Time-sharing electric quantity decomposition method based on sensitivity analysis

Abstract

The invention provides a time-sharing electric quantity decomposition method based on sensitivity analysis, and relates to the technical field of electric quantity decomposition, historical and real-time time-sharing electric quantity data and time-sharing meteorological data of a target area are acquired, meteorological factors and a joint data set are obtained after meshing processing, and a key meteorological factor set is screened out; the method comprises the following steps: constructing a basic time-sharing electric quantity decomposition model of quantile regression by using historical time-sharing electric quantity data, inputting real-time data to obtain a basic component and a time-sharing electric quantity residual error, and performing secondary modal decomposition on the residual error to obtain a meteorological sensitive modal component set; constructing an MOEA / D multi-objective optimization model, taking the meteorological sensitive mode component set as input, solving by using a Chebyshev decomposition method, and updating the weight based on the key meteorological factor to obtain a corrected meteorological electric quantity component; and finally, superposing the basic component and the correction component to obtain a final decomposition result. According to the method, the electric quantity decomposition precision is improved through multi-model combination, and powerful support is provided for power dispatching and demand prediction.

Description

Technical Field

[0001] The invention relates to a time-sharing electricity decomposition method based on sensitivity analysis, and belongs to the technical field of electricity decomposition. Background Art

[0002] In the field of power load forecasting and management, the impact of meteorological factors on power demand has received increasing attention. In recent years, with the advancement of data analysis technology and meteorology, researchers have begun to explore incorporating meteorological factors into power decomposition and forecasting models to improve the accuracy and reliability of forecasts.

[0003] For example, the Chinese invention patent with the patent publication number CN109299814A discloses a systematic method for predicting the impact of meteorological factors on electricity. This method systematically analyzes the impact of meteorological factors on power load by setting the base month, calculating the meteorological correlation and the growth rate of meteorological impact on electricity. Specifically, the method divides the year into winter and summer, calculates the correlation separately, and derives the importance of meteorological factors based on the obtained data. This process helps to significantly improve the accuracy of load forecasting and reduce errors.

[0004] However, the above patent adopts seasonal division (winter / summer) and linear growth rate calculation method, which can only handle the linear relationship between a single meteorological factor and electricity volume, and cannot effectively capture the nonlinear coupling effect between multiple meteorological factors (temperature, humidity, wind speed, etc.). Its screening method based on the Pearson correlation coefficient is difficult to identify meteorological factor combinations with significant interactions, resulting in limited prediction accuracy of the model under complex climate conditions. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention proposes a time-sharing electricity decomposition method based on sensitivity analysis.

[0006] The technical solution of the present invention is as follows: The present invention provides a time-sharing electricity decomposition method based on sensitivity analysis, the method comprising: Obtain the time-sharing electricity data related to the target area, including historical time-sharing electricity data, real-time time-sharing electricity data and time-sharing meteorological data, and perform grid processing on the time-sharing electricity data related to obtain meteorological factors and time-sharing electricity-meteorological joint data sets; Based on the time-sharing electricity and meteorological joint data set, the sensitivity index of each meteorological factor to the time-sharing electricity is calculated, and the meteorological factors with sensitivity index greater than the sensitivity threshold are screened to obtain the key meteorological factor set; Building a basic time-sharing electricity decomposition model based on historical time-sharing electricity data, inputting real-time time-sharing electricity data into the basic time-sharing electricity decomposition model to obtain basic time-sharing electricity components; and obtaining time-sharing electricity residuals based on the basic time-sharing electricity components; The improved fully adaptive noise ensemble decomposition algorithm ICEEMDAN is used to perform secondary modal decomposition on the time-sharing power residual to obtain the meteorological sensitive modal component set. A MOEA / D multi-objective optimization model is constructed, and a set of meteorological sensitive modal components is used as input. The MOEA / D multi-objective optimization model is solved by using the Chebyshev decomposition method to obtain a corrected meteorological power component, wherein the MOEA / D multi-objective optimization model updates a weight vector based on a set of key meteorological factors; The basic time-sharing electricity component is superimposed with the corrected meteorological electricity component to obtain the final time-sharing electricity decomposition result.

[0007] As a preferred implementation manner of the present invention, the time-sharing meteorological data includes temperature, humidity, wind speed, sunshine intensity and precipitation.

[0008] As a preferred implementation of the present invention, the gridding process of the time-sharing electricity-related data to obtain the meteorological factors and the time-sharing electricity-meteorological joint data set is specifically as follows: The time-sharing meteorological data is gridded according to the city dimension to obtain gridded meteorological indicators, namely meteorological factors; the historical time-sharing electricity data and real-time time-sharing electricity data are gridded according to the city dimension and industry dimension, and the time-sharing electricity-meteorological joint data set is constructed in combination with meteorological factors, where the grid resolution is 1km×1km; The time-sharing electricity-meteorological joint data set is subjected to data preprocessing, including data alignment, outlier removal and feature standardization.

[0009] As a preferred embodiment of the present invention, the sensitivity index of each meteorological factor to the time-sharing electricity is calculated based on the time-sharing electricity-meteorological joint data set as follows: The global sensitivity of the combined time-sharing electricity and meteorological data set is calculated by the Sobol index and expressed as follows: ; In the formula, For the The global sensitivity of each meteorological factor; For the Meteorological factors; For Other meteorological factors besides is the variance; is the conditional expectation; It is the time-sharing power output variable; Calculating the Mutual Information of Meteorological Factors and Time-of-day Electricity Using Kernel Density Estimation , expressed as: ; In the formula, for and The joint probability density of and They are and The marginal probability density of and They are and Specific value of Based on global sensitivity and mutual information The sensitivity index is calculated and expressed as follows: ; In the formula, is a sensitivity index; is the weight coefficient; is the information entropy of time-sharing electricity.

[0010] As a preferred implementation of the present invention, the basic time-sharing electricity decomposition model is constructed based on the historical time-sharing electricity data as follows: The basic time-sharing electricity decomposition model is a quantile regression model, which is expressed as follows: ; In the formula, for The basic time-sharing power at all times; during training, for The historical time-sharing power consumption at all times, real-time monitoring, for Real-time time-sharing power consumption at all times; and for The regression coefficient of time; and Respectively represent Cities and industries; and They are the total number of cities and the total number of industries respectively; for The moment The coefficient of the variable of each city; for The moment The coefficient of the industry variable for each industry; is the error term; is the quantile; The basic time-sharing electricity decomposition model is trained based on historical time-sharing electricity data to obtain a trained basic time-sharing electricity decomposition model, and real-time time-sharing electricity data is input into the basic time-sharing electricity decomposition model to obtain real-time basic time-sharing electricity components.

[0011] As a preferred implementation of the present invention, the time-sharing power residual is obtained based on the basic time-sharing power component and is expressed as follows: ; In the formula, for Time-sharing power residual at the moment; for Real-time time-sharing power consumption at all times.

[0012] As a preferred implementation of the present invention, the improved fully adaptive noise set decomposition algorithm ICEEMDAN is used to perform secondary modal decomposition on the time-sharing power residual to obtain a set of meteorological sensitive modal components, specifically: The first ICEEMDAN decomposition is performed on the time-sharing power residual, which is decomposed into The intrinsic mode functions and a residual component are expressed as: ; In the formula, is the first decomposition result Intrinsic mode functions; is the residual component obtained from the first decomposition; All the intrinsic mode functions obtained from the first decomposition are selected for the second decomposition, which can be expressed as: ; In the formula, Express The second decomposition yields The intrinsic mode functions, is the total number of intrinsic mode functions obtained by the second decomposition; is the residual component obtained from the second decomposition; All the intrinsic mode functions obtained by the second decomposition are combined to obtain a set of meteorological sensitive mode components , expressed as: .

[0013] As a preferred embodiment of the present invention, the constructed MOEA / D multi-objective optimization model is expressed as follows: ; In the formula, The residual fitting error is minimized; For the Nonlinear mapping function of meteorological factors; For the moment The index of The length of the real-time time-sharing power signal; For the The weight coefficient of each meteorological factor; The goal is to maximize the consistency between weights and sensitivity coefficients; is the total number of meteorological factors; For the The sensitivity index of meteorological factors.

[0014] As a preferred implementation of the present invention, the MOEA / D multi-objective optimization model is solved using the Chebyshev decomposition method, specifically: Defining Chebyshev distance , expressed as: ; In the formula, is a decision variable, used to refer to the The weight coefficient of each meteorological factor; is a set of multi-objective trade-off weights, ; is the target index; is the set of ideal points, ; is the decision variable The corresponding objective function value; A set of uniformly distributed initial weight vectors is generated according to the set of key meteorological factors, which can be expressed as follows: ; In the formula, is the initial weight vector set; is the bit index of the initial weight vector; is the total number of initial weight vectors, each of which corresponds to a subproblem; Calculate the weight vector in the initial weight vector set The distance to other weight vectors, select the initial weight vector whose distance meets the distance threshold to form the corresponding neighborhood ; Randomly generate an initial population set , expressed as: ; In the formula, corresponds to the initial weight vector The solution; From the neighborhood Randomly select two solutions, perform genetic operations on the two selected solutions, obtain new solutions, and calculate the objective function value of the new solution; Perform iterative solution, including calculating the Chebyshev distances of the current solution and the new solution relative to the multi-objective trade-off weights and the ideal point for each subproblem in the neighborhood in each iteration, and replacing the current solution with the new solution if the Chebyshev distance corresponding to the new solution is less than or equal to the Chebyshev distance corresponding to the current solution; If the objective function value is less than the ideal point, then the ideal point is equal to the objective function value; Repeat the above iterative solution process until the preset termination condition is met, then stop the iteration. At this time, the solution in the population is the approximate optimal solution of the MOEA / D multi-objective optimization model, and the corrected meteorological power component is obtained based on the approximate optimal solution.

[0015] As a preferred embodiment of the present invention, the final time-sharing power decomposition result is expressed as follows: ; In the formula, The final time-sharing power decomposition result; is the corrected meteorological electricity component.

[0016] The present invention has the following beneficial effects: 1. The present invention is a time-sharing electricity decomposition method based on sensitivity analysis, which calculates the sensitivity index of each meteorological factor to the time-sharing electricity, and screens out the key meteorological factors whose sensitivity index is greater than the threshold, focusing on the meteorological factors that have a greater impact on the time-sharing electricity, reducing the interference of irrelevant information, improving the calculation efficiency and accuracy of the subsequent model, so that the model can decompose the electricity more specifically; 2. The present invention is a time-sharing electricity decomposition method based on sensitivity analysis. The basic time-sharing electricity decomposition model (quantile regression model) is constructed by using historical time-sharing electricity data, and the real-time time-sharing electricity data is input into the model to obtain the basic time-sharing electricity components. The quantile regression model can consider the electricity distribution under different quantiles, comprehensively reflect the change characteristics of electricity, and more accurately calculate the basic electricity components; 3. The present invention is a time-sharing electricity decomposition method based on sensitivity analysis. The improved fully adaptive noise ensemble decomposition algorithm (ICEEMDAN) is used to perform secondary modal decomposition on the time-sharing electricity residual to obtain a set of meteorological sensitive modal components, further separate and refine the meteorological sensitive information in the residual, and dig out the meteorological influencing factors hidden in the residual, so as to provide more detailed information for the subsequent correction of the meteorological electricity component, which is helpful to improve the accuracy of electricity decomposition; 4. The present invention is a time-sharing electricity decomposition method based on sensitivity analysis, constructs a MOEA / D multi-objective optimization model, and uses the Chebyshev decomposition method to solve it. The weight vector is updated based on a set of key meteorological factors to obtain the corrected meteorological electricity component. The MOEA / D multi-objective optimization model comprehensively considers the two objectives of minimizing the residual fitting error and maximizing the consistency between the weight and the sensitivity coefficient. It can balance and optimize between multiple objectives, so that the corrected meteorological electricity component is more in line with the actual situation, further improving the accuracy of electricity decomposition. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0020] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0021] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0022] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0023] Embodiment 1: See also Figure 1 This embodiment provides a time-based electricity decomposition method based on sensitivity analysis, including the following steps: S1. Obtain the time-sharing electricity data related to the target area through the marketing business system and the data center, including historical time-sharing electricity data, real-time time-sharing electricity data and time-sharing meteorological data, and perform grid processing on the time-sharing electricity data to obtain meteorological factors and time-sharing electricity-meteorological joint data sets; S11, the time-sharing meteorological data includes temperature, humidity, wind speed, sunshine intensity and precipitation; S12, gridding the time-sharing meteorological data according to the city dimension to obtain gridded meteorological indicators, i.e., meteorological factors; gridding the historical time-sharing electricity data and the real-time time-sharing electricity data according to the city dimension and the industry dimension, and constructing a time-sharing electricity-meteorological joint data set in combination with the meteorological factors, wherein the grid resolution is 1 km × 1 km; Furthermore, in this embodiment, the prefecture-level city dimension is divided by administrative region; the industry dimension includes industry, commerce, agriculture, animal husbandry, fishery and urban and rural residents classification; S13, performing data preprocessing on the time-sharing electricity quantity-meteorological joint data set, including data alignment, outlier removal and feature standardization, and using the preprocessed time-sharing electricity quantity-meteorological joint data set for subsequent decomposition calculation; S2. Calculate the sensitivity index of each meteorological factor to the time-sharing electricity based on the time-sharing electricity-meteorological joint data set, and select meteorological factors whose sensitivity index is greater than the sensitivity threshold to obtain a set of key meteorological factors; S21. The global sensitivity corresponding to the combined time-sharing electricity and weather data set is calculated by the Sobol index and expressed as follows: ; In the formula, For the The global sensitivity of each meteorological factor; For the Meteorological factors; For Other meteorological factors besides is the variance; is the conditional expectation; It is the time-sharing power output variable; Calculating the Mutual Information of Meteorological Factors and Time-of-day Electricity Using Kernel Density Estimation , expressed as: ; In the formula, for and The joint probability density of and They are and The marginal probability density of and They are and Specific value of Based on global sensitivity and mutual information The sensitivity index is calculated and expressed as follows: ; In the formula, is a sensitivity index; is the weight coefficient; is the information entropy of time-sharing electricity; S22, the sensitivity threshold is dynamically determined by Monte Carlo cross validation, which is a conventional technical means in the art and will not be described in detail here; S3. Build a basic time-sharing electricity decomposition model based on historical time-sharing electricity data, input real-time time-sharing electricity data into the basic time-sharing electricity decomposition model to obtain basic time-sharing electricity components; and obtain time-sharing electricity residuals based on the basic time-sharing electricity components; S31. In this embodiment, the historical time-sharing electricity data of winter (December-February) and summer (June-August) can be further selected, and a quantile regression model is established for each of the 24 time-sharing points in a day, that is, a basic time-sharing electricity decomposition model, which is expressed as follows: ; In the formula, for The basic time-sharing power at all times; during training, for The historical time-sharing power consumption at all times, real-time monitoring, for Real-time time-sharing power consumption at all times; and for The regression coefficient of time; and Respectively represent Cities and industries; and They are the total number of cities and the total number of industries respectively; for The moment The coefficient of the variable of each city; for The moment The coefficient of the industry variable for each industry; is the error term; is the quantile; The reasons for choosing winter and summer include: 1) these two seasons are the seasons with the most obvious temperature fluctuations, and the impact of temperature and humidity on electricity consumption is particularly prominent; 2) temperature and electricity consumption are negatively correlated and positively correlated in these two seasons, respectively, which is representative; The selection of quantiles is also crucial. Different quantiles give very different regression results. Specifically: 1) When =0.1 and =0.9, used to capture extreme power usage scenarios: =0.1: reflects the minimum electricity demand under extremely low temperature or low humidity conditions (such as cold waves in winter and heavy rains in summer), corresponding to the "bottom line" load of the power system; =0.9: reflects the peak power demand under extremely high temperature or high humidity conditions (such as summer heat and winter heating peak), corresponding to the "overload" risk point of the power grid; when =0.25 and =0.75, analyze the normal fluctuation range: =0.25: represents the lower range of power distribution and can be used to evaluate base load in mild weather; =0.75: represents the higher range of power distribution, corresponding to the increase in power consumption when high temperature or humidity rises; when =0.5 (median), used for robust basic power estimation: =0.5 is the middle value of the power distribution, which is insensitive to abnormal values ​​(such as holidays and equipment failures). It is used to calculate the basic power and separate the baseline of meteorological influences; In this embodiment, the median of the power distribution is selected to provide a robust baseline load and avoid the interference of extreme events on the model results, so the quantile is selected as ; S32, training the basic time-sharing power decomposition model based on historical time-sharing power data to obtain a trained basic time-sharing power decomposition model, and inputting real-time time-sharing power data into the basic time-sharing power decomposition model to obtain real-time basic time-sharing power components; S33, based on the basic time-sharing power component, the time-sharing power residual is obtained and expressed as: ; In the formula, for Time-sharing power residual at the moment; for Real-time time-sharing power consumption at all times; S4. Use the improved fully adaptive noise set decomposition algorithm ICEEMDAN to perform secondary modal decomposition on the time-sharing power residual to obtain a set of meteorological sensitive modal components. Specifically: S41. Perform the first ICEEMDAN decomposition on the time-sharing power residual and decompose it into The intrinsic mode functions and a residual component are expressed as: ; In the formula, is the first decomposition result Intrinsic mode functions; is the residual component obtained from the first decomposition; S42, select all the intrinsic mode functions obtained by the first decomposition for the second decomposition, and express it as follows: ; In the formula, Express The second decomposition yields The intrinsic mode functions, is the total number of intrinsic mode functions obtained by the second decomposition; is the residual component obtained from the second decomposition; S43. Combine all the intrinsic mode functions obtained from the second decomposition to obtain a set of meteorological sensitive mode components. , expressed as: ; S5. Construct a decomposition-based multi-objective evolutionary algorithm MOEA / D multi-objective optimization model, take a set of meteorological sensitive modal components as input, solve the MOEA / D multi-objective optimization model using the Chebyshev decomposition method, and obtain a corrected meteorological power component, wherein the MOEA / D multi-objective optimization model updates a weight vector based on a set of key meteorological factors; S51. In this embodiment, the two objectives set include minimizing the residual fitting error and maximizing the consistency of the weight and sensitivity coefficient. The constructed MOEA / D multi-objective optimization model is expressed as follows: ; In the formula, The residual fitting error is minimized; For the Nonlinear mapping function of meteorological factors; For the moment The index of The length of the real-time time-sharing power signal; For the The weight coefficient of each meteorological factor; The goal is to maximize the consistency between weights and sensitivity coefficients; is the total number of meteorological factors; For the Sensitivity index of each meteorological factor; Among them, the minimization of residual fitting error is to make the combination of meteorological factors fit the meteorological sensitive signals in the residual to the greatest extent; the maximization of the consistency of weights and sensitivity coefficients is to prevent the optimized weights from completely deviating from the physical meaning (for example, a meteorological factor with little actual impact is given too high a weight). Through the above two goals, the residual signals in the data are utilized, and the physical impact mechanism of meteorological factors is respected, avoiding the overfitting risk of pure data-driven models; S52. The MOEA / D multi-objective optimization model is solved using the Chebyshev decomposition method, specifically: Defining Chebyshev distance , expressed as: ; In the formula, is a decision variable, used to refer to the The weight coefficient of each meteorological factor; is a set of multi-objective trade-off weights. Specifically, in this embodiment, since there are two objectives in this embodiment, the corresponding ; is the target index; is the set of ideal points, ; is the decision variable The corresponding objective function value; A set of uniformly distributed initial weight vectors is generated according to the set of key meteorological factors, which can be expressed as follows: ; In the formula, is the initial weight vector set; is the bit index of the initial weight vector; is the total number of initial weight vectors, each of which corresponds to a subproblem; Calculate the weight vector in the initial weight vector set The distance to other weight vectors, select the initial weight vector whose distance meets the distance threshold to form the corresponding neighborhood ; Randomly generate an initial population set , expressed as: ; In the formula, corresponds to the initial weight vector The solution; From the neighborhood Randomly select two solutions, perform genetic operations on the two selected solutions, obtain new solutions, and calculate the objective function value of the new solution; Perform iterative solution, including calculating the Chebyshev distances of the current solution and the new solution relative to the multi-objective trade-off weights and the ideal point for each subproblem in the neighborhood in each iteration, and replacing the current solution with the new solution if the Chebyshev distance corresponding to the new solution is less than or equal to the Chebyshev distance corresponding to the current solution; If the objective function value is less than the ideal point, then the ideal point is equal to the objective function value; Repeat the above iterative solution process until the preset termination condition is met, then stop the iteration. At this time, the solution in the population is the approximate optimal solution of the MOEA / D multi-objective optimization model, and the corrected meteorological power component is obtained based on the approximate optimal solution. S6. Superimpose the basic time-sharing electricity component and the corrected meteorological electricity component to obtain the final time-sharing electricity decomposition result, which is expressed as follows: ; In the formula, The final time-sharing power decomposition result; is the corrected meteorological electricity component.

[0024] In summary, this embodiment provides a time-based electricity decomposition method based on sensitivity analysis.

[0025] In the embodiments of the present invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. A and B may be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c may represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c may be single or multiple.

[0026] Those skilled in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0027] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0028] In several embodiments provided by the present invention, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk and other media that can store program codes.

[0029] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A time-based electricity decomposition method based on sensitivity analysis, characterized in that: The method comprises: Obtain the time-sharing electricity data related to the target area, including historical time-sharing electricity data, real-time time-sharing electricity data and time-sharing meteorological data, and perform grid processing on the time-sharing electricity data related to obtain meteorological factors and time-sharing electricity-meteorological joint data sets; Based on the time-sharing electricity and meteorological joint data set, the sensitivity index of each meteorological factor to the time-sharing electricity is calculated, and the meteorological factors with sensitivity index greater than the sensitivity threshold are screened to obtain the key meteorological factor set; Building a basic time-sharing electricity decomposition model based on historical time-sharing electricity data, inputting real-time time-sharing electricity data into the basic time-sharing electricity decomposition model to obtain basic time-sharing electricity components; and obtaining time-sharing electricity residuals based on the basic time-sharing electricity components; The improved fully adaptive noise ensemble decomposition algorithm ICEEMDAN is used to perform secondary modal decomposition on the time-sharing power residual to obtain the meteorological sensitive modal component set. A MOEA / D multi-objective optimization model is constructed, and a set of meteorological sensitive modal components is used as input. The MOEA / D multi-objective optimization model is solved by using the Chebyshev decomposition method to obtain a corrected meteorological power component, wherein the MOEA / D multi-objective optimization model updates a weight vector based on a set of key meteorological factors; The basic time-sharing electricity component is superimposed with the corrected meteorological electricity component to obtain the final time-sharing electricity decomposition result.

2. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 1 is characterized in that: The time-sharing meteorological data include temperature, humidity, wind speed, sunshine intensity and precipitation.

3. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 1 is characterized in that: The gridding process of the time-sharing electricity-related data to obtain the meteorological factors and the time-sharing electricity-meteorological joint data set is specifically: The time-sharing meteorological data is gridded according to the city dimension to obtain gridded meteorological indicators, namely meteorological factors; the historical time-sharing electricity data and real-time time-sharing electricity data are gridded according to the city dimension and industry dimension, and the time-sharing electricity-meteorological joint data set is constructed in combination with meteorological factors, where the grid resolution is 1km×1km; The time-sharing electricity-meteorological joint data set is subjected to data preprocessing, including data alignment, outlier removal and feature standardization.

4. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 1 is characterized in that: The sensitivity index of each meteorological factor to time-sharing electricity is calculated based on the time-sharing electricity-meteorological joint data set as follows: The global sensitivity of the combined time-sharing electricity and meteorological data set is calculated by the Sobol index and expressed as follows: ； In the formula, For the The global sensitivity of each meteorological factor; For the Meteorological factors; For Other meteorological factors besides is the variance; is the conditional expectation; It is the time-sharing power output variable; Calculating the Mutual Information of Meteorological Factors and Time-of-day Electricity Using Kernel Density Estimation , expressed as: ； In the formula, for and The joint probability density of and They are and The marginal probability density of and They are and Specific value of Based on global sensitivity and mutual information The sensitivity index is calculated and expressed as follows: ； In the formula, is a sensitivity indicator; is the weight coefficient; is the information entropy of time-sharing electricity.

5. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 1 is characterized in that: The basic time-sharing electricity decomposition model based on historical time-sharing electricity data is constructed as follows: The basic time-sharing electricity decomposition model is a quantile regression model, which is expressed as follows: ； In the formula, for The basic time-sharing power at all times; during training, for The historical time-sharing power consumption at all times, real-time monitoring, for Real-time time-sharing power consumption at all times; and for The regression coefficient of time; and Respectively represent Cities and industries; and They are the total number of cities and the total number of industries respectively; for The moment The coefficient of the variable of each city; for The moment The coefficient of the industry variable for each industry; is the error term; is the quantile; The basic time-sharing electricity decomposition model is trained based on historical time-sharing electricity data to obtain a trained basic time-sharing electricity decomposition model, and real-time time-sharing electricity data is input into the basic time-sharing electricity decomposition model to obtain real-time basic time-sharing electricity components.

6. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 5 is characterized in that: The time-sharing power residual obtained based on the basic time-sharing power component is expressed as follows: ； In the formula, for Time-sharing power residual at the moment; for Real-time time-sharing power consumption at all times.

7. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 6 is characterized in that: The improved fully adaptive noise set decomposition algorithm ICEEMDAN is used to perform secondary modal decomposition on the time-sharing power residual to obtain a set of meteorological sensitive modal components, specifically: The first ICEEMDAN decomposition is performed on the time-sharing power residual, which is decomposed into The intrinsic mode functions and a residual component are expressed as: ； In the formula, is the first decomposition result Intrinsic mode functions; is the residual component obtained from the first decomposition; All the intrinsic mode functions obtained from the first decomposition are selected for the second decomposition, which can be expressed as: ； In the formula, Express The second decomposition yields The intrinsic mode functions, is the total number of intrinsic mode functions obtained by the second decomposition; is the residual component obtained from the second decomposition; All the intrinsic mode functions obtained by the second decomposition are combined to obtain the set of meteorological sensitive mode components , expressed as: 。 8. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 7 is characterized in that: The constructed MOEA / D multi-objective optimization model is expressed as follows: ； In the formula, The residual fitting error is minimized; For the Nonlinear mapping function of meteorological factors; For the moment The index of The length of the real-time time-sharing power signal; For the The weight coefficient of each meteorological factor; The goal is to maximize the consistency between weights and sensitivity coefficients; is the total number of meteorological factors; For the The sensitivity index of meteorological factors.

9. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 8 is characterized in that: The Chebyshev decomposition method is used to solve the MOEA / D multi-objective optimization model, specifically: Defining Chebyshev distance , expressed as: ； In the formula, is a decision variable, used to refer to the The weight coefficient of each meteorological factor; is a set of multi-objective trade-off weights, ; is the target index; is the set of ideal points, ; is the decision variable The corresponding objective function value; A set of uniformly distributed initial weight vectors is generated according to the set of key meteorological factors, which can be expressed as follows: ； In the formula, is the initial weight vector set; is the bit index of the initial weight vector; is the total number of initial weight vectors, each of which corresponds to a subproblem; Calculate the weight vector in the initial weight vector set The distance to other weight vectors, select the initial weight vector whose distance meets the distance threshold to form the corresponding neighborhood ; Randomly generate an initial population set , expressed as: ； In the formula, corresponds to the initial weight vector The solution; From the neighborhood Randomly select two solutions, perform genetic operations on the two selected solutions, obtain new solutions, and calculate the objective function value of the new solution; Perform iterative solution, including calculating the Chebyshev distances of the current solution and the new solution relative to the multi-objective trade-off weights and the ideal point for each subproblem in the neighborhood in each iteration, and replacing the current solution with the new solution if the Chebyshev distance corresponding to the new solution is less than or equal to the Chebyshev distance corresponding to the current solution; If the objective function value is less than the ideal point, then the ideal point is equal to the objective function value; Repeat the above iterative solution process until the preset termination condition is met, then stop the iteration. At this time, the solution in the population is the approximate optimal solution of the MOEA / D multi-objective optimization model, and the corrected meteorological power component is obtained based on the approximate optimal solution.

10. The time-sharing electricity decomposition method based on sensitivity analysis according to claim 9 is characterized in that: The final time-sharing power decomposition result is expressed as follows: ； In the formula, The final time-sharing power decomposition result; is the corrected meteorological electricity component.

Images