Customer portrait-based power retail package pushing method and system

Abstract

The invention discloses an electric power retail package pushing method and system based on customer portraits, and relates to the technical field of electric power markets. The method comprises the following steps: constructing a customer peak clipping and valley filling portrait model based on DDPM, mining the historical load of a customer, extracting the peak clipping and valley filling characteristics of the customer, and constructing a peak clipping and valley filling time difference model and a peak clipping and valley filling electric quantity difference model in combination with the peak value characteristics and valley value characteristics of a power grid; according to the peak clipping and valley filling time difference value and the peak clipping and valley filling electric quantity difference value, respectively formulating four electric power retail packages; and constructing an electric power retail package judgment model based on the GPT, judging the peak clipping and valley filling time difference and the peak clipping and valley filling electric quantity difference of the customer, and selecting one of the packages 1 to 4 to be pushed to the customer according to the judgment result of the GPT. According to the method, the appropriate power retail package can be pushed to the user in a customized manner according to the peak clipping and valley filling characteristics of the customer, the pushing accuracy is improved, and the enthusiasm of the user to participate in the power package activity is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power market, and in particular to a method and system for pushing power retail packages based on customer portraits. Background Art

[0002] With the gradual opening of the power market and the application of intelligent technology, the power retail industry is facing increasingly fierce market competition. In order to improve the service quality and customer satisfaction of power companies, personalized power package push based on user portraits has become an important development trend. Through accurate user portraits, power retailers can provide more personalized package options based on customers' power needs and habits, and enhance user stickiness and satisfaction. However, the existing power retail package push methods still have a series of shortcomings and need to be improved.

[0003] First, traditional user portrait construction methods mostly rely on simple statistical information, such as basic information of users, electricity consumption, etc. Customers' electricity consumption behavior is usually volatile, and these simple statistical information cannot fully reflect the actual electricity demand of customers during different electricity consumption periods (such as peak and valley periods), and it is difficult to capture the deep-level characteristics of users in terms of peak load shaving and valley filling, resulting in the recommendation of electricity packages being difficult to accurately match users' actual needs.

[0004] Secondly, the existing package formulation methods fail to fully consider the differences in customers' electricity demand in specific time periods. For example, during peak electricity demand periods, customers may face higher electricity prices, while during off-peak periods, they may receive lower electricity price discounts. Existing methods often ignore these differences in time dimensions and cannot dynamically adjust according to customers' peak-shaving and valley-filling characteristics, resulting in package recommendations that cannot optimize electricity consumption and may even increase users' electricity costs.

[0005] In addition, traditional package push methods lack pertinence and personalization. Some existing technologies push packages based only on basic user information (such as monthly electricity consumption, electricity costs, etc.), fail to deeply analyze the changes in customer electricity demand in different time periods, fail to effectively match customer load fluctuations with grid load, and fail to provide truly personalized and optimized electricity package recommendations, resulting in poor user experience.

[0006] In addition, the current power package push method lacks dynamic adjustment and real-time optimization functions. With the changes in the power market and the rapid fluctuations in user demand, the traditional push method cannot respond quickly to these changes. It is difficult for power companies to adjust package recommendation strategies based on real-time data, resulting in package push lagging behind market and customer demand, affecting the optimal use of power resources and customer satisfaction.

[0007] Therefore, a new method and system for pushing electricity retail packages is needed to deeply explore users' electricity demand patterns, including the characteristics of peak shaving and valley filling, provide electricity retailers with more detailed user portraits, better meet the peak shaving and valley filling needs of the power grid, and improve the attractiveness and conversion rate of push. Summary of the invention

[0008] In view of the problems in the prior art such as overly simple user portraits, ignoring time period differences, lacking personalized recommendations, and lacking dynamic adjustment functions, the present invention proposes a method and system for pushing electricity retail packages based on customer portraits. By conducting an in-depth analysis of the customer's historical load, the peak-shaving and valley-filling features of the customer's electricity consumption behavior are extracted based on the DDPM technology, and different electricity package models are constructed in combination with the load characteristics of the power grid, and customized electricity packages are pushed through GPT judgment.

[0009] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0010] On the one hand, the present invention proposes a method for pushing a power retail package based on a customer portrait, the method comprising:

[0011] Construct a customer peak-shaving and valley-filling portrait model based on DDPM, including a customer peak feature portrait model and a customer valley feature portrait model, which are used to mine customer historical loads, extract customer peak-shaving and valley-filling features, and realize the portrait of customer peak-shaving and valley-filling features;

[0012] Based on the peak characteristics of the power grid, the valley characteristics of the power grid and the customer's peak-shaving and valley-filling portrait model, a peak-shaving and valley-filling time difference model and a peak-shaving and valley-filling electricity quantity difference model are constructed. According to the high and low peak-shaving and valley-filling time difference and the high and low peak-shaving and valley-filling electricity quantity difference, four types of electricity retail packages are formulated respectively, specifically:

[0013] Package 1: Electricity retail package when the time difference between peak shaving and valley filling is large and the difference in electricity volume between peak shaving and valley filling is large;

[0014] Package 2: Electricity retail package when the time difference between peak shaving and valley filling is high and the difference in electricity volume between peak shaving and valley filling is low;

[0015] Package 3: Electricity retail package when the time difference between peak shaving and valley filling is low and the difference between peak shaving and valley filling electricity is high;

[0016] Package 4: Electricity retail package when the time difference between peak shaving and valley filling is low and the difference in electricity volume between peak shaving and valley filling is low;

[0017] Build a GPT-based electricity retail package judgment model, including a peak shaving and valley filling time difference judgment model and a peak shaving and valley filling electricity quantity difference judgment model, judge the customer's peak shaving and valley filling time difference and peak shaving and valley filling electricity quantity difference, and choose to push one of the packages one to four to the customer based on the judgment results of GPT.

[0018] As a preferred solution of the present invention, the customer peak feature portrait model mines the customer historical load based on the DDPM algorithm to extract the customer peak feature, which is expressed as:

[0019]

[0020] The customer peak feature portrait is expressed as:

[0021]

[0022] In the formula, Peak characteristics for customers; represents the customer peak feature portrait model based on DDPM; P L (t) historical load of the customer; is the customer load peak start time, is the end time of the customer load peak, is the extracted customer load peak;

[0023] The customer valley feature portrait model mines the customer's historical load based on the DDPM algorithm and extracts the customer valley feature, which is expressed as:

[0024]

[0025] The customer valley feature portrait is expressed as:

[0026]

[0027] In the formula, It is the customer valley feature; It represents the customer valley feature profiling model based on DDPM; The customer load valley start time. The end time of the customer load valley. is the extracted customer load valley value.

[0028] As a preferred solution of the present invention, the grid peak characteristics include grid power peak start time, grid power peak end time and grid peak power, expressed as:

[0029]

[0030] In the formula, Indicates the peak characteristics of the power grid; is the starting time of the grid power peak, is the end time of the power grid peak, is the peak power of the power grid;

[0031] The grid valley characteristics include the grid power valley start time, grid power valley end time and grid valley power, and the expression is:

[0032]

[0033] In the formula, Indicates the valley value characteristics of the power grid; is the starting time of the grid power valley value, is the end time of the grid power valley value, is the grid valley power.

[0034] As a preferred solution of the present invention, the peak shaving and valley filling time difference model includes a peak time difference model and a valley time difference model;

[0035] The peak time difference model specifically calculates the peak time difference according to the grid power peak start time, grid power peak end time, customer load peak start time, and customer load peak end time. The expression is:

[0036]

[0037] The valley time difference model specifically calculates the valley time difference according to the grid power valley start time, grid power valley end time and customer load valley start time and customer load valley end time, and the expression is:

[0038]

[0039] The peak shaving and valley filling time difference model is expressed as:

[0040] In the formula, is the peak time difference; is the valley time difference; T C It is the time difference of peak shaving and valley filling.

[0041] As a preferred solution of the present invention, the peak shaving and valley filling electricity quantity difference model includes a peak electricity quantity difference model and a valley electricity quantity difference model;

[0042] The peak power difference model specifically calculates the peak power difference according to the peak power accumulation of the power grid and the peak power accumulation of the customer load, and the expression is:

[0043]

[0044] The valley power difference model specifically calculates the valley power difference according to the valley power cumulative power of the power grid and the valley power cumulative power of the customer load, and the expression is:

[0045]

[0046] The peak shaving and valley filling electricity difference model is expressed as:

[0047] In the formula, is the peak power difference; is the valley power difference; Q C is the difference in power consumption between peak load reduction and valley load filling; t represents time.

[0048] As a preferred solution of the present invention, the four electricity retail packages are respectively expressed as:

[0049]

[0050] In the formula, Y1 represents package one, R1 is the basic fee of package one, and M1 is the included electricity of package one; Y2 represents package two, R2 is the basic fee of package two, and M2 is the included electricity of package two; Y3 represents package three, R3 is the basic fee of package three, and M3 is the included electricity of package three; Y4 represents package four, R4 is the basic fee of package four, and M4 is the included electricity of package four; T C is the peak shaving and valley filling time difference, T θ Q is the peak shaving and valley filling time difference threshold; C Q is the peak load difference, θ The power difference threshold for peak shaving and valley filling;

[0051] Among them, the calculation formulas for the basic fees and included electricity consumption of the four electricity retail packages are:

[0052]

[0053]

[0054] Where R0 is the basic fee of the electricity retail package; α1, α2, α3, and α4 are the time difference cost coefficients corresponding to packages one, two, three, and four, respectively; β1, β2, β3, and β4 are the electricity difference cost coefficients corresponding to packages one, two, three, and four, respectively; M0 is the basic included electricity volume of the electricity retail package; δ1, δ2, δ3, and δ4 are the time difference electricity coefficients corresponding to packages one, two, three, and four, respectively; ε1, ε2, ε3, and ε4 are the electricity difference electricity coefficients corresponding to packages one, two, three, and four, respectively.

[0055] As a preferred solution of the present invention, the peak shaving and valley filling time difference judgment model is based on the historical load P of customer i. L,i (t), grid load P G (t), customer i's historical electricity basic cost R i 、Customer i's historical electricity consumption Q i and the peak shaving and valley filling time difference threshold Tθ As input, the result T is determined by the peak-shaving and valley-filling time difference of customer i. δ,i For output, the expression is:

[0056]

[0057] When T δ,i When the value is 0, it means the peak-shaving and valley-filling time difference T of customer i C,i Less than or equal to the peak shaving and valley filling time difference threshold T θ , when T δ,i When the value is 1, it means the peak-shaving and valley-filling time difference T of customer i C,i Greater than the peak shaving and valley filling time difference threshold T θ , the expression is:

[0058]

[0059] The peak load reduction and valley filling power difference judgment model is based on the historical load P of customer i. L,i (t), grid load P G (t), customer i's historical electricity basic cost R i 、Customer i's historical electricity consumption Q i And the peak-shaving and valley-filling power difference threshold Q θ As input, the result Q is determined by the peak-shaving and valley-filling power difference of customer i. δ,i For output, the expression is:

[0060]

[0061] When Q δ,i When the value is 0, it means the peak load difference Q of customer i C,i Less than or equal to the peak-shaving and valley-filling power difference threshold Q θ , when Q δ,i When the value is 1, it indicates the peak load-filling power difference Q of customer i C,i Greater than the peak-shaving and valley-filling power difference threshold Q θ , the expression is:

[0062]

[0063] In the formula, It represents the time difference judgment model of peak shaving and valley filling based on GPT; represents the peak-shaving and valley-filling power difference judgment model based on GPT; t represents time.

[0064] As a preferred solution of the present invention, the selection of pushing one of Packages 1 to 4 to the customer according to the judgment result of GPT is specifically:

[0065] When the time difference judgment result T of customer i isδ,i Take 1, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When 1 is taken, package 1 is pushed to customer i;

[0066] When the time difference judgment result T of customer i is δ,i Take 1, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 0, package 2 is pushed to customer i;

[0067] When the time difference judgment result T of customer i is δ,i Take 0, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 1, package 3 is pushed to customer i;

[0068] When the time difference judgment result T of customer i is δ,i Take 0, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 0, package 4 is pushed to customer i.

[0069] On the other hand, the present invention also proposes a system for pushing a power retail package based on a customer portrait. Based on the method for pushing a power retail package based on a customer portrait as described above, the system comprises:

[0070] The customer peak-shaving and valley-filling portrait module is used to build a customer peak-shaving and valley-filling portrait model, including a customer peak feature portrait model and a customer valley feature portrait model. Based on the DDPM algorithm, the customer's historical load is mined to extract customer peak features and customer valley features respectively, thus realizing the portrait of customer peak-shaving and valley-filling features.

[0071] A power grid feature extraction module, used to extract power grid peak features and power grid valley features, wherein the power grid peak features include power grid power peak start time, power grid power peak end time and power grid peak power, and the power grid valley features include power grid power valley start time, power grid power valley end time and power grid valley power;

[0072] The peak shaving and valley filling time difference module is used to construct a peak shaving and valley filling time difference model to calculate the peak shaving and valley filling time difference, including a peak value time difference model and a valley value time difference model, which are used to calculate the peak value time difference and the valley value time difference respectively;

[0073] The peak shaving and valley filling electricity difference module is used to construct a peak shaving and valley filling electricity difference model to calculate the peak shaving and valley filling electricity difference, including a peak electricity difference model and a valley electricity difference model, to calculate the peak electricity difference and the valley electricity difference respectively;

[0074] The electricity retail package formulation module is used to formulate four electricity retail packages according to the difference in peak shaving and valley filling time and the difference in peak shaving and valley filling electricity volume;

[0075] The power retail package judgment module is used to build a power retail package judgment model based on GPT, including a peak-shaving and valley-filling time difference judgment model and a peak-shaving and valley-filling electricity quantity difference judgment model, to judge the customer's peak-shaving and valley-filling time difference and peak-shaving and valley-filling electricity quantity difference;

[0076] The electricity retail package push module is used to select and push one of the packages 1 to 4 to the customer based on the judgment result of GPT.

[0077] In addition, the present invention also proposes a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for pushing electricity retail packages based on customer portraits as described above are implemented.

[0078] Compared with the prior art, the beneficial effects of the present invention are as follows: by modeling the customer's historical load through the DDPM algorithm, the customer's peak shaving and valley filling characteristics are extracted, and the customer's electricity consumption pattern is accurately characterized, so that the electricity retail package can be accurately customized according to the customer's actual needs, thereby improving user satisfaction; the peak shaving and valley filling time difference and electricity quantity difference models are constructed in combination with the grid load characteristics and the customer's peak shaving and valley filling characteristics, and four typical personalized electricity retail packages are designed to meet the needs of different users and reduce waste while reducing user electricity bills; through the GPT-based intelligent judgment model, the most suitable package is automatically pushed according to the customer's electricity consumption characteristics, which improves the accuracy of the electricity retail package push, optimizes the allocation of power resources, reduces customer electricity bills, and greatly improves service efficiency and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0080] in:

[0081] Figure 1 is a flow chart of the method of the present invention;

[0082] Figure 2 It is a schematic diagram of the modular structure of the system of the present invention. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0084] like Figure 1 As shown in FIG. 1 , an embodiment of the present invention is provided, which provides a method for pushing a power retail package based on a customer portrait, comprising the following steps:

[0085] S1: Build a customer peak-shaving and valley-filling portrait model based on DDPM, including a customer peak feature portrait model and a customer valley feature portrait model, which are used to mine the customer's historical load, extract the customer's peak-shaving and valley-filling features, and realize the portrait of the customer's peak-shaving and valley-filling features (representing the customer's overall electricity consumption fluctuation pattern);

[0086] DDPM (Denoising Diffusion Probabilistic Model) is a probability model based on the diffusion process. The core idea is to gradually transform data into noise through a forward diffusion process, and then gradually transform noise into data through a reverse process. This reverse process is achieved by learning a conditional probability distribution that describes how to recover data from noise given the current state and the target state.

[0087] The customer's historical electricity consumption data (historical load) is the customer's actual electricity load over the past period of time, recording the customer's electricity consumption in different time periods. The DDPM algorithm is used to model and analyze the customer's historical load to extract the customer's peak-shaving and valley-filling characteristics, that is, the characteristics during the peak and valley periods of electricity demand. Customer peak characteristics refer to the various features extracted from the customer's historical load for the "load peak" period (such as peak start time, peak end time, peak power, peak duration, etc.). These features represent the customer's electricity consumption behavior during the peak period of electricity demand and can be used to describe the customer's electricity consumption pattern during high-load periods. Customer valley characteristics are similar to peak characteristics, referring to the customer's electricity consumption characteristics during the valley period (such as valley start time, valley end time, valley power, valley duration, etc.). These features describe the customer's electricity consumption pattern during the valley period of electricity demand.

[0088] The term "profile" generally refers to a comprehensive description of a customer's electricity usage behavior through the integration of multiple features. For example, a customer peak feature profile includes not only the customer's peak features, but also the patterns and trends obtained by analyzing these features (for example, the frequency of peaks, the intensity of each peak, etc.), which can be understood as a customer's electricity usage profile during peak hours, including but not limited to a single peak feature.

[0089] In one embodiment, the customer peak feature portrait model mines the customer historical load based on the DDPM algorithm and extracts the customer peak feature, which is expressed as:

[0090]

[0091] The customer peak feature portrait is expressed as:

[0092]

[0093] In the formula, Peak characteristics for customers; represents the customer peak feature portrait model based on DDPM; P L (t) historical load of the customer; is the customer load peak start time, is the end time of the customer load peak, is the extracted customer load peak;

[0094] The customer valley feature portrait model mines the customer's historical load based on the DDPM algorithm and extracts the customer valley feature, which is expressed as:

[0095]

[0096] The customer valley feature portrait is expressed as:

[0097]

[0098] In the formula, It is the customer valley feature; It represents the customer valley feature profiling model based on DDPM; The customer load valley start time. The end time of the customer load valley. is the extracted customer load valley value.

[0099] S2: Based on the peak characteristics of the power grid, the valley characteristics of the power grid and the customer's peak-shaving and valley-filling portrait model, a peak-shaving and valley-filling time difference model and a peak-shaving and valley-filling electricity difference model are constructed. According to the high and low peak-shaving and valley-filling time difference and the high and low peak-shaving and valley-filling electricity difference, four types of electricity retail packages are formulated respectively, specifically:

[0100] Package 1: Electricity retail package when the time difference between peak shaving and valley filling is large and the difference in electricity volume between peak shaving and valley filling is large;

[0101] Package 2: Electricity retail package when the time difference between peak shaving and valley filling is high and the difference in electricity volume between peak shaving and valley filling is low;

[0102] Package 3: Electricity retail package when the time difference between peak shaving and valley filling is low and the difference between peak shaving and valley filling electricity is high;

[0103] Package 4: Electricity retail package when the time difference between peak shaving and valley filling is low and the electricity quantity difference between peak shaving and valley filling is low.

[0104] Furthermore, the grid peak characteristics include the grid power peak start time, the grid power peak end time and the grid peak power, and the expression is:

[0105]

[0106] In the formula, Indicates the peak characteristics of the power grid; is the starting time of the grid power peak, is the end time of the power grid peak, is the peak power of the power grid;

[0107] The grid valley characteristics include the grid power valley start time, grid power valley end time and grid valley power, and the expression is:

[0108]

[0109] In the formula, Indicates the valley value characteristics of the power grid; is the starting time of the grid power valley value, is the end time of the grid power valley value, is the grid valley power.

[0110] In one of the embodiments, the peak shaving and valley filling time difference model includes a peak time difference model and a valley time difference model;

[0111] The peak time difference model specifically calculates the peak time difference according to the grid power peak start time, grid power peak end time, customer load peak start time, and customer load peak end time. The expression is:

[0112]

[0113] The valley time difference model specifically calculates the valley time difference according to the grid power valley start time, grid power valley end time, customer load valley start time, and customer load valley end time. The expression is:

[0114]

[0115] Therefore, the peak shaving and valley filling time difference model can be expressed as:

[0116] In the formula, is the peak time difference; is the valley time difference; T C It is the time difference of peak shaving and valley filling.

[0117] The peak load shaving and valley load filling power difference model includes a peak power difference model and a valley power difference model;

[0118] The peak power difference model specifically calculates the peak power difference based on the peak power accumulation of the power grid and the peak power accumulation of the customer load. The expression is:

[0119]

[0120] The valley power difference model specifically calculates the valley power difference based on the valley power cumulative power of the grid and the valley power cumulative power of the customer load. The expression is:

[0121]

[0122] Therefore, the peak shaving and valley filling electricity difference model can be expressed as:

[0123] In the formula, is the peak power difference; is the valley power difference; Q C is the difference in power consumption between peak load reduction and valley load filling; t represents time.

[0124] The four electricity retail packages are represented as follows:

[0125]

[0126] In the formula, Y1 represents package one, R1 is the basic fee of package one, and M1 is the included electricity of package one; Y2 represents package two, R2 is the basic fee of package two, and M2 is the included electricity of package two; Y3 represents package three, R3 is the basic fee of package three, and M3 is the included electricity of package three; Y4 represents package four, R4 is the basic fee of package four, and M4 is the included electricity of package four; T C is the peak shaving and valley filling time difference, T θ Q is the peak shaving and valley filling time difference threshold; C Q is the peak load difference, θ The power difference threshold for peak shaving and valley filling;

[0127] Among them, the calculation formulas for the basic fees and included electricity consumption of the four electricity retail packages are:

[0128]

[0129]

[0130] Where R0 is the basic fee of the electricity retail package; α1, α2, α3, and α4 are the time difference cost coefficients corresponding to packages one, two, three, and four, respectively; β1, β2, β3, and β4 are the electricity difference cost coefficients corresponding to packages one, two, three, and four, respectively; M0 is the basic included electricity volume of the electricity retail package; δ1, δ2, δ3, and δ4 are the time difference electricity coefficients corresponding to packages one, two, three, and four, respectively; ε1, ε2, ε3, and ε4 are the electricity difference electricity coefficients corresponding to packages one, two, three, and four, respectively.

[0131] S3: Construct a GPT-based electricity retail package judgment model, including a peak shaving and valley filling time difference judgment model and a peak shaving and valley filling electricity quantity difference judgment model, judge the customer's peak shaving and valley filling time difference and peak shaving and valley filling electricity quantity difference, and choose to push one of the packages 1 to 4 to the customer based on the judgment result of GPT.

[0132] GPT (Generative Pre-trained Transformer) is a language model based on artificial intelligence technology. GPT technology is based on the Transformer architecture in deep learning, pre-trained through unsupervised learning, and fine-tuned on specific tasks to achieve efficient and accurate language processing.

[0133] In one specific embodiment, the peak load shaving and valley filling time difference judgment model is based on the historical load P of customer i. L,i (t), grid load P G (t), customer i's historical electricity basic cost R i 、Customer i's historical electricity consumption Q i and the peak shaving and valley filling time difference threshold T θ As input, the result T is determined by the peak-shaving and valley-filling time difference of customer i. δ,i For output, the expression is:

[0134]

[0135] When T δ,i When the value is 0, it means the peak-shaving and valley-filling time difference T of customer i C,i Less than or equal to the peak shaving and valley filling time difference threshold T θ , when T δ,i When the value is 1, it means the peak-shaving and valley-filling time difference T of customer i C,i Greater than the peak shaving and valley filling time difference threshold T θ , the expression is:

[0136]

[0137] The peak load reduction and valley filling electricity difference judgment model is based on the historical load P of customer i. L,i (t), grid load P G (t), customer i's historical electricity basic cost R i 、Customer i's historical electricity consumption Q i And the peak-shaving and valley-filling power difference threshold Q θ As input, the result Q is determined by the peak-shaving and valley-filling power difference of customer i. δ,i For output, the expression is:

[0138]

[0139] When Q δ,i When the value is 0, it means the peak load difference Q of customer i C,i Less than or equal to the peak-shaving and valley-filling power difference threshold Q θ , when Q δ,i When the value is 1, it indicates the peak load-filling power difference Q of customer i C,i Greater than the peak-shaving and valley-filling power difference threshold Q θ , the expression is:

[0140]

[0141] In the formula, It represents the time difference judgment model of peak shaving and valley filling based on GPT; represents the peak-shaving and valley-filling power difference judgment model based on GPT; t represents time.

[0142] Furthermore, according to the judgment result of GPT, one of the packages 1 to 4 is selected to be pushed to the customer, specifically:

[0143] When the time difference judgment result T of customer i is δ,i Take 1, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When 1 is taken, package 1 is pushed to customer i;

[0144] When the time difference judgment result T of customer i is δ,i Take 1, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 0, package 2 is pushed to customer i;

[0145] When the time difference judgment result T of customer i is δ,i Take 0, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 1, package 3 is pushed to customer i;

[0146] When the time difference judgment result T of customer i isδ,i Take 0, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 0, package 4 is pushed to customer i.

[0147] Step S1 uses the DDPM algorithm to obtain the customer's peak feature portrait and valley feature portrait as the basic data for the subsequent formulation of electricity retail packages; Step S2 combines these features with the grid load characteristics to formulate different electricity retail packages; Step S3 pushes matching electricity retail packages to customers based on the judgment of the GPT model. In this way, the entire process finally realizes personalized electricity retail package push through multi-level data input and output relationships, improving the efficiency of power resource use and customer satisfaction.

[0148] like Figure 2 As shown, another embodiment of the present invention provides a system for pushing a power retail package based on a customer portrait, based on the above-mentioned method for pushing a power retail package based on a customer portrait, including:

[0149] The customer peak-shaving and valley-filling portrait module is used to build a customer peak-shaving and valley-filling portrait model, including a customer peak feature portrait model and a customer valley feature portrait model. Based on the DDPM algorithm, the customer's historical load is mined to extract customer peak features and customer valley features respectively, thus realizing the portrait of customer peak-shaving and valley-filling features.

[0150] A power grid feature extraction module, used to extract power grid peak features and power grid valley features, the power grid peak features include power grid power peak start time, power grid power peak end time and power grid peak power, the power grid valley features include power grid power valley start time, power grid power valley end time and power grid valley power;

[0151] The peak shaving and valley filling time difference module is used to construct a peak shaving and valley filling time difference model to calculate the peak shaving and valley filling time difference, including a peak value time difference model and a valley value time difference model, which are used to calculate the peak value time difference and the valley value time difference respectively;

[0152] The peak shaving and valley filling electricity difference module is used to construct a peak shaving and valley filling electricity difference model to calculate the peak shaving and valley filling electricity difference, including a peak electricity difference model and a valley electricity difference model, to calculate the peak electricity difference and the valley electricity difference respectively;

[0153] The electricity retail package formulation module is used to formulate four electricity retail packages according to the difference in peak shaving and valley filling time and the difference in peak shaving and valley filling electricity volume;

[0154] The power retail package judgment module is used to build a power retail package judgment model based on GPT, including a peak-shaving and valley-filling time difference judgment model and a peak-shaving and valley-filling electricity quantity difference judgment model, to judge the customer's peak-shaving and valley-filling time difference and peak-shaving and valley-filling electricity quantity difference;

[0155] The electricity retail package push module is used to select and push one of the packages 1 to 4 to the customer based on the judgment result of GPT.

[0156] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or other arbitrary combinations. When implemented using software, it can be implemented in whole or in part in the form of a computer program product, which includes one or more computer instructions. When loading and executing computer program instructions on a computer, the process or function according to the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0157] In addition, each functional unit in each embodiment of the present application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. If the above-mentioned integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium can be a read-only memory, a disk or an optical disk, etc.

[0158] Therefore, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for pushing a power retail package based on a customer portrait as described above.

[0159] In summary, the present invention realizes the accurate extraction of customer's peak shaving and valley filling characteristics through in-depth mining of customer's historical load, and successfully characterizes the customer's electricity demand characteristics during peak and valley periods by constructing customer peak characteristic and valley characteristic models, providing personalized data support for the subsequent formulation of electricity packages; through accurate profiling of customer's peak shaving and valley filling characteristics, it can effectively identify the customer's electricity consumption pattern, thereby providing customized electricity packages according to the differences in electricity consumption behavior of different customers.

[0160] In terms of package formulation, the present invention combines the characteristics of power grid load, constructs a time difference and electricity quantity difference model, and designs four electricity retail packages by comparing the peak shaving and valley filling characteristics of customers and the power grid. These packages are pushed according to different combinations of customer's peak shaving and valley filling time difference and electricity quantity difference, ensuring that each customer can obtain the electricity package that best meets their actual needs, thereby optimizing the allocation of power resources and reducing customers' electricity costs.

[0161] In addition, the present invention adopts an intelligent judgment model based on GPT to automatically select the most suitable electricity package and push it to the customer according to the customer's peak shaving and valley filling characteristics. Through the intelligent push method, the accuracy and personalization level of package selection are further improved, manual intervention is reduced, and customer satisfaction is improved. In general, the present invention not only improves the efficiency of power resource utilization, but also enhances the personalized experience of customers and promotes the rationalization and economy of power consumption.

[0162] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various 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.

Claims

1. A method for pushing electricity retail packages based on customer portraits, characterized in that: The method comprises: Construct a customer peak-shaving and valley-filling portrait model based on DDPM, including a customer peak feature portrait model and a customer valley feature portrait model, which are used to mine customer historical loads, extract customer peak-shaving and valley-filling features, and realize the portrait of customer peak-shaving and valley-filling features; Based on the peak characteristics of the power grid, the valley characteristics of the power grid and the customer's peak-shaving and valley-filling portrait model, a peak-shaving and valley-filling time difference model and a peak-shaving and valley-filling electricity quantity difference model are constructed. According to the high and low peak-shaving and valley-filling time difference and the high and low peak-shaving and valley-filling electricity quantity difference, four types of electricity retail packages are formulated respectively, specifically: Package 1: Electricity retail package when the time difference between peak shaving and valley filling is large and the difference in electricity volume between peak shaving and valley filling is large; Package 2: Electricity retail package when the time difference between peak shaving and valley filling is high and the difference between peak shaving and valley filling electricity is low; Package 3: Electricity retail package when the time difference between peak shaving and valley filling is low and the difference between peak shaving and valley filling electricity is high; Package 4: Electricity retail package when the time difference between peak shaving and valley filling is low and the difference in electricity volume between peak shaving and valley filling is low; Build a GPT-based electricity retail package judgment model, including a peak shaving and valley filling time difference judgment model and a peak shaving and valley filling electricity quantity difference judgment model, judge the customer's peak shaving and valley filling time difference and peak shaving and valley filling electricity quantity difference, and choose to push one of the packages one to four to the customer based on the judgment results of GPT.

2. A method for pushing electricity retail packages based on customer portraits as claimed in claim 1, characterized in that: The customer peak feature portrait model mines the customer historical load based on the DDPM algorithm and extracts the customer peak features, which are expressed as: The customer peak feature portrait is expressed as: In the formula, Peak characteristics for customers; It represents the customer peak feature profiling model based on DDPM; P L (t) historical load of the customer; is the customer load peak start time, is the end time of the customer load peak, is the extracted customer load peak; The customer valley feature portrait model mines the customer's historical load based on the DDPM algorithm and extracts the customer valley feature, which is expressed as: The customer valley feature portrait is expressed as: In the formula, It is the customer valley feature; It represents the customer valley feature profiling model based on DDPM; The customer load valley start time. The end time of the customer load valley. is the extracted customer load valley value.

3. A method for pushing electricity retail packages based on customer portraits as claimed in claim 2, characterized in that: The grid peak characteristics include the grid power peak start time, the grid power peak end time and the grid peak power, and the expression is: In the formula, Indicates the peak characteristics of the power grid; is the starting time of the grid power peak, is the end time of the power grid peak, is the peak power of the power grid; The grid valley characteristics include the grid power valley start time, grid power valley end time and grid valley power, and the expression is: In the formula, Indicates the valley value characteristics of the power grid; is the starting time of the grid power valley value, is the end time of the grid power valley value, is the grid valley power.

4. A method for pushing electricity retail packages based on customer portraits as claimed in claim 3, characterized in that: The peak shaving and valley filling time difference model includes a peak time difference model and a valley time difference model; The peak time difference model specifically calculates the peak time difference according to the grid power peak start time, grid power peak end time, customer load peak start time, and customer load peak end time. The expression is: The valley time difference model specifically calculates the valley time difference according to the grid power valley start time, grid power valley end time and customer load valley start time and customer load valley end time, and the expression is: The peak shaving and valley filling time difference model is expressed as: In the formula, is the peak time difference; is the valley time difference; T C It is the time difference of peak shaving and valley filling.

5. The method for pushing a power retail package based on a customer portrait as claimed in claim 3, characterized in that: The peak shaving and valley filling electricity quantity difference model includes a peak electricity quantity difference model and a valley electricity quantity difference model; The peak power difference model specifically calculates the peak power difference according to the peak power accumulation of the power grid and the peak power accumulation of the customer load, and the expression is: The valley power difference model specifically calculates the valley power difference according to the valley power cumulative power of the power grid and the valley power cumulative power of the customer load, and the expression is: The peak shaving and valley filling electricity difference model is expressed as: In the formula, is the peak power difference; is the valley power difference; Q C is the difference in power consumption between peak load reduction and valley load filling; t represents time.

6. A method for pushing electricity retail packages based on customer portraits as claimed in claim 1, characterized in that: The four electricity retail packages are respectively represented as follows: In the formula, Y1 represents package one, R1 is the basic fee of package one, and M1 is the included electricity of package one; Y2 represents package two, R2 is the basic fee of package two, and M2 is the included electricity of package two; Y3 represents package three, R3 is the basic fee of package three, and M3 is the included electricity of package three; Y4 represents package four, R4 is the basic fee of package four, and M4 is the included electricity of package four; T C is the peak shaving and valley filling time difference, T θ Q is the peak shaving and valley filling time difference threshold; C Q is the peak load difference, θ The power difference threshold for peak shaving and valley filling; Among them, the basic fees and calculation formulas for the included electricity consumption of the four electricity retail packages are: Where R0 is the basic fee of the electricity retail package; α1, α2, α3, and α4 are the time difference cost coefficients corresponding to packages one, two, three, and four, respectively; β1, β2, β3, and β4 are the electricity difference cost coefficients corresponding to packages one, two, three, and four, respectively; M0 is the basic included electricity volume of the electricity retail package; δ1, δ2, δ3, and δ4 are the time difference electricity coefficients corresponding to packages one, two, three, and four, respectively; ε1, ε2, ε3, and ε4 are the electricity difference electricity coefficients corresponding to packages one, two, three, and four, respectively.

7. The method for pushing a power retail package based on a customer portrait according to claim 1, characterized in that: The peak shaving and valley filling time difference judgment model is based on the historical load P of customer i. L,i (t), grid load P G (t), customer i's historical electricity basic cost R i 、Customer i's historical electricity consumption Q i and the peak shaving and valley filling time difference threshold T θ As input, the result T is determined by the peak-shaving and valley-filling time difference of customer i. δ,i For output, the expression is: When T δ,i When the value is 0, it means the peak-shaving and valley-filling time difference T of customer i C,i Less than or equal to the peak shaving and valley filling time difference threshold T θ , when T δ,i When the value is 1, it means the peak-shaving and valley-filling time difference T of customer i C,i Greater than the peak shaving and valley filling time difference threshold T θ , the expression is: The peak load reduction and valley filling power difference judgment model is based on the historical load P of customer i. L,i (t), grid load P G (t), customer i's historical electricity basic cost R i 、Customer i's historical electricity consumption Q i And the peak-shaving and valley-filling power difference threshold Q θ As input, the result Q is determined by the peak-shaving and valley-filling power difference of customer i. δ,i For output, the expression is: When Q δ,i When the value is 0, it means the peak load difference Q of customer i C,i Less than or equal to the peak-shaving and valley-filling power difference threshold Q θ , when Q δ,i When the value is 1, it indicates the peak load-filling power difference Q of customer i C,i Greater than the peak-shaving and valley-filling power difference threshold Q θ , the expression is: In the formula, It represents the peak shaving and valley filling time difference judgment model based on GPT; represents the peak-shaving and valley-filling power difference judgment model based on GPT; t represents time.

8. A method for pushing a power retail package based on a customer portrait as claimed in claim 7, characterized in that: The method of selecting one of Packages 1 to 4 to push to the customer based on the judgment result of GPT is as follows: When the time difference judgment result T of customer i is δ,i Take 1, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When 1 is taken, package 1 is pushed to customer i; When the time difference judgment result T of customer i is δ,i Take 1, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 0, package 2 is pushed to customer i; When the time difference judgment result T of customer i is δ,i Take 0, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 1, package 3 is pushed to customer i; When the time difference judgment result T of customer i is δ,i Take 0, and the peak-shaving and valley-filling power difference judgment result of customer i is Q δ,i When the value is 0, package 4 is pushed to customer i.

9. A system for pushing electric retail packages based on customer portraits, based on a method for pushing electric retail packages based on customer portraits as claimed in any one of claims 1 to 8, characterized in that: The system comprises: The customer peak-shaving and valley-filling portrait module is used to build a customer peak-shaving and valley-filling portrait model, including a customer peak feature portrait model and a customer valley feature portrait model. Based on the DDPM algorithm, the customer's historical load is mined to extract customer peak features and customer valley features respectively, thus realizing the portrait of customer peak-shaving and valley-filling features. A power grid feature extraction module, used to extract power grid peak features and power grid valley features, wherein the power grid peak features include power grid power peak start time, power grid power peak end time and power grid peak power, and the power grid valley features include power grid power valley start time, power grid power valley end time and power grid valley power; The peak shaving and valley filling time difference module is used to construct a peak shaving and valley filling time difference model to calculate the peak shaving and valley filling time difference, including a peak value time difference model and a valley value time difference model, which are used to calculate the peak value time difference and the valley value time difference respectively; The peak shaving and valley filling electricity difference module is used to construct a peak shaving and valley filling electricity difference model to calculate the peak shaving and valley filling electricity difference, including a peak electricity difference model and a valley electricity difference model, to calculate the peak electricity difference and the valley electricity difference respectively; The electricity retail package formulation module is used to formulate four electricity retail packages according to the difference in peak shaving and valley filling time and the difference in peak shaving and valley filling electricity volume; The electricity retail package judgment module is used to build an electricity retail package judgment model based on GPT, including a peak shaving and valley filling time difference judgment model and a peak shaving and valley filling electricity quantity difference judgment model, to judge the customer's peak shaving and valley filling time difference and peak shaving and valley filling electricity quantity difference; the electricity retail package push module is used to select one of packages one to four to push to the customer based on the judgment results of GPT.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of a method for pushing electricity retail packages based on customer portraits as described in any one of claims 1 to 8 are implemented.

Images