A user carbon profiling method based on electricity usage behavior

Abstract

The present invention discloses a method for user carbon profiling based on electricity usage behavior, comprising the following steps: user data preprocessing; constructing a multi-dimensional user feature label model, establishing a user multi-source feature label system from three dimensions: user electricity usage characteristics, user low-carbon electricity usage behavior, and user electricity carbon production and consumption characteristics; calculating comprehensive indicators of various characteristic indicators according to the divided time periods; visualizing the results of the carbon profiling according to the user clusters, and using the profiling results as the basis for the power company to formulate user demand-side response policies and low-carbon regulation. The present invention makes the implicit characteristics of users explicit by analyzing the refined characteristics of users, that is, studying the user carbon profiling, and helps power companies formulate reasonable low-carbon regulation measures based on the low-carbon electricity behavior characteristics of different residential users from the perspective of the production and consumption characteristics of users' electricity carbon emissions.

Description

Technical Field

[0001] The present invention relates to a user carbon profiling method based on electric energy usage behavior, belonging to the technical field of power supply and distribution. Background Art

[0002] With my country's dual carbon goals of achieving peak carbon emissions and achieving carbon neutrality, energy conservation and carbon reduction have become a hot topic. In recent years, the proportion of residential electricity consumption in my country has continued to rise, and the carbon dioxide generated by daily electricity use cannot be ignored. Analyzing residential electricity usage behavior and adjusting it is the most direct and effective way to reduce carbon emissions from residential electricity use.

[0003] With the increasing adoption and functional upgrades of smart grids and the development of big data technologies, it's now possible to collect multidimensional user data and create behavioral profiles. Carbon profiling, in particular, describes the comprehensive carbon characteristics of a user's electricity usage. Power companies can optimize residential demand-side response models by adjusting marketing methods and electricity pricing policies, achieving precise regulation of user electricity usage and ultimately reducing energy consumption and emissions. Summary of the Invention

[0004] The purpose of the present invention is to provide a user carbon profiling method based on electricity usage behavior, with the goal of helping power companies to formulate reasonable low-carbon regulatory measures. The user's attributes are labeled and abstracted from three aspects: the user's electricity usage characteristics, the user's low-carbon electricity usage behavior, and the user's electricity production and consumption carbon characteristics (the user's electricity production and consumption carbon characteristics are the comprehensive characteristics of the user's electricity in the process of generating and consuming carbon emissions). It is proposed to calculate the attribute labels of users in each dimension and improve the k-means clustering algorithm for application in the user's comprehensive carbon profiling method.

[0005] The purpose of the present invention is achieved through the following technical solutions:

[0006] A user carbon profiling method based on electricity usage behavior includes the following steps:

[0007] S1. User data preprocessing, including: collecting and filtering multi-dimensional user behavior data, and removing error values and wrong user data;

[0008] S2. Build a multi-dimensional user feature labeling model, establishing a multi-source user feature labeling system based on three dimensions: user electricity consumption characteristics, user low-carbon electricity consumption behavior, and user electricity production and carbon consumption characteristics;

[0009] S3. Calculate comprehensive indicators of various characteristic indicators based on the divided time periods. The comprehensive indicators include: user electricity consumption characteristic labels, user low-carbon electricity consumption behavior labels, and user electricity consumption carbon production and consumption characteristic labels. Different residential users are divided into clusters with different attributes based on the user electricity consumption characteristic labels, user low-carbon electricity consumption behavior labels, and user electricity consumption carbon production and consumption characteristic labels.

[0010] S4. Visualize the results of the carbon profiling based on user clusters, and use the profiling results as a basis for power companies to formulate user demand-side response policies and low-carbon regulation.

[0011] The present invention further includes the following preferred embodiments.

[0012] In the aforementioned user carbon profiling method based on electricity usage behavior, in step 1, the user's multi-dimensional behavior data includes:

[0013] Based on the multi-dimensional electricity usage behavior data of users collected by non-home terminals, the electricity consumption data of various loads identified by non-home terminals are collected with a sampling time of 15 minutes and 96 points of electricity consumption data of various loads of users on a daily basis;

[0014] The loads collected by non-home terminals include refrigerators, long-term electric heating, short-term electric heating, variable-frequency air conditioners, non-variable-frequency air conditioners, induction cookers, rice cookers, electric water heaters, air conditioners, microwave ovens, and washing machines.

[0015] Based on the user electricity bill data collected by the marketing system, on a daily basis, including: the user's total electricity bill, the electricity bill generated during the user's peak and off-peak hours;

[0016] User online behavior statistics collected through the online business hall, mobile app, and power grid customer service phone number include: the number of times users set their home air conditioners to energy-saving temperatures, the number of times users checked their electricity bills, the number of times users checked their electricity bills by logging into the mobile app, the number of times users checked their electricity bills by logging into the online business hall webpage, and the number of times users checked their electricity bills by calling the power supply service hotline 95598;

[0017] In the aforementioned user carbon profiling method based on electricity usage behavior, in step 2, the multi-dimensional user feature label includes the user electricity usage characteristic label, the user low-carbon electricity usage behavior label, and the user electricity production and consumption carbon characteristic label.

[0018] In step 2, the user electricity usage characteristic label is intended to characterize the user's electricity usage. By studying the user's load characteristics and electricity usage, the user's regulation potential is explored. The user electricity usage characteristic label includes: power characteristic label, load characteristic label, and electrical appliance composition characteristic label. Among them:

[0019] The power characteristic label L E , defined as the ratio of the user's electricity consumption to the average household electricity consumption during the calculation period, as shown in the following formula:

[0020]

[0021] Where W i is the power consumed by the user during the computing cycle, W av is the average power consumption of all users within the statistical range, and n is the number of users within the statistical range;

[0022] The load characteristic label is defined as the maximum load power, average load rate, maximum load utilization hours, and peak-to-valley difference rate within the analysis period, which are expressed as:

[0023] Maximum load power:

[0024] L max-Ti =P max-Ti (2)

[0025] Average load factor:

[0026]

[0027] Maximum load utilization hours:

[0028]

[0029] Peak-to-valley rate:

[0030]

[0031] Where, P Ti is the power of the i-th type adjustable load, P max-Ti is the maximum power in the statistical Ti period, P av-Ti is the average power in the Ti period, P max is the maximum power within the statistical time, P min is the minimum power within the statistical time;

[0032] The electrical appliance characteristic label is defined as the ratio of the total power of the load that can respond to the demand side to the reference power, as shown in the following formula:

[0033] L app =(∑k i P Ti +∑k j P zj ) / 1000 (6)

[0034] Where, P Ti is the power of the i-th type adjustable load, P zjis the power of the jth type of load that can be reduced, k represents the number of each type of load, k i is the number of adjustable loads of type i, k j is the number of the jth type of reducible loads; where adjustable loads refer to loads that do not stop running during user use but whose power can be adjusted, and reducible loads refer to loads that can be interrupted during user use;

[0035] In step 2, the user's low-carbon electricity consumption behavior label is intended to evaluate the user's ability to interact with the power grid from a subjective perspective, thereby reflecting the difficulty of the power company in achieving the goal of low-carbon emission reduction by regulating the user's electricity consumption behavior. The user's low-carbon electricity consumption behavior label includes: a peak-valley electricity price sensitivity estimation label, an electricity consumption tendency estimation label, and a user's low-carbon electricity use awareness estimation label;

[0036] The peak-valley electricity price sensitivity estimation label is defined as the proportion of peak-valley electricity charges and the change in electricity consumption at the turning point of peak-valley electricity prices, which are expressed as:

[0037] Peak and valley electricity charges:

[0038]

[0039] Where C f represents the peak electricity charge during the calculation period, C g Indicates the off-peak electricity price within the calculation period;

[0040] Changes in electricity consumption at the turning point of peak-valley electricity prices:

[0041]

[0042] Where W i (t n +1) represents the electricity consumption one hour after the peak-valley electricity price turns in the evening of day i, W i (t n -1) represents the amount of electricity consumed one hour before the peak-valley electricity price turns on the morning of day i, T is the calculation period, and L cs Reflects the user's sensitivity to electricity prices;

[0043] The power consumption tendency estimation tag is defined as the valley power coefficient, which is expressed as:

[0044]

[0045] Where, L v Indicates the user's electricity consumption during off-peak hours, L z Indicates the total electricity consumption of the user, and the valley electricity coefficient reflects the user's tendency to consume electricity during off-peak hours;

[0046] The estimation of users' awareness of low-carbon electricity use is based on data collected and analyzed by the power system marketing department and other related network systems, including the number of times users set their home air conditioners to energy-saving temperatures, the number of times users checked their electricity bills, the number of times users checked their electricity bills by logging into the mobile app, the number of times users checked their electricity bills by logging into the online business hall website, and the number of times users checked their electricity bills by calling 95598:

[0047] L es =N esc +N ce (10)

[0048] N ce =N App +N net +N 598 (11)

[0049] Where N esc The number of times the user sets the home air conditioner to the energy-saving temperature, N ce The number of times the user checks the electricity usage, N App The number of times the user checks the electricity bill by logging into the mobile APP, N net The number of times users check their electricity bills by logging into the online business hall, N 598 The number of times users checked their electricity bills by dialing 95598, L es represents the low-carbon electricity usage awareness estimate, L es The larger the value, the more concerned the user is about his or her low-carbon electricity consumption behavior and the better the low-carbon electricity consumption awareness.

[0050] In step 2, the user's electricity production and consumption carbon characteristics label is intended to characterize the potential of residential users to reduce carbon dioxide emissions by responding to the relevant electricity energy conservation and carbon reduction policies formulated by the power grid and rationally allocating and using electricity. The user's electricity production and consumption carbon characteristics label includes: the user's electricity clean energy ratio label, the green appliance energy consumption ratio label, the user's electricity carbon emission characteristics label, the user's carbon emission regulation coefficient label, and the user's effective alternative electricity label;

[0051] The user's electricity clean energy ratio label L cl , defined as the proportion of clean energy in the electricity used by the user during the calculation period, as shown in the following formula:

[0052]

[0053] Where, L p The amount of photovoltaic energy used in the electricity used by users, L t The amount of thermal power used in the electricity consumed by users, L clIt reflects the proportion of clean energy in the daily electricity used by users, and its value is positively correlated with the user's carbon reduction potential;

[0054] The green appliance energy consumption ratio label L g , defined as the proportion of the electric energy consumption of green household appliances used by the user to the total electric energy consumption of the user during the calculation period, as shown in the following formula:

[0055]

[0056] Where, L ge Monthly consumption of green appliances by users, L t The amount of thermal power used in the electricity consumed by users, L p The amount of photovoltaic energy used in the electricity used by users and the proportion of energy consumption of green appliances by users can reflect the potential for carbon reduction in users' electricity.

[0057] The user electricity carbon emission characteristic label L ce , which is defined as the ratio of the carbon emissions generated by a user’s personal electricity consumption to the average carbon emissions generated by electricity consumption by users within the statistical sample range during the calculation period, as shown in the following formula:

[0058]

[0059] L tc =L t ×a t (15)

[0060] L avc =L av ×a t (16)

[0061] Where, L tc Carbon emissions generated by electricity consumption, L av is the carbon emissions generated by the average electricity consumption of users in the community, L t The amount of thermal power used in the electricity consumed by users, L av is the average electricity consumption per user in the sample community, a t The carbon emission coefficient corresponding to electric energy is 0.96kg / kWh. The user's electricity carbon emission characteristic label is the most direct expression of the user's potential to reduce carbon dioxide emissions through electricity regulation.

[0062] The user carbon emission control coefficient label L rce , which is defined as the proportion of the electricity consumed by the user during the calculation period that is left over from photovoltaic power after the base load is met, which is the proportion of the demand response load, as shown in the following formula:

[0063]

[0064] Where, L b is the user's basic load, P Ti is the power of the i-th type adjustable load, P zj is the power of class j that can be reduced, K i is the number of adjustable loads of type i, K j is the amount of load that can be reduced in category j, and the user carbon emission regulation coefficient reflects the potential of the clean energy used by the user to participate in demand-side response;

[0065] The user's effective replacement power label E es , defined as the effective amount of electricity replaced by electric energy during the calculation period, as shown in the following formula:

[0066]

[0067] Where, α rem M is the proportion of clean energy electricity consumption of users in the calculation period, es is the number of electric energy replacement devices in the user's household, E esm The label for the user's effective electricity replacement is the amount of electricity replaced by electric energy replacement equipment during the calculation period. It indicates the amount of electricity replaced by convenient, safe, low-carbon and clean electricity in the user's household energy structure to replace traditional primary energy sources such as coal, oil and natural gas, objectively reflecting the user's low-carbon emission reduction potential.

[0068] In the aforementioned user carbon profiling method based on electricity usage behavior, in step 3, the method for calculating the comprehensive index of various characteristic indicators includes: data standardization processing, analysis of various comprehensive index values, K-means clustering algorithm, and user cluster analysis;

[0069] The data standardization process includes:

[0070] (1) Establish a comprehensive control matrix for the same period, which includes three categories of 15-dimensional labels: 6-dimensional user electricity consumption characteristic labels L1-L6, 4-dimensional user low-carbon electricity consumption behavior labels L7-L 10 , user electricity production and consumption carbon characteristics label 5-dimensional L 11 -L 15 , that is, the data sample of N users is expressed as:

[0071]

[0072] (2) All users’ data of the same category are represented as column vectors:

[0073] {L i (1), L i (2), …L i (15)}

[0074] (3) Perform data cleaning and standardize the data using min-max normalization:

[0075]

[0076] Where, L(j) min represents the minimum value of L(j), L(j) max represents the maximum value of L(j);

[0077] The various comprehensive index value analyses include:

[0078] (1) User power characteristics label L rp

[0079] The user power consumption characteristic label includes user power consumption characteristic labels L1-L6, a total of 6-dimensional data. The comprehensive label value L of the power consumption characteristic is obtained using the following formula: rp , where w j is the weight of each dimension data:

[0080]

[0081] (2) Power regulation willingness label L rw

[0082] The power regulation intention label includes the user power consumption characteristic label L7-L 10 , a total of 4-dimensional data, use the following formula to calculate the comprehensive label value L of the power consumption characteristics rw , where w j is the weight of each dimension data:

[0083]

[0084] (3) Electricity Carbon Reduction Potential Label L crp

[0085] The electricity carbon reduction potential label includes the user electricity characteristics label L 11 -L 15 , a total of 5-dimensional data, use the following formula to calculate the comprehensive label value L of the power consumption characteristics crp , where w j is the weight of each dimension data:

[0086]

[0087] The weights of each dimension are determined by the entropy weight method. The weights of L1-L6 are: 0.23, 0.25, 0.05, 0.06, 0.03, 0.39, and L7-L 10 The weights are: 0.26, 0.14, 0.14, 0.46, L 11 -L15 The weights are: 0.01, 0.19, 0.09, 0.01, 0.70.

[0088] Analyze and determine the user's cluster based on the K-Means clustering algorithm, where:

[0089] For user power consumption characteristics label L rp (i) Based on the Euclidean distance K-means clustering algorithm, cluster the clusters into m1 categories and find the cluster center points L of each category. rp (k);

[0090]

[0091] For the power regulation willingness label L rw (i) Based on the Euclidean distance K-means clustering algorithm, clustering is done into m2 categories, and the cluster center points L of each category are obtained. rw (k);

[0092]

[0093] Carbon reduction potential label for electricity crp (i) Based on the Euclidean distance K-means clustering algorithm, the clustering idea is clustered into m3 categories, and the cluster center point L of each category is obtained. crp (k);

[0094]

[0095] According to L rp (i) L rw (i) L crp (i) The cluster center points of the three types of data are combined to form m1×m2×m3 three-dimensional surface center points, corresponding to m1×m2×m3 clusters respectively;

[0096] Among them, m1=m2=m3=3.

[0097] In the aforementioned user carbon profiling method based on electricity usage behavior, the visualization of the carbon profiling results described in step 4 refers to the visualization of the calculated user electricity usage characteristic labels, user low-carbon electricity usage behavior labels, user low-carbon electricity production and consumption carbon characteristic labels, and user clusters determined by k-means clustering.

[0098] Compared with the prior art, the present invention has the following beneficial technical effects:

[0099] The present invention makes the implicit characteristics of users explicit by analyzing their refined characteristics, namely studying their carbon profiles, and helps power companies formulate reasonable low-carbon regulatory measures based on the production and consumption characteristics of users' electricity carbon emissions and the low-carbon electricity consumption behavior characteristics of different residential users. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 This is a basic flow chart of the user carbon profiling method based on electricity usage behavior of the present invention;

[0101] Figure 2 It is the overall control analysis flow chart of the present invention;

[0102] Figure 3 It is a visual presentation diagram of the present invention. DETAILED DESCRIPTION

[0103] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0104] according to Figure 1 The basic process of the user carbon portrait method is shown.

[0105] First, users' electricity usage behavior data is collected and filtered. The collected data includes: multi-dimensional user electricity usage behavior data collected by non-home terminals, and electricity consumption data of various loads identified by non-home terminals. With a sampling interval of 15 minutes, 96 points of electricity consumption data of various load types are collected on a daily basis.

[0106] The loads collected by non-home terminals include refrigerators, long-term electric heating, short-term electric heating, variable-frequency air conditioners, non-variable-frequency air conditioners, induction cookers, rice cookers, electric water heaters, air conditioners, microwave ovens, and washing machines.

[0107] Based on the user electricity bill data collected by the marketing system, on a daily basis, including: the user's total electricity bill, the electricity bill generated during the user's peak and off-peak hours;

[0108] User online behavior statistics collected through the online business hall, mobile app, and power grid customer service phone number include: the number of times users set their home air conditioners to energy-saving temperatures, the number of times users checked their electricity bills, the number of times users checked their electricity bills by logging into the mobile app, the number of times users checked their electricity bills by logging into the online business hall webpage, and the number of times users checked their electricity bills by calling the power supply service hotline 95598;

[0109] Based on the existing user electricity usage data, combined with the profiling purpose and user electricity usage attributes, a labeling system for user carbon profiling is designed. The labeling system includes: user electricity usage characteristic labels, user low-carbon electricity usage behavior labels, and user electricity production and consumption carbon characteristic labels. At the same time, specific calculation methods for each dimension of labels are designed.

[0110] The user's electricity characteristic labels include: power characteristic labels, load characteristic labels, and electrical appliance characteristic labels; B , defined as the ratio of the user's electricity consumption to the average household electricity consumption during the calculation period, as shown in the following formula:

[0111]

[0112] Where W i is the power consumed by the user during the computing cycle, W av is the average power consumption of all users within the statistical range, and n is the number of users within the statistical range;

[0113] The load characteristic label is defined as the maximum load power, average load rate, maximum load utilization hours, and peak-to-valley difference rate within the analysis period, which are expressed as:

[0114] Maximum load power:

[0115] L max-Ti =P max-Ti (2)

[0116] Average load factor:

[0117]

[0118] Maximum load utilization hours:

[0119]

[0120] Peak-to-valley rate:

[0121]

[0122] Where, P max-Ti is the maximum power in the statistical Ti period, P av-Ti is the average power in the Ti period, P max is the maximum power within the statistical time, P min is the minimum power within the statistical time;

[0123] The appliance component characteristic label is defined as the ratio of the total power of the load that can respond to the demand side to the reference power, as shown in the following formula:

[0124] L app =(∑k iP Ti +∑k j P zj ) / 1000 (6)

[0125] Where, P Ti is the power of the i-th type adjustable load, P zj is the power of the jth type of load that can be reduced, k represents the number of each type of load, k i is the number of adjustable loads of type i, k j is the number of the jth type of reducible loads. Adjustable loads refer to loads that do not stop running during user use but whose power can be adjusted, and reducible loads refer to loads that can be interrupted during user use.

[0126] The peak-valley electricity price sensitivity estimation label is defined as the proportion of peak-valley electricity charges and the change in electricity consumption at the turning point of peak-valley electricity prices, which are expressed as:

[0127] Peak and valley electricity charges:

[0128]

[0129] Where C f represents the peak electricity charge during the calculation period, C g Indicates the off-peak electricity price within the calculation period;

[0130] Changes in electricity consumption at the turning point of peak-valley electricity prices:

[0131]

[0132] Where W i (t n +1) represents the electricity consumption one hour after the peak-valley electricity price turns in the evening of day i, W i (t n -1) represents the amount of electricity consumed one hour before the peak-valley electricity price turns on the morning of day i, T is the calculation period, and L cs It reflects the user's sensitivity to electricity prices.

[0133] The power consumption tendency estimation label is defined as the valley power coefficient, which is expressed as:

[0134]

[0135] Where, L v Indicates the user's electricity consumption during off-peak hours, L z It represents the total electricity consumption of the user, and the valley electricity coefficient reflects the user's electricity consumption tendency during off-peak hours.

[0136] The estimated labels for users' low-carbon electricity usage awareness include: the number of times users set their home air conditioners to energy-saving temperatures, the number of times users checked their electricity bills, the number of times users checked their electricity bills by logging into the mobile app, the number of times users checked their electricity bills by logging into the online business hall website, and the number of times users checked their electricity bills by calling 95598.

[0137] L es =N esc +N ce (10)

[0138] N ce =N App +N net +N 598 (11)

[0139] Where N esc The number of times the user sets the home air conditioner to the energy-saving temperature, N ce The number of times the user checks the electricity usage, N App The number of times the user checks the electricity bill by logging into the mobile APP, N net The number of times users check their electricity bills by logging into the online business hall, N 598 The number of times a user checked their electricity bill by calling 95598. es represents the low-carbon electricity usage awareness estimate, L es The larger the value, the more concerned the user is about his or her low-carbon electricity consumption behavior and the better the low-carbon electricity consumption awareness.

[0140] User electricity production and consumption carbon characteristics labels include: user electricity clean energy ratio label, green appliance energy consumption ratio label, user electricity carbon emission characteristics label, user carbon emission regulation coefficient label, user effective alternative power label;

[0141] User electricity clean energy ratio label L cl , defined as the proportion of clean energy in the electricity used by the user during the calculation period, as shown in the following formula:

[0142]

[0143] Where, L p The amount of photovoltaic energy used in the electricity used by users, L t The amount of thermal power used in the electricity consumed by users, L cl It reflects the proportion of clean energy in the user's daily electricity use, and its value is positively correlated with the user's carbon reduction potential.

[0144] Green appliances energy consumption ratio label L g, defined as the proportion of the electric energy consumption of green household appliances used by the user to the total electric energy consumption of the user during the calculation period, as shown in the following formula:

[0145]

[0146] Where, L ge Monthly consumption of green appliances by users, L t The amount of thermal power used in the electricity consumed by users, L p The amount of photovoltaic energy used in the electricity used by users and the proportion of energy consumption of green appliances by users can reflect the potential for carbon reduction in users' electricity.

[0147] User electricity carbon emission characteristic label L ce , which is defined as the ratio of the carbon emissions generated by a user’s personal electricity consumption to the average carbon emissions generated by electricity consumption by users within the statistical sample range during the calculation period, as shown in the following formula:

[0148]

[0149] L tc =L t ×a t (15)

[0150] L avc =L av ×a t (16)

[0151] Where, L tc Carbon emissions generated by electricity consumption, L av is the carbon emissions generated by the average electricity consumption of users in the community, L t The amount of thermal power used in the electricity consumed by users, L av is the average electricity consumption per user in the sample community, a t The carbon emission coefficient for electricity is 0.96 kg / kWh. The user's electricity carbon emission characteristic label is the most direct indicator of the user's potential to reduce carbon dioxide emissions through electricity regulation.

[0152] User carbon emission control coefficient label L rce , which is defined as the proportion of the electricity consumed by the user during the calculation period that is left over from photovoltaic power after the base load is met, which is the proportion of the demand response load, as shown in the following formula:

[0153]

[0154] Where, L b is the user's basic load, P Ti is the power of the i-th type adjustable load, P zjis the power of class j that can be reduced, K i is the number of adjustable loads of type i, K j is the amount of load that can be reduced in type j, and the user carbon emission regulation coefficient reflects the potential of the clean energy used by the user to participate in demand-side response.

[0155] User effective replacement power label E es , defined as the effective amount of electricity replaced by electric energy during the calculation period, as shown in the following formula:

[0156]

[0157] Where, α rem M is the proportion of clean energy electricity consumption of users in the calculation period, es is the number of electric energy replacement devices in the user's household, E esm The label for the user's effective electricity replacement is the amount of electricity replaced by electric energy replacement equipment during the calculation period. It indicates the amount of electricity replaced by convenient, safe, low-carbon and clean electricity in the user's household energy structure to replace traditional primary energy sources such as coal, oil and natural gas, objectively reflecting the user's low-carbon emission reduction potential.

[0158] After determining the specific calculation method of each dimension label system, such as Figure 2 As shown in the figure, the sub-labels in each label system are processed to obtain the labels of each dimension, and then according to Figure 2 The steps shown provide a comprehensive presentation of the carbon image.

[0159] First, the data of each dimensional sub-label is standardized, and the calculated label data of each dimension is represented by a data sample matrix. A comprehensive control matrix for the same period is established, and then the same data is represented by a vector. Then, data cleaning is performed to eliminate data that does not meet the requirements, and finally the cleaned data is standardized.

[0160] (1) Establish a comprehensive control matrix for the same period, which includes three categories of 14-dimensional labels: 6-dimensional user electricity consumption characteristic labels L1-L6, 4-dimensional user low-carbon electricity consumption behavior labels L7-L 10 , user electricity production and consumption carbon characteristics label 5-dimensional L 11 -L 15 , that is, the data sample of N users is expressed as:

[0161]

[0162] (2) All users’ data of the same category are represented as column vectors:

[0163] {L i (1), L i (2), …L i (15)}

[0164] (3) Perform data cleaning and standardize the data using min-max normalization:

[0165]

[0166] Where, L(j) min represents the minimum value of L(j), L(j) max Indicates the maximum value of L(j).

[0167] After completing the data standardization process, it is necessary to perform various comprehensive index analyses on each dimension sub-tag. The various comprehensive index value analyses include:

[0168] (1) User power characteristics label L rp

[0169] The user power consumption characteristic label includes user power consumption characteristic labels L1-L6, a total of 6-dimensional data. The comprehensive label value L of the power consumption characteristic is obtained using the following formula: rp , where w j is the weight of each dimension data.

[0170]

[0171] (2) Power regulation willingness label L rw

[0172] The power regulation intention label includes the user power consumption characteristic label L7-L 10 , a total of 4-dimensional data, use the following formula to calculate the comprehensive label value L of the power consumption characteristics rw , where w j is the weight of each dimension data.

[0173]

[0174] (3) Electricity Carbon Reduction Potential Label L crp

[0175] The power regulation intention label includes the user power consumption characteristic label L 11 -L 15 , a total of 5-dimensional data, use the following formula to calculate the comprehensive label value L of the power consumption characteristics crp , where w j is the weight of each dimension data.

[0176]

[0177] The weights of each dimension are determined by the entropy weight method. The weights of L1-L6 are: 0.23, 0.25, 0.05, 0.06, 0.03, 0.39, and L7-L 10The weights are: 0.26, 0.14, 0.14, 0.46, L 11 -L 15 The weights are: 0.01, 0.19, 0.09, 0.01, 0.70.

[0178] After completing the analysis of various comprehensive indicators, the k-means clustering algorithm is used to determine the user's cluster. First, the Euclidean distance is combined with the k-means clustering idea to cluster the data according to the classification requirements. Then, the cluster center points of each type of data are obtained, and the user's cluster type is determined based on each cluster center point. The specific process is as follows:

[0179] For user power consumption characteristics label L rp (i) Based on the Euclidean distance K-means clustering algorithm, cluster the clusters into m1 categories and find the cluster center points L of each category. rp (k);

[0180]

[0181] For the power regulation willingness label L rw (i) Based on the Euclidean distance K-means clustering algorithm, clustering is done into m2 categories, and the cluster center points L of each category are obtained. rw (k);

[0182]

[0183] Carbon reduction potential label for electricity crp (i) Based on the Euclidean distance K-means clustering algorithm, the clustering idea is clustered into m3 categories, and the cluster center point L of each category is obtained. crp (k);

[0184]

[0185] According to L rp (i) L rw (i) L crp (i) The cluster center points of the three types of data are combined to form m1×m2×m3 three-dimensional surface center points, corresponding to m1×m2×m3 clusters respectively;

[0186] After completing the user clustering judgment through the k-means clustering algorithm, user cluster analysis is required to finally determine the user cluster.

[0187] First, according to the three-dimensional labels, they are divided into three categories: low, medium, and high. Then the order of clusters is determined. The higher the cluster, the more favorable the user's characteristics are for carbon reduction regulation. Finally, Figure 3 As shown, after determining the user's cluster according to the order of clusters, a comprehensive presentation of the user's carbon portrait is performed.

[0188] When constructing user electricity consumption characteristic labels, the present invention divides the daily load into four periods according to the typical power grid load: morning peak, noon peak, evening peak, and night valley. These periods correspond to T1 period: 6:00-10:00, T2 period: 10:00-16:00, T3 period: 16:00-21:00, and T4 period: 21:00-6:00. Different users are divided into 27 clusters based on their electricity consumption characteristics, low-carbon electricity consumption behavior characteristics, and electricity production and consumption characteristics. A three-dimensional comprehensive presentation of each type of user is performed. The portraits clearly show the position of each type of user in the three-dimensional coordinate system, the number of each type of user, and their approximate proportion. Power companies can formulate reasonable policies to regulate user behavior based on the user characteristics displayed in the portraits, thereby achieving the goal of energy conservation and emission reduction.

[0189] The present invention maps the behaviors of various types of users onto a two-dimensional plane of the user's electricity production and consumption carbon characteristics and the user's electricity consumption characteristics, a two-dimensional plane of the user's electricity consumption characteristics and the user's low-carbon electricity consumption behavior, and a two-dimensional plane of the user's electricity production and consumption carbon characteristics and the user's low-carbon electricity consumption behavior. The user's position in each plane can be clearly seen, which facilitates power companies to conduct control variable analysis on various aspects of user characteristics and refine carbon reduction regulation strategies.

[0190] The present invention can help power companies grasp the carbon characteristics and regulation potential of each user's electricity use, as well as the interaction and regulation capabilities between users and the power grid, from both objective and subjective levels during each typical period. This allows them to more reasonably formulate user-side electricity regulation plans, regulate and guide the carbon dioxide emitted during the generation and use of electricity used by residential users on a daily basis, and ultimately achieve the goal of reducing carbon dioxide emissions from residential electricity use.

[0191] In addition to the above embodiments, the present invention may also have other implementation methods. Any technical solutions formed by equivalent replacement or equivalent transformation fall within the protection scope required by the present invention.

Claims

1. A user carbon profiling method based on electricity usage behavior, characterized in that: The following steps are involved: S1. User data preprocessing, including: collecting and filtering multi-dimensional user behavior data, and removing error values and wrong user data; S2. Build a multi-dimensional user feature labeling model, establishing a multi-source user feature labeling system based on three dimensions: user electricity consumption characteristics, user low-carbon electricity consumption behavior, and user electricity production and carbon consumption characteristics; The user electricity consumption carbon emission characteristic label is designed to indicate the potential of residential users to reduce carbon dioxide emissions by responding to the relevant power grid energy conservation and carbon reduction policies and rationally allocating and using electricity. The user electricity consumption carbon emission characteristic label includes: the user electricity clean energy ratio label, the green appliance energy consumption ratio label, the user electricity carbon emission characteristic label, the user carbon emission regulation coefficient label, and the user effective alternative electricity quantity label. in: The user's electricity clean energy ratio label L cl , defined as the proportion of clean energy in the electricity used by the user during the calculation period, as shown in the following formula: Where, L p The amount of photovoltaic energy used in the electricity used by users, L t The amount of thermal power used in the electricity consumed by users, L cl It reflects the proportion of clean energy in the daily electricity used by users, and its value is positively correlated with the user's carbon reduction potential; The green appliance energy consumption ratio label L g , defined as the proportion of the electric energy consumption of green household appliances used by the user to the total electric energy consumption of the user during the calculation period, as shown in the following formula: Where, L ge Monthly consumption of green appliances by users, L t The amount of thermal power used in the electricity consumed by users, L p The amount of photovoltaic energy used in the electricity consumed by users and the proportion of energy consumption of green appliances by users reflect the potential for carbon reduction in electricity consumption by users; The user electricity carbon emission characteristic label L ce , which is defined as the ratio of the carbon emissions generated by a user’s personal electricity consumption to the average carbon emissions generated by electricity consumption by users within the statistical sample range during the calculation period, as shown in the following formula: L tc =L t ×a t (15) L avc =L av ×a t (16) Where, L tc Carbon emissions generated by electricity consumption, L av is the carbon emissions generated by the average electricity consumption of users in the community, L t The amount of thermal power used in the electricity consumed by users, L av is the average electricity consumption per user in the sample community, a t The carbon emission coefficient corresponding to electric energy; the user's electricity carbon emission characteristic label is the most direct expression of the user's potential to reduce carbon dioxide emissions through electricity regulation; The user carbon emission control coefficient label L rce , which is defined as the proportion of the electricity consumed by the user during the calculation period that is left over from photovoltaic power after the base load is met, which is the proportion of the demand response load, as shown in the following formula: Where, L b is the user's basic load, P Ti is the power of the i-th type adjustable load, P zj is the power of class j that can be reduced, K i is the number of adjustable loads of type i, K j is the amount of load that can be reduced in category j, and the user carbon emission regulation coefficient reflects the potential of the clean energy used by the user to participate in demand-side response; The user's effective replacement power label E es , defined as the effective amount of electricity replaced by electric energy during the calculation period, as shown in the following formula: Where, α rem M is the proportion of clean energy electricity consumption of users in the calculation period, es is the number of electric energy replacement devices in the user's household, E esm The effective electricity replacement label represents the amount of electricity used by electric energy replacement equipment during the calculation period. It indicates the amount of electricity used in a user's household energy structure to replace traditional primary energy sources such as coal, oil, and natural gas using convenient, safe, low-carbon, and clean electricity, objectively reflecting the user's low-carbon emission reduction potential. S3. Calculate comprehensive indicators of various characteristic indicators based on the divided time periods. The comprehensive indicators include: user electricity consumption characteristic labels, user low-carbon electricity consumption behavior labels, and user electricity consumption carbon production and consumption characteristic labels. Different residential users are divided into clusters with different attributes based on the user electricity consumption characteristic labels, user low-carbon electricity consumption behavior labels, and user electricity consumption carbon production and consumption characteristic labels. S4. Visualize the results of the carbon profiling based on user clusters, and use the profiling results as a basis for power companies to formulate user demand-side response policies and low-carbon regulation.

2. A user carbon profiling method based on electricity usage behavior according to claim 1, characterized in that: In step S1, the user's multi-dimensional behavior data includes: the user's multi-dimensional electricity usage behavior data collected based on the non-household terminal, the electricity consumption data of various loads identified by the non-household terminal, with a sampling time of 15 minutes, and 96 points of electricity consumption data of various loads of the user obtained in units of days; the user's electricity bill data collected based on the marketing system, in units of days, including: the user's total electricity bill, and the electricity bill generated by the user during peak and valley periods.

3. The user carbon profiling method based on electricity usage behavior according to claim 1, characterized in that: In step S2, the user's electricity characteristic tags include: power characteristic tags, load characteristic tags, and electrical appliance characteristic tags; wherein: The power characteristic label L E , defined as the ratio of the user's electricity consumption to the average household electricity consumption during the calculation period, as shown in the following formula: Where W i is the power consumed by the user during the computing cycle, W av is the average power consumption of all users within the statistical range, and n is the number of users within the statistical range; The load characteristic label is defined as the maximum load power, average load rate, maximum load utilization hours, and peak-to-valley difference rate within the analysis period, which are expressed as: Maximum load power: L max-Ti =P max-Ti (2) Average load factor: Maximum load utilization hours: Peak-to-valley rate: Where, P Ti is the power of the i-th type adjustable load, P max-Ti is the maximum power in the statistical Ti period, P av-Ti is the average power in the statistical Ti period, P max is the maximum power within the statistical time, P min is the minimum power within the statistical time; The electrical appliance characteristic label is defined as the ratio of the total power of the load that can respond to the demand side to the reference power, as shown in the following formula: L app =(∑k i P Ti +∑k j P zj ) / 1000 (6) Where, P Ti is the power of the i-th type adjustable load, P zj is the power of the jth type of load that can be reduced, k represents the number of each type of load, k i is the number of adjustable loads of type i, k j is the number of the jth type of reducible loads; the adjustable load refers to the load that does not stop running during user use but the power can be adjusted, and the reducible load refers to the load that can be interrupted during user use.

4. The user carbon profiling method based on electricity usage behavior according to claim 1, characterized in that: In step S2, the user's low-carbon electricity consumption behavior tag includes: a peak-valley electricity price sensitivity estimation tag, an electricity consumption tendency estimation tag, and a user's low-carbon electricity consumption awareness estimation tag; wherein: The peak-valley electricity price sensitivity estimation label is defined as the proportion of peak-valley electricity charges and the change in electricity consumption at the turning point of peak-valley electricity prices, which are expressed as: Peak and valley electricity charges: Where C f represents the peak electricity charge during the calculation period, C g Indicates the off-peak electricity price within the calculation period; Changes in electricity consumption at the turning point of peak-valley electricity prices: Where W i (t n +1) represents the electricity consumption one hour after the peak-valley electricity price turns in the evening of day i, W i (t n -1) represents the amount of electricity consumed one hour before the peak-valley electricity price turns on the morning of day i, T is the calculation period, and L cs Reflects the user's sensitivity to electricity prices; The power consumption tendency estimation tag is defined as the valley power coefficient, which is expressed as: Where, L v Indicates the user's electricity consumption during off-peak hours, L z Indicates the total electricity consumption of the user, and the valley electricity coefficient reflects the user's tendency to consume electricity during off-peak hours; The estimation of users' awareness of low-carbon electricity use is based on data collected and analyzed by the power system marketing department and other related network systems, including the number of times users set their home air conditioners to energy-saving temperatures, the number of times users checked their electricity bills, the number of times users checked their electricity bills by logging into the mobile app, the number of times users checked their electricity bills by logging into the online business hall website, and the number of times users checked their electricity bills by calling 95598: L es =N esc +N ce (10) N ce =N App +N net +N 598 (11) Where N esc The number of times the user sets the home air conditioner to the energy-saving temperature, N ce The number of times the user checks the electricity usage, N App The number of times the user checks the electricity bill by logging into the mobile APP, N net The number of times users check their electricity bills by logging into the online business hall, N 598 The number of times users checked their electricity bills by dialing 95598, L es represents the low-carbon electricity usage awareness estimate, L es The larger the value, the more concerned the user is about his or her low-carbon electricity consumption behavior and the better the low-carbon electricity consumption awareness.

5. The user carbon profiling method based on electricity usage behavior according to claim 1, characterized in that: In step S3, the method for calculating the comprehensive index of various characteristic indicators includes: data standardization processing, analysis of various comprehensive index values, K-means clustering algorithm, and user cluster analysis.

6. A user carbon profiling method based on electricity usage behavior according to claim 5, characterized in that: The data standardization process includes: (1) Establish a comprehensive control matrix for the same period, which includes three categories of 15-dimensional labels: 6-dimensional user electricity consumption characteristic labels L1-L6, 4-dimensional user low-carbon electricity consumption behavior labels L7-L 10 , user electricity production and consumption carbon characteristics label 5-dimensional L 11 —L 15 , that is, the data sample of N users is expressed as: (2) All users’ data of the same category are represented as column vectors: {L i (1),L i (2),…L i (15)} (3) Perform data cleaning and standardize the data using min-max normalization: Where, L(j) min represents the minimum value of L(j), L(j) max Indicates the maximum value of L(j).

7. The user carbon profiling method based on electricity usage behavior according to claim 5, characterized in that: The comprehensive index value analysis includes: (1) User power characteristics label L rp The user power consumption characteristic label includes user power consumption characteristic labels L1-L6, a total of 6-dimensional data. The comprehensive label value L of the power consumption characteristic is obtained using the following formula: rp , where w j is the weight of each dimension data: (2) Power regulation willingness label L rw The power regulation intention label includes the user power consumption characteristic label L7-L 10 , a total of 4-dimensional data, use the following formula to calculate the comprehensive label value L of the power consumption characteristics rw , where w j is the weight of each dimension data: (3) Electricity Carbon Reduction Potential Label L crp The electricity carbon reduction potential label includes the user electricity characteristics label L 11 —L 15 , a total of 5-dimensional data, use the following formula to calculate the comprehensive label value L of the power consumption characteristics crp , where w j is the weight of each dimension data: The weights of each dimension are determined by the entropy weight method. The weights of L1-L6 are: 0.23, 0.25, 0.05, 0.06, 0.03, 0.39, and L7-L 10 The weights are: 0.26, 0.14, 0.14, 0.46, L 11 —L 15 The weights are: 0.01, 0.19, 0.09, 0.01, 0.

70.

8. The user carbon profiling method based on electricity usage behavior according to claim 5, characterized in that: The K-Means clustering algorithm is used to analyze and determine the cluster to which the user belongs, where: For user electricity characteristics label L rp (i) Based on the Euclidean distance K-means clustering algorithm, cluster the clusters into m1 categories and find the cluster center points L of each category. rp (k); For the power regulation willingness label L rw (i) Based on the Euclidean distance K-means clustering algorithm, clustering is done into m2 categories, and the cluster center points L of each category are obtained. rw (k); Carbon reduction potential label for electricity crp (i) Based on the Euclidean distance K-means clustering algorithm, the clustering idea is clustered into m3 categories, and the cluster center point L of each category is obtained. crp (k); According to L rp (i) L rw (i) L crp (i) The cluster center points of the three types of data are combined to form m1×m2×m3 three-dimensional surface center points, corresponding to m1×m2×m3 clusters respectively; Among them, m1=m2=m3=3.

9. The user carbon profiling method based on electricity usage behavior according to claim 1, characterized in that: The presentation of the user carbon portrait described in step S4 refers to the visualization of the user electricity usage characteristic labels, user low-carbon electricity usage behavior labels, user electricity production and consumption carbon characteristic labels, and the determined clusters to which each user belongs, which are calculated based on the collected user electricity usage behavior data.

Images