A demand-side carbon response based anti-manipulation user partitioning method

Abstract

The present application relates to a kind of based on demand side carbon response's anti-manipulation user division method, comprising: S1, set parameter MEI;S2, the MEI value of each user is calculated, form combination (i, MEI i,T ) And ascending arrangement;S3, user i and category k are assigned initial value 1;S4, with MEI i,T As starting value, with 2p length interval, cover its adjacent combination pair in the direction of MEI increases, the user number of maximum MEI is assigned to user j;S5, user i to user j in interval are all divided into category k, and assignment is carried out;S6, repeat S4 and S5 step, until i>N, obtain clustering result C1,…,C λ The present application is by setting marginal carbon emission intensity index, formulates user division method, selects key user to participate carbon response plan, according to the emission mode of user, user is divided into different categories, and then prevent the manipulation behavior that can destroy the effectiveness of carbon response plan.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for anti-manipulation user segmentation based on demand-side carbon response. Background Technology

[0002] The power industry is a major contributor to carbon emissions in my country. As of 2021, the power industry accounted for 40% of the total carbon emissions from all industries. Currently, numerous technologies and policies focus on carbon reduction in the power system. For example, replacing traditional coal-fired generators with renewable energy generators and levying carbon taxes on power generation companies to reduce their energy consumption. However, these technologies and policies primarily focus on carbon metering and carbon reduction on the generation side. Specifically, given the generator type (e.g., coal-fired, natural gas, nuclear, and wind power) and corresponding output at a given moment, the generator's average carbon emission factor can be calculated, thus providing further information on the overall carbon emissions within the power supply area. In practice, the average carbon emission factor for a region is often given based on empirical values. This method is widely adopted by some institutions due to its simplicity, straightforwardness, and low computational complexity. However, this method overlooks a crucial factor: the impact of demand on power system carbon emissions. Because the power system needs to maintain a real-time supply-demand balance, its power generation plan essentially depends on users' electricity demand, making users the fundamental "drivers" of power system carbon emissions. Fluctuations in user-side electricity demand affect power generation plans and generator scheduling, ultimately determining the system's carbon emissions. Because existing methods do not deeply consider the significant role of dynamic changes in demand on the power system's carbon emissions, they cannot dynamically reflect how carbon emissions are generated, transmitted, and ultimately distributed to energy consumers based on user electricity consumption behavior characteristics.

[0003] While leveraging the carbon reduction potential of the demand side is crucial for reducing carbon emissions in the power system, designing demand-side carbon reduction measures still faces significant challenges. These challenges primarily include: 1) Due to the massive number of users on the demand side, and the vast differences in their load curves, accurately reflecting the impact of each user's electricity consumption behavior on system carbon emissions, and precisely measuring each user's carbon emissions, is no easy task. 2) When users are categorized based on load characteristics, and different categories are assigned different probabilities of participating in carbon response (demand response based on carbon reduction), users may have an incentive to deliberately manipulate their load curves to reduce their own likelihood of participating in carbon response. Therefore, designing methods for user segmentation to prevent such manipulation is a problem that needs to be addressed.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a user segmentation method based on demand-side carbon response to prevent manipulation, thereby solving the manipulation problem of users in the context of demand-side carbon response.

[0006] The objective of this invention is achieved through the following technical solution: a user segmentation method based on demand-side carbon response to prevent manipulation, wherein the user segmentation method includes:

[0007] S1. Set parameter MEI: The marginal carbon emission intensity, i.e., MEI, is obtained by multiplying the user's load curve by the carbon emission intensity of the marginal generator;

[0008] S2. Determine the objective function of user manipulation, calculate the MEI value for each user, and then combine it with the user's own ID to form a combination (i, MEI). i,T ), where i represents user number i, i = 1, ..., N;

[0009] S3, for the combination (i, MEI) i,T )Press MEI i,T Sort in ascending order, and initialize user i and category k to 1 respectively;

[0010] S4, with MEI i,T Starting with a value of 2ρ, the adjacent combination pairs are covered in the direction of increasing MEI. Within the interval, the user number with the largest MEI is assigned to user j. ρ represents the threshold value for setting the distance between a user's MEI and its corresponding cluster center.

[0011] S5. All users i to j within the interval are assigned to category k, and k+1 is assigned to k, and j+1 is assigned to j;

[0012] S6. Repeat steps S4 and S5 until i > N, where N represents the number of users, i.e., all users are covered. Assign k to λ to obtain the clustering results C1,…,C λ .

[0013] The S1 step specifically includes:

[0014] The power system has N electricity users. The carbon emissions at different times are generated by the corresponding generators, and the total carbon emissions at different times are... The total carbon emissions for the time period [0, T] are then obtained as follows: Among them, e t p represents the total carbon emissions of the power system at time t. t q t and w t These are parameters that are specific to a particular system. L is the electricity consumption of user i at time t. t Let L be the total power consumption of the system, where L = (L0, ..., L...). T );

[0015] Set the user's load curve for the time period [0, T] as follows: in, Let ||·||1| be the vector form of user i's electricity consumption, ||·||1| be the l1 norm, and Δt be the time interval. Then, the formula for obtaining MEI is: in, This represents the overall offset of the load curve while maintaining the same shape as the user's electricity load curve. This indicates the carbon emission intensity of the marginal generator;

[0016] The total carbon emissions of a single user within the time period [0,T] are calculated as CE. i,T =MEI i,T ·||l i ||1, where CE i,T It represents the total carbon emissions of a single user.

[0017] The objective function for determining user manipulation includes:

[0018] Set the cluster center of category n to p. n And the corresponding load curve is Let user i belong to category u(i), represented by the scalar μ. i,n The effort value of user i to leave category u(i) to another category n ≠ u(i) through manipulation is measured. After manipulation, the user's MEI changes as follows: and

[0019] Use CR i Indicates satisfaction The smallest scalar μ i,n By solving the optimization problem, we can obtain p u(i) Represents the cluster center μ i,n The corresponding MEI;

[0020] When CR i If the threshold is less than θ, then user i can successfully manipulate the system. θ represents the threshold value to ensure CR. i If ≥θ, then the objective function of the user operation is Θ i This indicates the category that user i can successfully manipulate.

[0021] This invention has the following advantages: a user segmentation method based on demand-side carbon response to prevent manipulation, setting a marginal carbon intensity (MEI) index, which can represent the impact of user electricity consumption behavior on system carbon emissions; in addition, based on this index, and by simultaneously considering user load curve information and system operating status, it provides accurate measurement of user carbon emissions; for the massive number of users on the demand side, a user segmentation scheme is formulated, selecting key users to participate in the carbon response plan. The proposed scheme divides users into different categories according to their emission patterns and can prevent manipulation behaviors that may undermine the effectiveness of the carbon response plan. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown herein can generally be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of this application provided below is not intended to limit the scope of protection of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The present invention will be further described below.

[0023] This invention specifically relates to a method for anti-manipulation user segmentation based on demand-side carbon response, which includes the following:

[0024] 1. Design carbon emission targets: Marginal carbon intensity (MEI);

[0025] A power system is configured with N electricity users. The carbon emissions of the system at different times are generated by the corresponding generators, and the total carbon emissions at different times satisfy the following relationship:

[0026]

[0027]

[0028] Among them, e t p represents the total carbon emissions of the power system at time t. t q t and w t These are parameters that are specific to a particular system. L is the electricity consumption of user i at time t. t Let be the total electricity consumption of the system. Then the total carbon emissions during the time period [0, T] can be written as:

[0029]

[0030] Where L=(L0,…,L T ).

[0031] Additionally, the load curve for users in the time period [0, T] is set as follows:

[0032]

[0033] in, Let ||·||1 be the vector form of user i's electricity consumption, ||·||1 be the l1 norm, and Δt be the time interval.

[0034] The definition of MEI is:

[0035]

[0036] in, This represents the overall offset of the load profile while maintaining the same shape as the user's electricity load curve. This indicates the carbon emission intensity of the marginal generator.

[0037] The marginal energy intensity (MEI) reflects the impact of user electricity consumption behavior on system carbon emissions by taking into account the heterogeneity of electricity load distribution. Specifically, the MEI is obtained by multiplying a user's load curve by the carbon emission intensity of the marginal generator. A higher MEI value indicates that the user consumes more electricity during the period when high-emission generators (e.g., fossil fuel generators) are operating, resulting in a greater impact on system carbon emissions. Conversely, a lower MEI value means that the user consumes more electricity during the period when low-emission generators (e.g., wind turbines or solar generators) are operating, resulting in a smaller impact on system carbon emissions.

[0038] Furthermore, according to the MEI definition, the total carbon emissions of a single user within the time period [0,T] are:

[0039] CE i,T =MEI i,T ·||l i ||1

[0040] Among them, CE i,T It represents the total carbon emissions of a single user.

[0041] 2. User segmentation to prevent manipulation:

[0042] (1) Demand-side carbon response:

[0043] Demand-side carbon response refers to the reduction of overall carbon emissions from the power system by users fully utilizing and exploring demand-side resources or changing their electricity consumption patterns. User behavior is set when participating in carbon response:

[0044]

[0045] The user's objective is to minimize their own carbon emissions. τ represents the user's load adjustment ratio. This is the original, raw electricity consumption. It is set that the user's total electricity consumption is fixed within the time period [0, T], and the load adjustment amount is proportional to the original electricity consumption.

[0046] (2) User manipulation behavior:

[0047] Given the large number of demand-side users, all users are divided into different categories, each with varying degrees of likelihood of being selected to participate in the carbon response program. Therefore, it is necessary to prevent users from manipulating their load curves to enter groups they do not belong to, thereby enhancing the emission reduction effect of the carbon response.

[0048] When segmenting users, one-dimensional clustering is performed based on their MEI (Mean Interval Intake) index. This is because users with high MEI are more likely to be identified as key users of system carbon emissions compared to those with low MEI, as they have a greater impact on system carbon emissions. However, in this scenario, some users strategically manipulate their load curves to lower their MEI values, thereby reducing their likelihood of participating in carbon response. This is because participating in carbon response requires users to adjust their electricity consumption behavior, which often involves changes in user habits, and many users may be unwilling to accept such changes. Obviously, when key users reduce their likelihood of participating in carbon response through manipulation, the overall effectiveness of the emissions reduction plan will be compromised. To further analyze this type of behavior, the clustering results, cluster centers, and emissions reduction strategies (i.e., the likelihood of users participating in carbon response) for each group are public information, making them available to all users.

[0049] (3) User operation process:

[0050] First, let p be the cluster center of category n. n And the corresponding load curve is User i belongs to category u(i). The effort user i makes to leave category u(i) to another category n≠u(i) through manipulation is measured. This effort value can be expressed by the scalar μ. i,n To measure this. Thus, after manipulation, the user's MEI changes as follows:

[0051]

[0052] At the same time, the following conditions also apply:

[0053]

[0054] Where, p nRepresents the MEI,p corresponding to cluster center n. u(i) Represents the cluster center μ i,n The corresponding MEI. The smallest μ that satisfies the above equation. i,n Represented as CR i It can be obtained by solving the following optimization problem:

[0055]

[0056] If user i can successfully manipulate, then condition CR i <θ needs to be satisfied, where θ is a preset threshold. Since users with a large MEI are more likely to participate in the power system carbon response, the following user manipulation objectives are set to reduce this likelihood:

[0057]

[0058] Where, Θ i Let i be the category that user i can successfully manipulate. The objective function can be rewritten as:

[0059]

[0060] (4) User anti-manipulation classification:

[0061] By giving the above equation b i,n Set upper and lower limits to restrict user manipulation behavior; specifically, if the user's MEI is close enough to p... u(i) Therefore, this partitioning method can effectively prevent user manipulation.

[0062] If a clustering algorithm can ensure So If a user manipulates their behavior to be classified into the nth category, then the following condition holds:

[0063]

[0064] Therefore, for any class n∈Θ i Then we have:

[0065]

[0066] Therefore, if the distance between a user's MEI and its corresponding cluster center does not exceed ρ, then the cluster center of this cluster center and the cluster centers of other categories that the user may manipulate can be restricted.

[0067] (5) Solving for anti-manipulation user segmentation:

[0068] First, calculate the MEI value for each user according to the aforementioned definition, and then combine it with the user's own ID to form a combination (i, MEI).i,T For the combination (i, MEI) i,T )Press MEI i,T Sort in ascending order. Next, start with the smallest combination (i, MEI). i,T ) begins by covering adjacent (i, MEI) regions using an interval of length 2ρ. i,T This process involves combining users to form different categories. This process is repeated until all users are covered and the clustering is complete. The algorithm has a time complexity of O(nlogn) and minimizes the manipulation space of users under the carbon response of the power system; specifically, it includes the following:

[0069] Step 1: Calculate the MEI value for each user, and then combine it with the user's own ID to form a combination (i, MEI). i,T ), where i represents user number i, i = 1, ..., N;

[0070] Step 2: For the combination (i, MEI) i,T )Press MEI i,T Sort in ascending order, and initialize user i and category k to 1 respectively;

[0071] Step 3, using MEI i,T Starting with a value of 2ρ, the adjacent pairs of combinations are covered in the direction of increasing MEI. Within the interval, the user number with the largest MEI is assigned to user j.

[0072] Step 4: Assign all users i to j within the interval to category k, and assign k+1 to k and j+1 to j;

[0073] Step 5: Repeat steps 3 and 4 until i > N, where N represents the number of users, i.e., all users are covered. Assign k to λ to obtain the clustering results C1,…,C λ .

[0074] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for anti-manipulation user segmentation based on demand-side carbon response, characterized in that: The user segmentation method includes: S1. Set parameter MEI: The marginal carbon emission intensity, i.e., MEI, is obtained by multiplying the user's load curve by the carbon emission intensity of the marginal generator; S2. Determine the objective function of user manipulation, calculate the MEI value for each user, and then combine it with the user's own ID to form a combination. , Indicates number users, ; S3, Combination according to Sort in ascending order and then sort the users separately. and categories The initial value is set to 1; S4, with As the starting value, with The length is such that the interval covers adjacent pairs in the direction of increasing MEI. Within the interval, the user ID with the largest MEI is assigned to the user. , This indicates setting a threshold for the distance between a user's MEI and its corresponding cluster center; S5, users within the interval To users All are classified into categories In, and will Assign to and will Assign to ; S6. Repeat steps S4 and S5 until... , This indicates the number of users, meaning all users are covered. Assign to Obtain clustering results ; The S1 step specifically includes: The power system has N electricity users. The carbon emissions at different times are generated by the corresponding generators, and the total carbon emissions at different times are... Then the time period is obtained. Total carbon emissions are ,in, express Total carbon emissions of the power system at any time , as well as These are parameters that are specific to a particular system. Electricity users exist Electricity consumption at all times This represents the total power consumption of the system. ; Set user time period The load curve is ,in, User Vector form of electricity consumption yes Norm, If it is a time interval, then the formula for obtaining MEI is: ,in, This represents the overall offset of the load curve while maintaining the same shape as the user's electricity load curve. This indicates the carbon emission intensity of the marginal generator; Get a single user in a time period The total carbon emissions within the region are ,in, It is the total carbon emissions of a single user; The user operation process includes: First, define the categories. The cluster centers are And the corresponding load curve is ,user The category it belongs to is Measure users Leave the category by manipulating behavior To another category The effort made, and the value of that effort, can be expressed as a scalar. To measure the change in a user's MEI after manipulation, the following is observed: ， At the same time, the following conditions also apply: ， in, Representing cluster centers corresponding , Representing cluster centers corresponding To satisfy the minimum of Represented as This is obtained by solving the following optimization problem: ， ， If user If successful manipulation is possible, then the conditions are... The need to be met The preset threshold; Since users with larger MEIs are more likely to participate in the carbon response of the power system, the objectives of user manipulation are as follows: ， in, User For categories that can be successfully manipulated, the objective function is rewritten as: ； User anti-manipulation measures include: By giving Set upper and lower limits to restrict user manipulation behavior; if the user's MEI is close enough... Therefore, this partitioning method can effectively prevent user manipulation if a clustering algorithm can ensure... ,So If a user manipulates their behavior to get themselves classified as a third party... In each category, the following condition holds true: ， Therefore, for any category Then we have: ， Therefore, if the distance between a user's MEI and its corresponding cluster center does not exceed This limits the cluster centers of this cluster and the cluster centers of other categories that users may be able to manipulate.

Images