernel density of the industrial daily load data in the preprocessed historical time period. The formula is: Among them, F(X) is the kernel density estimation of industrial daily load, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, X is the target position of the kernel density estimation, and X i is the i-th industrial daily load data after preprocessing, σ is the standard deviation of the industrial daily load data, 5. A method for identifying potential load correlation factors according to claim 1, characterized in that: The exponential kernel function is used to estimate the kernel density of each potential correlation factor after preprocessing. The formula is: Among them, F(x l ) is the kernel density estimate of the lth potential correlation factor, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, x l is the target position of the kernel density estimation of the lth potential correlation factor, is the i-th data of the l-th potential correlation factor after preprocessing, σ l is the standard deviation of the lth potential correlation factor, 6. A method for identifying potential load correlation factors according to claim 1, characterized in that: The Copula function is any one or more of a Gumbel Copula function, a Clayton Copula function, a Frank Copula function, and a Gaussian Copula.

7. A method for identifying potential load correlation factors according to claim 6, characterized in that: The marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Gumbel Copula function as the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Gumbel Copula function, the formula is: Y1(μ(X),μ(x l ))=C1(μ(X),μ(x l )) Among them, Y1(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, C1(.) is the Gumbel Copula function, exp{.} is the exponential function with the natural constant as the base, θ1 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

8. A method for identifying potential load correlation factors according to claim 6, characterized in that: The marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Clayton Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Clayton Copula function, and the formula is: Y2(μ(X),μ(x l ))=C2(μ(X),μ(x l )) Among them, Y2(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, C2(.) is the Clayton Copula function, θ2 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

9. A method for identifying potential load correlation factors according to claim 1, characterized in that: The parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor are obtained, including: Based on the pre-processed industrial daily load data in the historical time period and the corresponding data of multiple potential correlation factors, the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor is calculated; The calculation formula is: Among them, τ l is the Kendall rank correlation coefficient between industrial daily load and the lth potential correlation factor, is the number of combinations of selecting 2 data from n data, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, sign{.} is the sign function, X i is the preprocessed daily load data of the ith industry, X j is the j-th industrial day load data after preprocessing, is the i-th data of the l-th potential correlation factor after preprocessing, is the jth data of the lth potential correlation factor after preprocessing; According to the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor, the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are calculated respectively.

10. A system for identifying potential load correlation factors, characterized in that: include: The data processing module is used to obtain the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, and perform preprocessing; The industrial daily load distribution estimation module is used to use the exponential kernel function to perform kernel density estimation on the industrial daily load data in the preprocessed historical time period, and then integrate the kernel density to obtain the marginal probability distribution function of the industrial daily load; A potential correlation factor distribution estimation module is used to use an exponential kernel function to perform kernel density estimation on the preprocessed data of each potential correlation factor, and then integrate the kernel density to obtain the marginal probability distribution function of each potential correlation factor; The joint probability distribution acquisition module couples the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor; The correlation coefficient solving module is used to obtain the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor, and solve the correlation coefficient between the industrial daily load and each potential correlation factor based on the joint probability distribution function between the industrial daily load and each potential correlation factor.

11. A load potential correlation factor identification system according to claim 10, characterized in that: Potential correlation factors include: at least one of meteorological data, calendar factors, random interference factors and production data of industrial enterprises; Among them, meteorological data include: at least one of the highest temperature, the lowest temperature, the air quality index and precipitation; calendar factors include: holidays; random interference factors include: at least one of new orders added on the day and policy intervention; production data of industrial enterprises include: at least one of generated output, raw material consumption and production process switching.

12. A load potential correlation factor identification system according to claim 10, characterized in that: Preprocess the industrial daily load data within the historical period and the corresponding data of multiple potential correlation factors, including: When the industrial daily load data in the historical time period is missing, the missing data is processed. For the missing industrial daily load data of a single date, the average industrial daily load of the two dates before and after the missing data date is taken as the industrial daily load data of the date of the missing data; For missing industrial daily load data for multiple consecutive dates, taking each date of missing data as the center, within the preset date range, select the industrial daily load data whose similarity with the industrial daily load curve of the date of missing data reaches a preset threshold or above, as the industrial daily load data of the date of missing data; After processing the missing data, if there are still missing data, the dates containing the missing data will be directly eliminated; The data of industrial daily load and its corresponding potential correlation factors in the historical time period after missing data processing are normalized, and the Kalman filtering algorithm is used to smooth the industrial daily load data in the historical time period after normalization to obtain the preprocessed industrial daily load data in the historical time period and its corresponding multiple potential correlation factors.

13. A load potential correlation factor identification system according to claim 10, characterized in that: The exponential kernel function is used to estimate the kernel density of the industrial daily load data in the preprocessed historical time period. The formula is: Among them, F(X) is the kernel density estimation of industrial daily load, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, X is the target position of the kernel density estimation, and X i is the i-th industrial daily load data after preprocessing, σ is the standard deviation of the industrial daily load data, 14. A load potential correlation factor identification system according to claim 10, characterized in that: The exponential kernel function is used to estimate the kernel density of each potential correlation factor after preprocessing. The formula is: Among them, F(x l ) is the kernel density estimate of the lth potential correlation factor, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, h is the bandwidth parameter, K{.} is the exponential kernel function, x l is the target position of the kernel density estimation of the lth potential correlation factor, is the i-th data of the l-th potential correlation factor after preprocessing, σ l is the standard deviation of the lth potential correlation factor, 15. A load potential correlation factor identification system according to claim 10, characterized in that: The Copula function is any one or more of a Gumbel Copula function, a Clayton Copula function, a Frank Copula function, and a Gaussian Copula.

16. A load potential correlation factor identification system according to claim 15, characterized in that: The marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Gumbel Copula function as the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Gumbel Copula function, the formula is: Y1(μ(X),μ(x l ))=C1(μ(X),μ(x l )) Among them, Y1(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, C1(.) is the Gumbel Copula function, exp{.} is the exponential function with the natural constant as the base, θ1 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Gumbel Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

17. A load potential correlation factor identification system according to claim 15, characterized in that: The marginal probability distribution function of the industrial daily load is coupled with the marginal probability distribution function of each potential correlation factor through the Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor, including: coupling the marginal probability distribution function of the industrial daily load with the marginal probability distribution function of each potential correlation factor through the Clayton Copula function to obtain the joint probability distribution function of the industrial daily load and each potential correlation factor corresponding to the Clayton Copula function, and the formula is: Y2(μ(X),μ(x l ))=C2(μ(X),μ(x l )) Among them, Y2(μ(X),μ(x l )) is the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, C2(.) is the Clayton Copula function, θ2 is the parameter of the joint probability distribution function of the industrial daily load and the jth potential correlation factor corresponding to the Clayton Copula function, μ(X) is the marginal probability distribution function of the industrial daily load, μ(x l ) is the marginal probability distribution function of the lth potential correlation factor.

18. A load potential correlation factor identification system according to claim 10, characterized in that: The parameters of the joint probability distribution function of the industrial daily load and each potential correlation factor are obtained, including: Based on the pre-processed industrial daily load data in the historical time period and the corresponding data of multiple potential correlation factors, the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor is calculated; The calculation formula is: Among them, τ l is the Kendall rank correlation coefficient between industrial daily load and the lth potential correlation factor, is the number of combinations of selecting 2 data from n data, n is the total number of industrial daily load data or the total number of data of each potential correlation factor, sign{.} is the sign function, X i is the preprocessed daily load data of the ith industry, X j is the j-th industrial day load data after preprocessing, is the i-th data of the l-th potential correlation factor after preprocessing, is the jth data of the lth potential correlation factor after preprocessing; According to the Kendall rank correlation coefficient between the industrial daily load and each potential correlation factor, the parameters of the joint probability distribution function between the industrial daily load and each potential correlation factor are calculated respectively.

19. A device for identifying potential load correlation factors, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of a method for identifying potential load correlation factors in claims 1 to 9 when executing the computer program.

20. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a method for identifying potential load correlation factors in claims 1 to 9 are implemented.

Images