Method and device for generating and extracting typical power consumption time sequence scene of large industrial user

Abstract

The invention discloses a large industrial user power consumption typical time sequence scene generation and extraction method, which comprises the following steps of: obtaining a large industrial user power consumption fluctuation range based on historical similar day data, establishing a large industrial user power consumption prediction error model, performing sampling probability interval correction of error sampling according to the large industrial user power consumption fluctuation range, and obtaining a large industrial user power consumption typical time sequence scene. Generating a power consumption time sequence scene of a large industrial user; performing reduction on the time sequence scene by using a K-means method to obtain a scene set including a typical scene and an extreme scene; the scene generation quality is evaluated according to the scene coverage rate, the scene width and the correlation approximation degree index; establishing a time sequence scene state matrix and a time sequence scene electricity consumption deviation matrix, analyzing the deviation between a generated scene and predicted electricity consumption, extracting an information matrix according to the state matrix and the deviation matrix scene, and carrying out typical time sequence scene extraction; according to the invention, the power consumption scene prediction accuracy of large industrial users can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of typical time sequence scene generation and extraction, and more specifically, to a method for generating and extracting typical time sequence scenes of electricity consumption by large industrial users. Background Art

[0002] In the power market environment, the electricity consumption of large industrial users is uncertain, which brings challenges to the planning, dispatching and load management of power systems by power companies. The power consumption scenario generation technology for large industrial users can generate a series of representative future scenarios based on historical data and prediction models. These scenarios can reflect the possible range of changes in the power consumption of large industrial users and their probability distribution, provide a rich information basis for power load managers, and help them formulate more scientific and reasonable load management plans to cope with the uncertainty of power consumption of large industrial users. Through in-depth analysis of the power consumption scenarios of large industrial users, key indicators such as the supply and demand balance, spare capacity requirements, and system stability of the power system under different scenarios can also be evaluated, thereby identifying potential risk points and weak links. The generation and extraction method of typical time series scenarios of power consumption of large industrial users in the power market environment can provide grid operation boundary constraints for the above-mentioned businesses, effectively improve the reliability and flexibility of the actual operation of the grid, and provide a practical reference for power load managers.

[0003] However, the above-disclosed technical solution has at least the following technical problems: the previous methods for generating and extracting electricity consumption scenarios of large industrial users often adopt random sampling methods, without considering the timing correlation of electricity consumption of large industrial users, which will cause no fluctuation or drastic fluctuation in the sampling before and after; therefore, in response to the above problems, the present invention proposes a method and device for generating and extracting typical timing scenarios of electricity consumption of large industrial users. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for generating and extracting typical time series scenarios of electricity consumption of large industrial users. By generating a method for generating typical time series scenarios of electricity consumption of large industrial users based on sampling probability interval correction, the problems of low accuracy and weak time series correlation of electricity consumption prediction scenarios of large industrial users are solved.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The method for generating and extracting typical time series scenarios of electricity consumption of large industrial users includes the following steps:

[0007] Based on historical similar day data, the power consumption fluctuation range of large industrial users is obtained, and a large industrial user power consumption prediction error model is established. The sampling probability interval of the error sampling is corrected according to the power consumption fluctuation range of large industrial users to generate a large industrial user power consumption time series scenario;

[0008] Use K-means method to simplify the time series scenarios and obtain a scenario set including typical scenarios and extreme scenarios;

[0009] The quality of scene generation is evaluated based on scene coverage, scene width, and correlation similarity indicators;

[0010] Establish a time series scenario state matrix and a time series scenario power consumption deviation matrix, analyze the deviation between the generated scenario and the predicted power consumption, extract the information matrix based on the state matrix and the deviation matrix scenario, and carry out typical time series scenario extraction.

[0011] Furthermore, the typical time series scene extraction includes:

[0012] Use the time series scenario electricity consumption deviation matrix to store the difference between the time series scenario electricity consumption of large industrial users and the predicted scenario electricity consumption of large industrial users;

[0013] Generate a time sequence scene state matrix according to the deviation of each element in the deviation matrix from the threshold;

[0014] According to the large industrial user electricity consumption timing scenario state matrix, a typical large industrial user electricity consumption timing scenario extraction information matrix is ​​obtained, and then the typical large industrial user electricity consumption timing scenario is extracted.

[0015] Furthermore, the method for obtaining the fluctuation range of electricity consumption of large industrial users based on historical similar day data is as follows:

[0016] Based on the weather forecast information for the next day, the actual operation data of electricity consumption of large industrial users on days with similar historical weather are collected, and the fluctuation curve of electricity consumption data of large industrial users is analyzed to obtain the fluctuation range of electricity consumption of large industrial users in adjacent time periods on days with similar weather. The calculation formula for the fluctuation of electricity consumption of large industrial users before and after is as follows:

[0017] ΔP t =P t+T -P t

[0018] Where: ΔP t is the fluctuation value of electricity consumption of large industrial users at adjacent moments before and after the interval T; T is the time interval; P t is the electricity consumption of large industrial users at time t;

[0019] The power consumption fluctuation set A of large industrial users is:

[0020]

[0021] Where: k t→t+1.dn , k t→t+1.up are the lower and upper limits of power fluctuation of large industrial users at adjacent sampling moments; t is the sampling moment; ΔP t.qis the electricity consumption fluctuation from time t to time t+1 on similar days q; S ander For similar day collection.

[0022] Furthermore, the establishment of a large industrial user electricity consumption prediction error model includes:

[0023] Based on historical forecast data and actual operation data of large industrial users, a large industrial user electricity consumption forecast error model is established:

[0024] P t =P t,predict +ξ t

[0025] Where: P t,predict is the predicted power consumption of large industrial users at time t; t The electricity consumption prediction error for large industrial users.

[0026] Furthermore, the sampling probability interval correction of error sampling according to the power consumption fluctuation range of the large industrial user to generate the power consumption time sequence scenario of the large industrial user includes:

[0027] The Latin hypercube sampling (LHS) method is used to extract the initial prediction error values ​​of all time series scenarios vertically, and the prediction values ​​are superimposed to calculate the power of all time series scenarios at the first moment;

[0028] For each time series scenario, the sampling interval is corrected by considering the fluctuation range of electricity consumption of large industrial users at adjacent moments. The quality of scenario generation is evaluated based on the scenario coverage, scenario width, and correlation approximation indicators.

[0029] Furthermore, when evaluating the quality of scene generation,

[0030] 1) Scene coverage P cover Indicates the coverage of the generated scenario power value to the actual scenario power value:

[0031]

[0032] Where: B is a 0-1 variable. If the actual scene power value is between the maximum and minimum values ​​of the generated scene, then B = 1, otherwise B = 0;

[0033] 2) Scene width P width Reflects the maximum power range between different scenarios in the generated scenario set:

[0034]

[0035] Where: P s,t ,P t,tare the generated time series scenario s and the power consumption value at time t respectively; S is the generated scenario set;

[0036] 3) Correlation approximation P corr Reflects the approximation degree of the correlation between the generated scene and the actual data at any time:

[0037] P corr =|s real,t,t+d -s s,t,t+d |

[0038] Where: s real,t,t+d ,s s,t,t+d are the correlation coefficients between the actual scene and the generated scene at any interval d.

[0039] Furthermore, when extracting typical time series scenarios, it includes:

[0040] Use the time series scenario electricity consumption deviation matrix to store the difference between the time series scenario electricity consumption generated by large industrial users and the predicted scenario electricity consumption of large industrial users:

[0041] ΔP=P sce,out -P′ pre,out

[0042]

[0043] Where: ΔP is the power consumption deviation matrix of the time series scenario; P sce,out is the total power consumption matrix of large industrial users in the generated time series scenario; P′ pre,out Because P sce,out The expanded total matrix of electricity consumption forecast for large industrial users;

[0044] The time sequence scene state matrix includes:

[0045]

[0046] Where: the matrix element S is a 0-1 variable, which represents the power consumption deviation state quantity of the time series scenario s at time c;

[0047] Set the threshold value of large industrial users' electricity consumption in the time series scenario deviating from the predicted scenario electricity consumption ξ p,pos ,ξ p,neg , respectively represent the positive deviation threshold and negative deviation threshold of electricity consumption of large industrial users:

[0048]

[0049] Where: p is the scene type, ΔP st and S st They are respectively the power consumption deviation matrix elements and deviation state matrix elements of large industrial users in time series scenarios.

[0050] Furthermore, when extracting typical time series scenes:

[0051] The product operation is performed on the large industrial user electricity consumption time series scenario state matrix and the time series scenario state matrix transpose matrix to obtain the typical large industrial user electricity consumption time series scenario extraction information matrix Y:

[0052]

[0053] The time period threshold ξ is set, and the large industrial user electricity consumption time series scenario m that satisfies Y>ξ is extracted as the typical time series scenario of large industrial user electricity consumption.

[0054] The device for generating and extracting typical time-series scenarios of electricity consumption of large industrial users includes:

[0055] A large industrial user electricity consumption time series scenario acquisition module is used to obtain the electricity consumption fluctuation range of large industrial users based on historical similar day data, establish a large industrial user electricity consumption prediction error model, perform error sampling sampling probability interval correction according to the large industrial user electricity consumption fluctuation range, and generate a large industrial user electricity consumption time series scenario;

[0056] The scene set acquisition module is used to simplify the time series scenes using the K-means method to obtain a scene set containing typical scenes and extreme scenes;

[0057] The scene generation quality evaluation module is used to evaluate the quality of scene generation based on scene coverage, scene width, and correlation similarity indicators;

[0058] The typical time series scenario extraction module is used to establish a time series scenario state matrix and a time series scenario power consumption deviation matrix, analyze the deviation between the generated scenario and the predicted power consumption, extract the information matrix based on the state matrix and the deviation matrix scenario, and carry out typical time series scenario extraction.

[0059] A computer-readable storage medium having a non-volatile program code executable by a processor, characterized in that when the computer program is executed by the processor, the steps of the method for generating and extracting typical time sequence scenarios of electricity consumption of large industrial users as described above are implemented.

[0060] A computing device comprising:

[0061] one or more processing units;

[0062] A storage unit for storing one or more programs,

[0063] Wherein, when the one or more programs are executed by the one or more processing units, the one or more processing units execute the method for generating and extracting typical time series scenarios of electricity consumption of large industrial users as described above.

[0064] The present invention has the following beneficial effects:

[0065] 1. The present invention comprehensively evaluates the impact of various factors on the generation of power consumption scenarios of large industrial users by considering the impact of time series correlation, and can improve the accuracy of power consumption scenario prediction for large industrial users;

[0066] 2. Through multi-faceted analysis, the present invention can accurately identify typical scenarios and extreme scenarios, thereby achieving the best prediction of electricity consumption scenarios for large industrial users, improving the economy and operating efficiency of the power system, and providing decision-making references for load managers. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 A schematic diagram of a flow chart of the time series scene generation based on sampling probability interval correction of the present invention;

[0068] Figure 2 The present invention is a schematic diagram of the time series scene sampling generation process. DETAILED DESCRIPTION

[0069] 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.

[0070] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of 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 to be beyond the scope of this application.

[0071] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0072] Example 1

[0073] The method for generating and extracting typical time series scenarios of electricity consumption of large industrial users includes the following steps:

[0074] Based on historical similar day data, the power consumption fluctuation range of large industrial users is obtained, and a large industrial user power consumption prediction error model is established. The sampling probability interval of the error sampling is corrected according to the power consumption fluctuation range of large industrial users to generate a large industrial user power consumption time series scenario;

[0075] Use K-means method to simplify the time series scenarios and obtain a scenario set including typical scenarios and extreme scenarios;

[0076] The quality of scene generation is evaluated based on scene coverage, scene width, and correlation similarity indicators;

[0077] Establish a time series scenario state matrix and a time series scenario power consumption deviation matrix, analyze the deviation between the generated scenario and the predicted power consumption, extract the information matrix based on the state matrix and the deviation matrix scenario, and carry out typical time series scenario extraction.

[0078] Generation of electricity consumption time series scenarios for large industrial users: The electricity consumption time series scenarios for large industrial users can be regarded as a random process X, which can be represented by a set of N-dimensional random variables:

[0079]

[0080] Where x is the random variable of electricity consumption of large industrial users. By performing multiple time series sampling on the electricity consumption of large industrial users, we can obtain a set of time series scenarios of electricity consumption of large industrial users.

[0081] The generation process of uncertain time series scenarios of electricity consumption of large industrial users based on sampling probability interval correction is as follows: Figure 1 As shown, it can be divided into three steps: determining the fluctuation range of electricity consumption of large industrial users, modeling electricity consumption of large industrial users, and generating time series scenario sampling.

[0082] Determination of fluctuation range of electricity consumption of large industrial users:

[0083] Based on the weather forecast information for the next day, the actual electricity consumption data of large industrial users on similar historical weather days are collected, and the fluctuation curve of the electricity consumption data of large industrial users is analyzed to obtain the fluctuation range of electricity consumption of large industrial users in each adjacent period on similar weather days. The calculation formula for the fluctuation of electricity consumption of large industrial users before and after is as follows:

[0084] ΔP t =P t+T -P t

[0085] Where: ΔP t is the fluctuation value of electricity consumption of large industrial users at adjacent moments before and after the interval T; T is the time interval; P t is the electricity consumption of large industrial users at time t.

[0086] The power consumption fluctuation set A of large industrial users is:

[0087]

[0088] Where: k t→t+1.dn , k t→t+1.up are the lower and upper limits of power fluctuation of large industrial users at adjacent sampling moments; t is the sampling moment; ΔP t.q S is the power consumption fluctuation of large industrial users from time t to time t+1 on similar days q; ander For similar day collection.

[0089] Electricity consumption modeling for large industrial users: Based on historical forecast data and actual electricity consumption data of large industrial users, a forecast error model for electricity consumption of large industrial users is established:

[0090] P t =P t,predict +ξ t

[0091] Where: P t,predict is the predicted power consumption of large industrial users at time t; t Electricity consumption prediction error for large industrial users.

[0092] Time series scenario sampling based on sampling probability interval correction: The Latin hypercube sampling (LHS) method is used to extract the initial prediction error values ​​of all time series scenarios "vertically", and the predicted values ​​are superimposed to calculate the power of all time series scenarios at the first moment; for each time series scenario, the sampling interval is corrected "horizontally" by considering the fluctuation range of electricity consumption of large industrial users at adjacent moments.

[0093] Assume that N time series scenarios need to be generated. Based on the cumulative distribution of the power consumption forecast error corresponding to the first moment of large industrial users, use LHS to extract N forecast error values, and superimpose the power consumption forecast values ​​to obtain the power consumption sample value corresponding to the first moment of each scenario. For each time series scenario s, the forecast error is sampled moment by moment. Considering the historical volatility of power consumption of large industrial users, the power at adjacent moments must meet the following requirements:

[0094] P s,t+1,sample -P s,t,sample ∈[k t→t+1,dn ,k t→t+1,up ]

[0095] Where: P s,t+1,sample , P s,t,sample are the power consumption and sampling value of the time series scenario s at time t+1 respectively.

[0096] Before sampling the prediction error, the sampling interval is corrected first. Under the premise that the power consumption sampling value at time t is known, the power consumption prediction value of large industrial users determines the power consumption prediction error sampling range at time t+1:

[0097]

[0098] Where: P s,t+1,error is the sample value of the forecast error of large industrial users in the time series scenario s at time t+1; P s,Δ,t→t+1 is the difference between the predicted power consumption values ​​of large industrial users in time series scenario s between two moments.

[0099] P s,t+1,error Substituting into the prediction error cumulative distribution function F, we get the corresponding probability interval [c1, c2], that is:

[0100]

[0101] Where c1 is the lower limit of the probability interval; c2 is the upper limit of the probability interval; the sampling probability interval is adjusted from [0, 1] to [c1, c2], and random sampling is performed in the interval [c1, c2] to obtain the large industrial user electricity consumption prediction error P at sampling time t+1 s,t+1,error , superimposed predicted value P t+1,predict Then, the sampling value P of large industrial users’ electricity consumption at time t+1 can be obtained considering the historical fluctuation characteristics of large industrial users’ electricity consumption. s,t+1,predict .

[0102] Repeat the above steps until the power sampling at all times of all time series scenarios is completed.

[0103] Construction of scene coverage, scene width, and correlation approximation indicators:

[0104] 1) Scene coverage P cover Indicates the coverage of the generated scenario power value to the actual scenario power value.

[0105]

[0106] Where: B is a 0-1 variable. If the actual scene power value is between the maximum and minimum values ​​of the generated scene, then B = 1, otherwise B = 0;

[0107] 2) Scene width P width Reflects the maximum power range between different scenarios in the generated scenario set.

[0108]

[0109] Where: P s,t ,P t,t are the generated time sequence scenario s and the active output value at time t respectively; S is the generated scenario set;

[0110] 3) Correlation approximation P corrReflects the approximation degree of the correlation between the generated scene and the actual data at any time:

[0111] P corr =|s real,t,t+d -s s,t,t+d |

[0112] Where: s real,t,t+d ,s s,t,t+d are the correlation coefficients between the actual scene and the generated scene at any interval d.

[0113] Time series scenario electricity consumption deviation matrix: The time series scenario electricity consumption deviation matrix is ​​used to store the difference between the time series scenario electricity consumption generated by large industrial users and the predicted scenario electricity consumption of large industrial users.

[0114] ΔP=P sce,out -P p ' re,out

[0115]

[0116] Where: ΔP is the power consumption deviation matrix of the time series scenario; P sce,out is the total power consumption matrix of large industrial users in the generated time series scenario; P p ' re,out Because P sce,out The expanded total matrix of electricity consumption forecast for large industrial users.

[0117] Timing scenario state matrix:

[0118]

[0119] Where: the matrix element S is a 0-1 variable, which represents the power consumption deviation state quantity of the time series scenario s at time c.

[0120] Set the threshold value of large industrial users' electricity consumption in the time series scenario deviating from the predicted scenario electricity consumption ξ p,pos ,ξ p,neg , respectively representing the positive deviation threshold and negative deviation threshold of electricity consumption of large industrial users.

[0121]

[0122] Where: p is the scene type, ΔP st and S st They are respectively the power consumption deviation matrix elements and deviation state matrix elements of the power consumption timing scenario of large industrial users.

[0123] Extraction of typical time series scenarios of electricity consumption by large industrial users: Multiply the state matrix of the time series scenarios of electricity consumption by large industrial users and the transposed matrix of the state matrix of the time series scenarios of electricity consumption by large industrial users to obtain the information matrix Y for extraction of typical time series scenarios of electricity consumption by large industrial users.

[0124]

[0125] Set a time period threshold ξ. Extract the large industrial user electricity consumption time sequence scenario m that satisfies Y>ξ as the typical large industrial user electricity consumption time sequence scenario.

[0126] Example 2

[0127] The device for generating and extracting typical time-series scenarios of electricity consumption of large industrial users includes:

[0128] A large industrial user electricity consumption time series scenario acquisition module is used to obtain the electricity consumption fluctuation range of large industrial users based on historical similar day data, establish a large industrial user electricity consumption prediction error model, perform error sampling sampling probability interval correction according to the large industrial user electricity consumption fluctuation range, and generate a large industrial user electricity consumption time series scenario;

[0129] The scene set acquisition module is used to simplify the time series scenes using the K-means method to obtain a scene set containing typical scenes and extreme scenes;

[0130] The scene generation quality evaluation module is used to evaluate the quality of scene generation based on scene coverage, scene width, and correlation similarity indicators;

[0131] The typical time series scenario extraction module is used to establish a time series scenario state matrix and a time series scenario power consumption deviation matrix, analyze the deviation between the generated scenario and the predicted power consumption, extract the information matrix based on the state matrix and the deviation matrix scenario, and carry out typical time series scenario extraction.

[0132] A computer-readable storage medium having a non-volatile program code executable by a processor, characterized in that when the computer program is executed by the processor, the steps of the method for generating and extracting typical time sequence scenarios of electricity consumption of large industrial users as described above are implemented.

[0133] A computing device is also provided, comprising:

[0134] one or more processing units;

[0135] A storage unit for storing one or more programs,

[0136] Among them, when the one or more programs are executed by the one or more processing units, the one or more processing units execute the method for generating and extracting typical timing scenarios of electricity consumption of large industrial users in this embodiment; it should be noted that the computing device may include but is not limited to a processing unit and a storage unit; those skilled in the art can understand that the computing device including a processing unit and a storage unit does not constitute a limitation on the computing device, and may include more components, or a combination of certain components, or different components. For example, the computing device may also include input and output devices, network access devices, buses, etc.

[0137] A computer-readable storage medium having a non-volatile program code executable by a processor, wherein the computer program, when executed by the processor, implements the steps of the method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users in this embodiment; it should be noted that the readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof; the program contained on the readable medium may be transmitted using any appropriate medium, including, but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof. For example, the program code for performing the operation of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and also conventional procedural programming languages ​​such as C language or similar programming languages. The program code may be executed entirely on a user computing device, partially on a user device, as an independent software package, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0138] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0139] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users, characterized in that: The steps include: Based on historical similar day data, the power consumption fluctuation range of large industrial users is obtained, and a large industrial user power consumption prediction error model is established. The sampling probability interval of the error sampling is corrected according to the power consumption fluctuation range of large industrial users to generate a large industrial user power consumption time series scenario; Use K-means method to simplify the time series scenarios and obtain a scenario set including typical scenarios and extreme scenarios; The quality of scene generation is evaluated based on scene coverage, scene width, and correlation similarity indicators; Establish a time series scenario state matrix and a time series scenario power consumption deviation matrix, analyze the deviation between the generated scenario and the predicted power consumption, extract the information matrix based on the state matrix and the deviation matrix scenario, and carry out typical time series scenario extraction.

2. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: The typical time series scene extraction includes: Use the time series scenario electricity consumption deviation matrix to store the difference between the time series scenario electricity consumption of large industrial users and the predicted scenario electricity consumption of large industrial users; Generate a time sequence scene state matrix according to the deviation of each element in the deviation matrix from the threshold; According to the large industrial user electricity consumption timing scenario state matrix, a typical large industrial user electricity consumption timing scenario extraction information matrix is ​​obtained, and then the typical large industrial user electricity consumption timing scenario is extracted.

3. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: The method for obtaining the fluctuation range of electricity consumption of large industrial users based on historical similar day data is as follows: Based on the weather forecast information for the next day, the actual operation data of electricity consumption of large industrial users on days with similar historical weather are collected, and the fluctuation curve of electricity consumption data of large industrial users is analyzed to obtain the fluctuation range of electricity consumption of large industrial users in adjacent time periods on days with similar weather. The calculation formula for the fluctuation of electricity consumption of large industrial users before and after is as follows: ΔP t =P t+T -P t Where: ΔP t is the fluctuation value of electricity consumption of large industrial users at adjacent moments before and after the interval T; T is the time interval; P t is the electricity consumption of large industrial users at time t; The power consumption fluctuation set A of large industrial users is: Where: k t→t+1.dn , k t→t+1.up They are the lower and upper limits of power consumption fluctuation of large industrial users at adjacent sampling moments respectively; t is the sampling time; ΔP t.q is the electricity consumption fluctuation from time t to time t+1 on similar days q; S ander For similar day collection.

4. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: The establishment of the large industrial user electricity consumption prediction error model includes: Based on historical forecast data and actual operation data of large industrial users, a large industrial user electricity consumption forecast error model is established: P t =P t,predict +ξ t Where: P t,predict is the predicted power consumption of large industrial users at time t; t The electricity consumption prediction error for large industrial users.

5. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: The sampling probability interval correction of error sampling according to the power consumption fluctuation range of the large industrial user to generate the power consumption time sequence scenario of the large industrial user includes: The Latin hypercube sampling method is used to extract the initial prediction error values ​​of all time series scenarios longitudinally, and the prediction values ​​are superimposed to calculate the power of all time series scenarios at the first moment; For each time series scenario, the sampling interval is corrected by considering the fluctuation range of electricity consumption of large industrial users at adjacent moments. The quality of scenario generation is evaluated based on the scenario coverage, scenario width, and correlation approximation indicators.

6. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: When evaluating the quality of scene generation, 1) Scene coverage P cover Indicates the coverage of the generated scenario power value to the actual scenario power value: Where: B is a 0-1 variable. If the actual scene power value is between the maximum and minimum values ​​of the generated scene, then B = 1, otherwise B = 0; 2) Scene width P width Reflects the maximum power range between different scenarios in the generated scenario set: Where: P s,t ,P t,t They are respectively generating time series scenario s and the power consumption value at time t; S is the set of generated scenes; 3) Correlation approximation P corr Reflects the approximation degree of correlation between the generated scene and the actual data at any time: P corr =|s real,t,t+d -s s,t,t+d | Where: s real,t,t+d ,s s,t,t+d are the correlation coefficients between the actual scene and the generated scene at any interval d.

7. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: When carrying out typical time series scene extraction, it includes: Use the time series scenario electricity consumption deviation matrix to store the difference between the time series scenario electricity consumption generated by large industrial users and the predicted scenario electricity consumption of large industrial users: ΔP=P sce,out -P′ pre,out Where: ΔP is the power consumption deviation matrix of the time series scenario; P sce,out is the total power consumption matrix of large industrial users in the generated time series scenario; P′ pre,out Because P sce,out The expanded total matrix of electricity consumption forecast for large industrial users; The time sequence scene state matrix includes: Where: Matrix element S is a 0-1 variable, which represents the power consumption deviation state quantity of time series scenario s at time c; Set the threshold value of large industrial users' electricity consumption in the time series scenario deviating from the predicted scenario electricity consumption ξ p,pos ,ξ p,neg , respectively represent the positive deviation threshold and negative deviation threshold of electricity consumption of large industrial users: Where: p is the scene type, ΔP st and S st They are respectively the power consumption deviation matrix elements and deviation state matrix elements of large industrial users in time series scenarios.

8. The method for generating and extracting typical time-series scenarios of electricity consumption of large industrial users according to claim 1 is characterized in that: When extracting typical time series scenarios: The product operation is performed on the large industrial user electricity consumption time series scenario state matrix and the time series scenario state matrix transpose matrix to obtain the typical large industrial user electricity consumption time series scenario extraction information matrix Y: The time period threshold ξ is set, and the large industrial user electricity consumption time series scenario m that satisfies Y>ξ is extracted as the typical time series scenario of large industrial user electricity consumption.

9. A device for generating and extracting typical time-series scenarios of electricity consumption of large industrial users, characterized in that: include: A large industrial user electricity consumption time series scenario acquisition module is used to obtain the electricity consumption fluctuation range of large industrial users based on historical similar day data, establish a large industrial user electricity consumption prediction error model, perform error sampling sampling probability interval correction according to the large industrial user electricity consumption fluctuation range, and generate a large industrial user electricity consumption time series scenario; The scene set acquisition module is used to simplify the time series scenes using the K-means method to obtain a scene set containing typical scenes and extreme scenes; The scene generation quality evaluation module is used to evaluate the quality of scene generation based on scene coverage, scene width, and correlation similarity indicators; The typical time series scenario extraction module is used to establish a time series scenario state matrix and a time series scenario power consumption deviation matrix, analyze the deviation between the generated scenario and the predicted power consumption, extract the information matrix based on the state matrix and the deviation matrix scenario, and carry out typical time series scenario extraction.

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

Images