Electricity utilization strategy intelligent optimization method and system based on big customer value mining, and medium

Abstract

The invention discloses a power utilization strategy intelligent optimization method based on big customer value mining. The intelligent optimization method in the electric power field comprises the following steps: establishing an evaluation dimension of a big customer value; acquiring electricity consumption historical data of the users under the evaluation dimension, and performing clustering analysis by adopting a badger optimized fuzzy C-means algorithm HBP-FCM to obtain electricity consumption characteristic models of various users; analyzing characteristic attributes of different types of customers based on the electricity utilization characteristic model; and establishing an electricity charge analysis model according to the characteristic attributes, and providing an optimized electricity utilization strategy. According to the intelligent optimization method in the power field provided by the invention, the electricity utilization cost of the customer is reduced, and the customer is helped to maintain long-term production and operation activities; meanwhile, the loss risk of high-quality customers is reduced, peak load shifting is facilitated, the stability of a power grid is improved, energy conservation and efficiency improvement are achieved, economic and social benefits of an electric power company are improved, and the win-win situation of mutual benefits and mutual benefits is achieved.

Description

Technical Field

[0001] The present invention relates to the electric power field, and in particular to an intelligent optimization method, system and medium for power consumption strategy based on key customer value mining in the electric power field. Background Art

[0002] As a special energy source, electric energy occupies a pivotal position in the energy market, and the power industry is a pillar industry that concerns the lifeline of the national economy. With the continuous development of the world economy and the intensification of global economic integration, various industries are carrying out marketing activities, and a global marketing system has emerged. The traditional vertically integrated business structure has been broken, and the main body of power generation and supply has been completely separated. The investment channels and entities of power resources of power generation enterprises tend to be diversified, and the supervision mechanisms such as bidding for access to the grid by independent power plants and fair competition are gradually being improved. For power grid companies, the gradual maturity of the operation of the simulated market system and the inclination towards alternative energy sources for electricity will make it possible for the demand for electricity to be relatively abundant. With the development of the economy, power grid companies can only continuously improve their economic benefits, enable them to survive and develop while enhancing their own social value and promoting social progress, by actively participating in market competition, attaching importance to customer resources, and carrying out effective marketing activities based on their actual conditions.

[0003] At present, the changes in the power system continue to deepen the power market. Power grid companies must abandon the traditional marketing model, continuously develop and occupy the market, actively participate in market competition, and attach importance to customer resources in order to achieve sustainable development of enterprises. The choice of marketing strategy oriented to customer demand has become the key to the future survival and development of power supply enterprises. Through the investigation and analysis of the power sales market of power supply enterprises, it is not difficult to find that the power purchase amount of 20% of the power purchase customers of most power supply companies accounts for 80% of the company's total power sales. According to the "20 / 80" theory, this type of customer is the customer who creates most of the profits for the power supply company, that is, it is a major customer with great significance to the power supply company. For power supply companies, the special status of major power customers makes it have a direct impact on the company's economic benefits while also causing indirect social impact. Therefore, how to carry out major customer marketing, provide them with high-quality services, and maintain a close partnership with them is of great practical significance for the sustainable operation of power supply companies, improving economic benefits and social value, actively fulfilling social responsibilities, and thus better serving the local area. Summary of the invention

[0004] In view of the above problems, the present invention provides an intelligent optimization method for power consumption strategy based on key customer value mining.

[0005] Establish the evaluation dimension of key customer value;

[0006] Obtain the historical electricity consumption data of users under this evaluation dimension, use the Honey Badger optimized fuzzy C-means algorithm HBA-FCM to perform cluster analysis, and obtain the electricity consumption characteristic model of each type of user;

[0007] Analyze the characteristic attributes of different types of customers based on the electricity consumption characteristic model;

[0008] Establish an electricity cost analysis model based on characteristic attributes and provide an optimized electricity consumption strategy.

[0009] Furthermore, the evaluation dimensions include the user's annual total electricity consumption, electricity load rate, electricity consumption growth rate, and electricity fee recovery timeliness rate.

[0010] Furthermore, the establishment of the electricity consumption characteristic model specifically includes:

[0011] Standardize the user's electricity consumption history data to obtain the electricity consumption information matrix;

[0012] Set the maximum number of clusters M and preset the number of clusters m = 2, use the HBA-FCM algorithm to solve, and obtain the clustering objective function J when the number of clusters is m m , fuzzy membership matrix U and cluster center C, and calculate the DBI index through the classification results;

[0013] Determine whether the clustering index DBI reaches the minimum value, take the number of clusters m and its clustering results under the minimum value and output them, otherwise increase the number of clusters and use the HBA-FCM algorithm again to solve;

[0014] The cluster center of the optimal clustering result is used as the power consumption characteristic model of each type of user, and curves are drawn and change characteristics are calculated to show their change rules.

[0015] Furthermore, the historical electricity consumption data is standardized using the Z-Score value, namely:

[0016]

[0017] Z ′ =[Z1 ′ ,Z2 ′ ,...,Z ′ D ]

[0018] Z t ′ is the value of the load data after standardization at time t. If the sampling interval is set to 15 minutes and the sampling time is 24 hours, then D = 96; f meav (Z t ) is Z at time t t The mean value of std (Z t ) is Z at time tt The standard deviation of Z ′ ∈N user ×D is the user load data set, N user is the number of users, Z ′ as clustering input.

[0019] Furthermore, the DBI indicator is obtained through the following steps:

[0020] 1) Initialize the maximum number of HBA iterations t max , population size n, and parameters λ and β, honey badger position y, i.e., the initial center c of the FCM algorithm; preset the number of FCM clusters m, the maximum number of iterations G, the allowable error e, and input the power data to be clustered;

[0021] y i =lb i +r1×(ub i -lb i ),r1∈[0,1]

[0022] Where, the position y of the honey bird in the i-th iteration is i refers to the candidate solutions in the N population, and lb i andub i are the lower and upper bounds of the search domain, respectively, and r1 is a random number between 0 and 1;

[0023] 2) Use the FCM algorithm for clustering and find the objective function J through iterative solution m Minimum value of:

[0024]

[0025] In the formula, m is the number of clusters, N is the number of input data, and x i is the i-th input data, c j represents the jth cluster center, which has the same form as the cluster data, u ij Represents x i With c j The greater the membership, the closer the relationship is, and the sum of the memberships of different cluster centers of the same data is 1; ‖·‖ is the Euclidean norm;

[0026] By repeatedly clustering, we can calculate J m Find the minimum value. When the difference between two iterations is within the set error range, the final result can be output. In this process, u also needs to be updated. ij With c j , to obtain the minimum J m , the specific calculation formula is:

[0027]

[0028] U = {u ij}

[0029] where c k represents the k-th cluster center;

[0030] 3) Determine whether the maximum number of iterations is reached, or whether J m satisfies the error condition |J m (t) - J m (t - 1)| < e. If the condition is satisfied, output the clustering objective function value; otherwise, go to step 2);

[0031] 4) Update the honey badger position y new , update the prey position value according to the objective function, and record the global optimal position. Specifically, it includes

[0032] a. Digging stage:

[0033] y new = y prey + FβI i y prey + Fr3αd i |cos(2πr4) × [1 - cos2πr5]|

[0034] where refers to the new position of honey badger i, y new is the position of the target, which is the optimal position found so far, i.e., the global optimal position; the foraging parameter β represents the foraging ability of the honey badger and is always greater than 1. r3, r4, and r5 are all random numbers between 0 and 1. F is used as a flag to change the search direction and is determined by the following formula:

[0035]

[0036] where r6 is a random number between 0 and 1;

[0037] b. Honey stage:

[0038] y new = y prey + Fr7αd i

[0039] where r7 is a random number between 0 and 1;

[0040] 5) Return to step 2) until the number of iterations is greater than the given threshold. According to the final global optimal position of the honey badger, calculate the fuzzy membership matrix U under this condition, determine the classification result, and output it.

[0041] Furthermore, whether the clustering index DBI reaches the minimum value is judged by the following formula

[0042]

[0043] In the formula, S i , S j Indicates c i 、c j The average intra-class distance of the corresponding cluster, N(A j ) represents cluster A j The number of samples contained in .

[0044] Furthermore, based on the D m categories of customers output by the cluster analysis, the attributes of customers in each category are analyzed.

[0045] Furthermore, according to different types of customer groups, based on their electricity load values ​​and electricity consumption periods, an optimized electricity consumption strategy is provided, that is, the electricity cost analysis model includes a basic electricity cost analysis model and a time-sharing model;

[0046] Basic electricity cost analysis model

[0047] Based on the historical electricity consumption data, the extrapolation method is used to obtain the monthly maximum load utilization hours for the next few months. By predicting the electricity consumption of major customers, the monthly maximum load forecast value is calculated.

[0048] According to the time series regression method, the basic electricity charges charged according to demand under each approved value are exhaustively enumerated.

[0049] Based on the actual situation on site, the basic electricity fee charged according to capacity is calculated based on the premise that the maximum load can be met after the transformer is suspended or reduced in capacity.

[0050] Under the premise that the actual demand can be calculated, according to the electricity price and billing rules, we can get:

[0051] If the approved demand is ≤ 67% capacity, charging based on the approved demand will result in a lower base rate.

[0052] If the rated demand is > 67% capacity, then capacity-based charging will result in a lower base rate.

[0053] When actual demand is < 40% capacity, charging according to actual demand will result in a lower basic electricity rate.

[0054] When 40% capacity < actual demand < 70% capacity, charging according to approved demand will result in a lower basic electricity rate.

[0055] 2. Time-sharing model

[0056] In order to achieve the profit and loss balance of peak and valley electricity prices and achieve the purpose of peak shifting, the balance point should meet the following conditions:

[0057]

[0058] Where P is the electricity price during peak, sharp, flat and valley periods; Y is the proportion of electricity consumption during each period.

[0059] According to the current electricity price floating coefficient, the ratio of the sum of peak and peak electricity consumption to the valley electricity consumption is obtained. The average price for users to use time-sharing electricity is the same as the average price for normal periods. If the ratio is greater than this value, the price for using time-sharing electricity will be higher than the average price. The user's historical time-sharing electricity consumption is analyzed and judged, and based on the ratio of the user's peak and peak electricity consumption to the ratio, it is recommended that the user arrange production shifts reasonably.

[0060] According to another aspect of the present invention, there is also provided a power consumption strategy intelligent optimization system based on key customer value mining, comprising:

[0061] Evaluation dimension establishment unit, establish the evaluation dimension of key customer value;

[0062] The cluster analysis unit obtains the electricity consumption history data of users under the evaluation dimension, and uses the Honey Badger optimized fuzzy C-means algorithm HBA-FCM to perform cluster analysis to obtain the electricity consumption characteristic model of each type of user;

[0063] The characteristic attribute analysis unit analyzes the characteristic attributes of different types of customers based on the electricity consumption characteristic model;

[0064] The electricity consumption strategy generation unit establishes an electricity cost analysis model based on characteristic attributes and provides an optimized electricity consumption strategy.

[0065] A computer storage medium, wherein the computer readable storage medium stores a computer program, and when the computer program is executed by a processor, any of the above methods is implemented.

[0066] Beneficial effects of the present invention: The present invention deeply explores the value of key customers of power supply companies, provides differentiated marketing strategies, reduces customers' electricity costs, and improves the economic and social benefits of power companies. Taking the electricity service of key customers as the research object, the fuzzy C-means algorithm based on honey badger optimization is used to efficiently classify power customers, and effective service marketing strategies are developed based on the clustering results. This will help power supply companies improve their market competitiveness, help expand the market scale, enable them to adapt to the development needs after the power reform, and allow power supply companies to seize opportunities and play to their strengths when facing the objective reality of the increasingly open power marketing market. At the same time, it provides power supply companies with good economic benefits and a positive social image, which has important theoretical and practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a flow chart of an intelligent optimization method for power consumption strategy for big customer value mining in the power field provided by the present invention.

[0068] Figure 2This is the overall process of the method provided by the present invention.

[0069] Figure 3 This is the clustering result diagram provided by the present invention. DETAILED DESCRIPTION

[0070] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0071] The technical solution of the present invention to solve the above technical problems is as follows:

[0072] The present invention provides an intelligent optimization method for power consumption strategy based on key customer value mining in the power field, referring to Figure 1 As shown, the intelligent optimization method in the power field includes:

[0073] Segment the value attributes of key customers and establish evaluation dimensions;

[0074] At present, there are two main ways for power supply companies to classify users. One is to implement electricity price segmentation according to different types of electricity consumption. This market segmentation standard is conducive to user management and has strong operability. Another way is to segment according to various industries of the national economy. In order to better meet the interests of customers, the above classification method can be further improved, and the power market can be segmented at two levels. The first level is to divide users into living users and production users based on the purpose of electricity consumption, and the second level is to further segment living users and production users respectively. Since production users are basically large customers of various power supply companies and have strong differences, power supply companies should focus on production users in market segmentation. The basic dimensions of production user segmentation mainly include industry, electricity price, power supply area, voltage level, power quality, power consumption, user nature, credit rating, etc.

[0075] The present invention takes production users as the main research object, and takes their annual total electricity consumption (electricity consumption), electricity load rate (volatility), electricity consumption growth rate (growth potential), and electricity fee recovery timeliness rate (creditworthiness) as evaluation indicators to form evaluation dimensions, as shown in Table 1.

[0076] Enterprise Number Power Consumption stability Growth potential Creditworthiness 1 396863 0.93 0.24 0.75 2 110690 0.62 -0.08 1 3 77091 0.66 -0.03 1 4 50779 0.45 0.02 0.95 5 80527 0.3 0.70 1 6 27345 0.80 - 1 7 13760 0.57 0.20 1 8 11254 0.52 0.13 1 9 10263 0.49 6 0.96 10 9751 0.54 0.29 1 11 7115 0.80 1.77 1 12 10542 0.52 0.08 1

[0077] Table 1

[0078] Make full use of the data in the existing system and determine the dimensions of value evaluation through automatic extraction or manual entry.

[0079] Honey Badger-optimized fuzzy C-means algorithm (HBA-FCM) is used to analyze user electricity consumption characteristics and screen high-quality customers;

[0080] According to the evaluation dimensions, we focus on analyzing the top 12 customers in terms of electricity consumption. The historical data of users' annual total electricity consumption (electricity consumption), electricity load rate (volatility), electricity growth rate (growth potential), and timeliness of electricity fee recovery rate (creditworthiness) are input into the fuzzy C-means algorithm based on Honey Badger optimization for cluster analysis.

[0081] Specific process based on HBA-FCM algorithm

[0082] In order to effectively solve the problem that the initial center of the traditional fuzzy C-means algorithm is randomly generated and the clustering result falls into the local optimum, the classification effect of power information is improved by combining it with the honey badger algorithm. This paper proposes a honey badger optimized fuzzy C-means classification algorithm (HBA-FCM) to improve clustering accuracy. The specific steps are as follows Figure 2 As shown:

[0083] 1. First, collect user electricity consumption information, standardize it, and obtain the electricity consumption information matrix.

[0084] There are various types of users in the distribution area, and the electricity consumption attributes of different users vary greatly. Therefore, standardization processing is required before cluster analysis. The present invention adopts Z-Score value standardization, that is:

[0085]

[0086] Z ′ =[Z1 ′ ,Z2 ′ ,...,Z ′ D ]

[0087] Z t ′ is the value of the load data after standardization at time t. If the sampling interval is set to 15 minutes and the sampling time is 24 hours, then D = 96; f mean (Z t ) is Z at time t t The mean value of std (Z t ) is Z at time t t The standard deviation of Z ′ ∈N user ×D is the user load data set, N user is the number of users, Z ′ as clustering input.

[0088] 2. Then, set the maximum number of clusters M and preset the number of clusters m = 2, use the HBA-FCM algorithm to solve, and obtain the clustering objective function J when the number of clusters is m m , fuzzy membership matrix U and cluster center C, and calculate the DBI index through the classification results.

[0089] (1) Initialize the maximum number of HBA iterations t max , population size n, and parameters λ and β, honey badger position y, i.e., the initial center c of the FCM algorithm; preset the number of FCM clusters m, the maximum number of iterations G, the allowable error e, and input the power data to be clustered;

[0090] y i =lb i +r1×(ub i -lb i ),r1∈[0,1]

[0091] Where, the position y of the honey bird in the i-th iteration is i refers to the candidate solutions in the N population, and lb i andub i are the lower and upper bounds of the search domain respectively, and r1 is a random number between 0 and 1.

[0092] (2) Use the FCM algorithm for clustering and find the objective function J through iterative solution m Minimum value of:

[0093]

[0094] In the formula, m is the number of clusters, N is the number of input data, and x i is the i-th input data, c j represents the jth cluster center, which has the same form as the cluster data, u ij Represents x i With c j The greater the membership, the closer the relationship is, and the sum of the memberships of different cluster centers of the same data is 1. ‖·‖ is the Euclidean norm.

[0095] By repeatedly clustering, we can calculate J m Find the minimum value. When the difference between two iterations is within a certain error range, the final result can be output. In this process, u ij With c j , to obtain the minimum J m , the specific calculation formula is:

[0096]

[0097] U={u ij}

[0098] In the formula, c k represents the kth cluster center.

[0099] (3) Determine whether the maximum number of iterations has been reached, or whether the number of iterations before and after the tth iteration is Jm Whether the error condition |J m (t)-J m (t - 1)| < e. If the condition is satisfied, output the clustering objective function value; otherwise, go to step (2).

[0100] (4) Update the honey badger position y new , update the prey position value according to the objective function, and record the global optimal position;

[0101] Excavation stage:

[0102] y new = y prey + FβI i y prey + Fr3αd i |cos(2πr4)×[1 - cos2πr5]|

[0103] In the formula, refers to the new position of honey badger i, and y new is the position of the target, which is the optimal position found so far, that is, the global optimal position. The foraging parameter β represents the foraging ability of the honey badger and is always greater than 1. r3, r4, and r5 are all random numbers between 0 and 1. F is used as a sign to change the search direction and is determined by the following formula:

[0104]

[0105] In the formula, r6 is a random number between 0 and 1.

[0106] Honey stage:

[0107] y new = y prey + Fr7αd i

[0108] In the formula, r7 is a random number between 0 and 1.

[0109] (5) Return to step (2) until the number of iterations is greater than the given threshold. According to the final global optimal position of the honey badger, calculate the fuzzy membership matrix U under this condition, determine the classification result and output it.

[0110] 3. Judge whether the clustering index DBI reaches the minimum value. Output the number of clusters m and its clustering result in the case of the minimum value; otherwise, increase the number of clusters and use the HBA - FCM algorithm to solve again.

[0111]

[0112] In the formula, S i 、S j represent c i 、c jThe average intra-class distance of the corresponding cluster, N(A j ) represents cluster A j The number of samples contained in .

[0113] Since the larger the index value, the smaller the distance between classes and the greater the dispersion of objects within the class, that is, the worse the clustering effect, the DBI index corresponding to different m is obtained by changing the number of clusters m, and the m corresponding to the DBI minimum value is selected as the optimal clustering number, that is, the adaptive minimum principle.

[0114] 4. Finally, the cluster center of the optimal clustering result is used as the electricity consumption characteristic model of each type of user, and curves are drawn and change characteristics are calculated to show their changing rules.

[0115] Experimental verification:

[0116] The data source is an electricity data set obtained from a smart distribution network-smart city project. The network is connected to distributed power sources, electric vehicles, and flexible loads, and the load characteristics are complex. 248 users were selected to provide daily electricity load data for user electricity consumption characteristics analysis, and data was selected at 1h intervals. First, the distribution network user electricity consumption data was processed and standardized using the Z-Score value to form a characteristic array of electricity consumption corresponding to each distribution network user. The maximum number of iterations of the FCM algorithm is set to 40, the maximum number of iterations of the HBA is set to 100, and the maximum number of clusters m is set to 5. The electricity consumption characteristic array is input into the HBA-FCM algorithm for calculation. The clustering results are shown in the figure below: Figure 3 shown.

[0117] Analyze the characteristic attributes of different categories of customers;

[0118] according to Figure 3 The four categories of medium output are divided into four groups, namely, high electricity and low credit group (category 1), medium electricity and negative growth group (category 2), high growth and low volatility group (category 3), and low growth and high volatility group (category 4). The characteristic attributes of different categories of customers are as follows:

[0119] 1. High power consumption and low credit customer group:

[0120] This group of enterprises belongs to the non-ferrous metal smelting industry, with an annual electricity consumption of nearly 4 billion kWh, accounting for one-third of the total electricity consumption in the region. This fully shows that the companies in this group are super large customers of the power supply company. The company has implemented direct purchase of electricity from large customers since October 2009. Since 2014, affected by the regulation of overcapacity industries and market environment, the operating situation of high-energy-consuming industries has deteriorated, funds are tight, electricity bills are difficult to pay, and there have been many arrears, resulting in a reduction in its credit rating, which has posed a serious challenge to the company's electricity bill recovery management. It reflects that the company has serious risks in customer credit and electricity bill recovery. In addition, the annual electricity load rate of enterprises in this group is stable, and the growth of electricity consumption is slow, reflecting that the company has made great contributions to the safe and stable operation of the power grid and the increase of the company's profits. The power supply company should formulate marketing strategies from aspects such as improving customer credit rating and increasing the timeliness of electricity bill recovery to ensure the market share of this customer group.

[0121] 2. CEC negative growth customer group:

[0122] This group is a ferrous metal smelting enterprise. The total electricity consumption is nearly 1.9 billion kWh, and the electricity consumption growth rate shows a negative growth distribution. The main reason is that due to the impact of the economic situation, orders have decreased and most equipment has been shut down, resulting in a decrease in electricity consumption. This shows that the company has a risk of customer loss. Due to the large proportion of their electricity consumption, the shrinkage or loss of this group of customers will have a serious impact on the operating income of the power supply company. At the same time, the electricity load rate of this group of customers fluctuates greatly, but their credit rating is very high, and they can automatically complete the electricity bill settlement business. The power supply company should analyze the reasons for the shrinkage of its production and operation and the large fluctuations in load from the perspective of the company's production and operation, and formulate a marketing strategy from the perspective of improving customer loyalty.

[0123] 3. High growth and low volatility customer group: This group of major customers mainly come from the chemical industry such as PetroChina Petrochemical Branch, Da Yichen and Cement Co., Ltd. Among them, PetroChina Petrochemical Branch has maintained a relatively high growth rate of electricity consumption, and the new electricity load of Da Yichen has brought certain economic benefits to the power supply company. In the future, with the economic growth of the petrochemical industry, the growth of electricity consumption will be further driven. At the same time, this group of customer groups has a high stability of electricity consumption and maintains a fairly high credit rating. All these fully demonstrate that this customer group is an important customer group for the power supply company to maintain the growth of electricity sales in the future. The power supply company should formulate corresponding marketing strategies from the demand side management and strive to expand the electricity market.

[0124] 4. Low-growth and high-volatility customer group: This group of large customers are all small and medium-sized industrial enterprises, and the industry is mainly concentrated in non-metallic mineral smelting, which are traditional industrial enterprises. Since this group of customer groups has a relatively low proportion of electricity consumption and a low growth rate of electricity consumption, it shows that it has weak development potential. The low stability of electricity consumption indicates that the company is highly unstable in the production and operation process. Due to its small production scale and the fact that it is a traditional high-energy-consuming enterprise, its production and operation are easily affected by the macro-economy, causing business shrinkage or suspension of production, which can easily cause economic losses to the power supply company. The power supply company is at great risk of losing customers. However, since it maintains a high credit rating, its electricity consumption accounts for about one-tenth of the total electricity sales of the power supply company, so the company should formulate a marketing strategy for general service guarantees for this group of customers.

[0125] Establish electricity cost analysis model and provide optimization strategy for electricity consumption;

[0126] Provide optimized electricity consumption strategies based on different customer groups, their electricity load values ​​and electricity consumption periods.

[0127] 1. Basic electricity cost analysis model

[0128] Based on the historical electricity consumption data, the extrapolation method is used to obtain the monthly maximum load utilization hours for the next few months. By predicting the electricity consumption of major customers, the monthly maximum load forecast value is calculated. According to the time series regression method, the basic electricity charges charged according to demand are exhaustively enumerated under each approved value. Combined with the actual situation on site, the basic electricity charges charged according to capacity are calculated based on the premise that the maximum load is met after the transformer is suspended or reduced in capacity. Under the premise that the actual demand can be measured, according to the electricity price and billing rules, analysis can be obtained:

[0129] If the approved demand is ≤ 67% capacity, charging based on the approved demand will result in a lower base rate.

[0130] If the rated demand is > 67% capacity, then capacity-based charging will result in a lower base rate.

[0131] When actual demand is < 40% capacity, charging according to actual demand will result in a lower basic electricity rate.

[0132] When 40% capacity < actual demand < 70% capacity, charging according to approved demand will result in a lower basic electricity rate.

[0133] 2. Time-sharing model

[0134] In order to achieve the profit and loss balance of peak and valley electricity prices and achieve the purpose of peak shifting, the balance point should meet the following conditions:

[0135]

[0136] Where P is the electricity price during peak, sharp, flat and valley periods; Y is the proportion of electricity consumption during each period.

[0137] According to the current electricity price floating coefficient, if the ratio of the sum of peak and high-peak electricity consumption to the valley electricity consumption is obtained, the average price of the user's time-sharing will be the same as the flat electricity price. If it is greater than the ratio, the time-sharing electricity price will be greater than the flat electricity price. Analyze and judge the user's historical time-sharing electricity consumption, and recommend users to arrange production shifts reasonably based on the ratio of the peak and peak electricity consumption that has occurred to the user.

[0138] Beneficial effects of the present invention: The present invention deeply explores the value of key customers of power supply companies, provides differentiated marketing strategies, reduces customers' electricity costs, and improves the economic and social benefits of power companies. Taking the electricity service of key customers as the research object, the fuzzy C-means algorithm based on honey badger optimization is used to efficiently classify power customers, and effective service marketing strategies are developed based on the clustering results. This will help power supply companies improve their market competitiveness, help expand the market scale, enable them to adapt to the development needs after the power reform, and allow power supply companies to seize opportunities and play to their strengths when facing the objective reality of the increasingly open power marketing market. At the same time, it provides power supply companies with good economic benefits and a positive social image, which has important theoretical and practical value.

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

Claims

1. An intelligent optimization method for power consumption strategy based on key customer value mining, characterized in that: Establish the evaluation dimension of key customer value; Obtain the historical electricity consumption data of users under this evaluation dimension, use the Honey Badger optimized fuzzy C-means algorithm HBA-FCM to perform cluster analysis, and obtain the electricity consumption characteristic model of each type of user; Analyze the characteristic attributes of different types of customers based on the electricity consumption characteristic model; Establish an electricity cost analysis model based on characteristic attributes and provide an optimized electricity consumption strategy.

2. The intelligent optimization method in the electric power field according to claim 1, characterized in that: The evaluation dimensions include the user's annual total electricity consumption, electricity load rate, electricity consumption growth rate, and electricity fee recovery timeliness rate.

3. The intelligent optimization method in the electric power field according to claim 1, characterized in that: The establishment of the electricity consumption characteristic model specifically includes: Standardize the user's electricity consumption history data to obtain the electricity consumption information matrix; Set the maximum number of clusters M and preset the number of clusters m = 2, use the HBA-FCM algorithm to solve, and obtain the clustering objective function J when the number of clusters is m m , fuzzy membership matrix U and cluster center C, and calculate the DBI index through the classification results; Determine whether the clustering index DBI reaches the minimum value, take the number of clusters m and its clustering results under the minimum value and output them, otherwise increase the number of clusters and use the HBA-FCM algorithm again to solve; The cluster center of the optimal clustering result is used as the electricity consumption characteristic model of each type of user, and curve drawing and change characteristic calculation are performed.

4. The intelligent optimization method in the electric power field according to claim 3, characterized in that: For the historical electricity consumption data, the Z-Score value is used for standardization, that is: WITH ′ =[Z1 ′ ,Z2 ′ ,...,WITH ′ D ] Z t ′ is the value of the load data after standardization at time t. If the sampling interval is set to 15 minutes and the sampling time is 24 hours, then D = 96; f mean (Z t ) is Z at time t t The mean value of std (Z t ) is Z at time t t The standard deviation of Z′∈N user ×D is the user load data set, N user is the number of users, and Z′ is used as the clustering input.

5. The intelligent optimization method in the electric power field according to claim 1, characterized in that: The DBI indicator is obtained through the following steps: 1) Initialize the maximum number of HBA iterations t max , population size n, and parameters λ and β, the honey badger position y is the initial center c of the FCM algorithm; Preset the FCM clustering number m, the maximum number of iterations G, the allowable error e, and input the power data to be clustered; y i =lb i +r1×(ub i -lb i ),r1∈[0,1] Where, the position y of the honey bird in the i-th iteration is i refers to the candidate solutions in the N population, and lb i andub i are the lower and upper bounds of the search domain, respectively, and r1 is a random number between 0 and 1; 2) Use the FCM algorithm for clustering and find the objective function J through iterative solution m Minimum value of: In the formula, m is the number of clusters, N is the number of input data, and x i is the i-th input data, c j represents the jth cluster center, which has the same form as the cluster data, u ij Represents x i With c j The greater the membership, the closer the relationship is, and the sum of the memberships of different cluster centers of the same data is 1; ||·|| is the Euclidean norm; By repeatedly clustering, we can calculate J m Find the minimum value. When the difference between two iterations is within the set error range, the final result can be output; You also need to update u during this process ij With c j , to obtain the minimum J m , the specific calculation formula is: In={in ij } In the formula, c k represents the kth cluster center; 3) Determine whether the maximum number of iterations is reached, or whether J before and after the t-th iteration m satisfies the error condition |J m (t) - J m (t - 1)| < e. If the condition is satisfied, output the clustering objective function value; otherwise, go to step 2). 4) Update the honey badger position y new , update the prey position value according to the objective function and record the global optimal position, including: a. Mining stage: y new =y prey +FβI i y prey +Fr3αd i |cos(2πr4)×[1-cos2πr5]| Where, refers to the new position of honey badger i, y new is the position of the target, which is the best position found so far, that is, the global optimal position; the foraging parameter β represents the foraging ability of the honey badger, which is always greater than 1; r3, r4 and r5 are all random numbers from 0 to 1, and F is used as a sign to change the search direction, which is determined by the following formula: Where r6 is a random number between 0 and 1; b. Honey stage: and new =and prey +Fr7αd i Where r7 is a random number between 0 and 1; 5) Return to step 2) until the number of iterations is greater than the given threshold. According to the final global optimal position of the honey badger, calculate the fuzzy membership matrix U under this condition, determine the classification result and output it.

6. The intelligent optimization method in the electric power field according to claim 1, characterized in that: Whether the clustering index DBI reaches the minimum value is determined by the following formula: In the formula, S i , S j Indicates c i 、c j The average intra-class distance of the corresponding cluster, N(A j ) represents cluster A j The number of samples contained in .

7. The intelligent optimization method in the electric power field according to claim 1, characterized in that: Based on the cluster analysis output of Dm categories of customers, the customer attributes of each category are analyzed.

8. The intelligent optimization method in the electric power field according to claim 1, characterized in that: According to different types of customer groups, according to their electricity load values ​​and electricity consumption periods, an optimized electricity consumption strategy is provided, that is, the electricity fee analysis model includes a basic electricity fee analysis model and a time-sharing model; 1) Basic electricity cost analysis model Based on the historical electricity consumption data, the extrapolation method is used to obtain the monthly maximum load utilization hours for the next few months. By predicting the electricity consumption of major customers, the monthly maximum load forecast value is calculated; According to the time series regression method, the basic electricity charges charged according to demand under each approved value are exhaustively enumerated; 2) Time-sharing model In order to achieve the profit and loss balance of peak and valley electricity prices and achieve the purpose of peak shifting, the balance point should meet the following conditions: In the formula, P is the electricity price in each period, such as peak, peak, flat and valley; Y is the proportion of electricity consumption in each period; According to the current electricity price floating coefficient, the ratio of the sum of peak and peak electricity consumption to the valley electricity consumption is obtained. The average price for users to use time-sharing electricity is the same as the normal electricity price. If it is greater than this ratio, the time-sharing electricity price will be higher than the normal electricity price. The user's historical time-sharing electricity consumption is analyzed and judged, and the user is advised to arrange production shifts reasonably based on the ratio of the user's peak and peak electricity consumption.

9. An intelligent optimization system for power consumption strategy based on key customer value mining, characterized in that: include, Evaluation dimension establishment unit, establish the evaluation dimension of key customer value; The cluster analysis unit obtains the electricity consumption history data of users under the evaluation dimension, and uses the Honey Badger optimized fuzzy C-means algorithm HBA-FCM to perform cluster analysis to obtain the electricity consumption characteristic model of each type of user; The characteristic attribute analysis unit analyzes the characteristic attributes of different types of customers based on the electricity consumption characteristic model; The electricity consumption strategy generation unit establishes an electricity cost analysis model based on characteristic attributes and provides an optimized electricity consumption strategy.

10. A computer storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images