A power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation

Abstract

The present invention discloses a power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation, and relates to a power supply service analysis method. At present, a single evaluation index is often used to evaluate the service of a power supply company, and the impact of various factors on the power supply service level is not considered more comprehensively. The present invention adopts the bottom-level evaluation index based on four aspects of customer demands, responsibility attribution, abnormal level, and service processing to establish a complete evaluation index system. The score distribution type is obtained through a double-layer clustering model, and the central score of each evaluation level under different types of distribution is determined. The evaluation index score is converted into an easy-to-understand evaluation level through a fuzzy evaluation method, and an intuitive evaluation is made on the power supply service quality of the power supply company. This customer-side power supply service analysis method based on structured work order data can quickly and effectively evaluate the service level of the power supply company according to the work order structured data, which is conducive to improving the customer-side power supply service level and accurately locating service weaknesses.

Description

Technical Field

[0001] The present invention relates to the field of power supply service evaluation, and in particular to a power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation. Background Art

[0002] As the reform of the power system continues to deepen, the power sales market is gradually liberalized, and the pressure on the quality and level of power supply services is increasing day by day. In order to continuously improve customer service capabilities, the Provincial Marketing Service Center has always insisted on being customer-centric, starting from customer demands, analyzing and mining hot businesses, quickly and effectively finding business shortcomings, and improving customer service quality. With the development and popularization of power customer service centers, 95598 work order data has become an important data source that fully reflects the quality of customer service; at the same time, with the continuous deepening of power system reform, continuously improving customer service capabilities is very important for power supply companies to fine-tune management and improve customer experience.

[0003] The existing power supply service quality evaluation mainly evaluates the power supply service quality through qualitative analysis such as questionnaire surveys, and conducts a comprehensive evaluation in combination with power grid operation data and safety indicators. Including the construction of a comprehensive evaluation model through methods such as hierarchical analysis method, fuzzy theory, and matter-element analysis model. On the one hand, the evaluation results of different power supply companies lack objectivity, and on the other hand, the evaluation results are affected by the survey samples. At present, most of the methods for power supply service evaluation only consider the subjective evaluation of the sample and the operation data of the overall power grid, without using the widely popular power customer service data, and considering the horizontal comparison between different power supply companies. In addition, most methods use a single evaluation indicator to evaluate the service of power supply companies, and fail to comprehensively consider the impact of various factors on the power supply service level. To prevent the occurrence of power supply service risks.

[0004] How to make full use of structured billing data and select scientific and effective evaluation methods to comprehensively evaluate the power supply service level is a problem that needs in-depth research in the current power supply service evaluation. It can be seen that the existing methods for power supply service evaluation need to be improved. Summary of the invention

[0005] The technical problem to be solved and the technical task proposed by the present invention are to improve and perfect the existing technical solutions, and provide a power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation, so as to comprehensively and accurately evaluate the power supply service; for this purpose, the present invention adopts the following technical solutions;

[0006] The method comprises the following steps:

[0007] 1) Input the structured data of 95598 work orders within a certain period of time;

[0008] Including types of demands, work order classification, user satisfaction, and whether the power supply company is responsible;

[0009] 2) Comprehensively consider the number of user demands, service responsibilities of power supply companies, work order change levels, and power supply service quality, and establish a power supply service level evaluation index system based on structured data;

[0010] 3) Calculate the index value of each power supply service business under various evaluation indicators, and form an evaluation decision matrix of the power supply service level based on multiple indicators;

[0011] 4) Based on the improved entropy weight method, an objective weighting model for power supply service level evaluation is established, and the entropy value and weight of each sub-business in the fields of marketing, operation and inspection, and infrastructure are obtained according to the objective weighting model. The sub-business includes business environment and electricity charges;

[0012] 5) Calculate and rank the scores of different power supply companies based on the improved TOPSIS method;

[0013] 6) Perform double-layer clustering based on the power supply service score to determine the cluster center of the score distribution;

[0014] 7) Confirm its fuzzy membership function and calculate the power supply service level membership matrix;

[0015]

[0016] In the formula, Indicates the index value r ip For state V j The membership degree, matrix A i The nth row in represents the single factor evaluation result of the nth indicator, which is a fuzzy subset of V;

[0017] 8) Determine the service level of the power supply company based on the principle of maximum subordination;

[0018] 9) Based on the result data of step 4), step 5), and step 8), use visualization to display the comprehensive analysis results of power supply services from the customer perspective and the analysis results of power supply services from the marketing service dimension. Display them in multiple dimensions, including time dimension, geographical dimension, and appeal dimension, accurately locate the shortcomings of marketing services, and provide reminders and warnings to facilitate the management of potential power supply service problems.

[0019] As the preferred technical means: in step 2), in the power supply service level evaluation index system, a number of underlying indicators for different service businesses of the power supply company are adopted at the indicator layer, including customer demand indicators, responsibility attribution indicators, work order change indicators, and service processing indicators; at the business layer, the hierarchical division is based on the power supply company's business; at the field layer, the power supply company's service business is divided into fields; in step 4), an objective weighting model for power supply service level evaluation is established based on the improved entropy weight method to determine the indicator weights at the indicator layer, business layer, and field layer.

[0020] As a preferred technical means: When using customer demand indicators to evaluate the power supply service level, based on the 95598 work orders are divided into three categories: complaints, opinions, and service application work orders, the customer demand indicators are defined as:

[0021]

[0022]

[0023] In the formula, S i is the score of the i-th type of work order, Ni is the number of i-th type of work orders per million households of the power supply company, and N i,min is the reference value of the number of work orders of the i-th category, α 1i is the correction coefficient of the i-th type of work order, 0<α 1i ≤1 and

[0024] As a preferred technical means: When the responsibility attribution index is used to evaluate the power supply service level, the causes of all customer demands are divided into three categories from the perspective of responsibility attribution: customer, power supply enterprise, and indirect personnel responsibility; and the responsibility of the power supply enterprise will cause a worse user experience than the other two reasons for responsibility, and it is also the direction that the power supply enterprise can optimize; therefore, the smaller the proportion of responsibility in all work orders is at the power supply enterprise, the better the performance of the power supply company, and the responsibility attribution index is defined as:

[0025]

[0026] Where N ZR N is the number of responsible work orders in the power supply company. ZR,min is the reference value of the number of responsible work orders, and N is the total number of work orders for this business.

[0027] As the preferred technical means: when the work order change index is used to evaluate the power supply service level, based on the fact that the work order traffic volume is a nonlinear time series data affected by external subjective factors including power load and weather, the number of work order changes is 95598, which is an abnormal growth of work orders outside the normal fluctuation. When the number of work order changes is too large, it indicates that the power supply company has abnormal conditions in this business area, resulting in a decline in the power supply service level; based on the average change, the normal range of the number of work orders is determined, and the work order change index is defined as:

[0028]

[0029] N warn =N -1 (1+G mean )(1+δ)

[0030] Where N warnis the threshold of abnormal quantity, N -1 is the number of work orders last month, G mean is the average work order growth rate, and δ is the confidence coefficient of work order changes.

[0031] As the preferred technical means: when the service processing index is used to evaluate the power supply service level, the processing result based on the power supply service application is an important part of the power supply service evaluation and an important basis for a good customer experience; among them, the service satisfaction rate of customer demands and the number of repeated work orders are important indicators reflecting the service level of the power supply company. The service satisfaction rate is the customer's feedback on the work order processing results. A high satisfaction rate indicates that the customers are generally satisfied with the processing results; the number of repeated work orders reflects that the same event has not been resolved in a timely manner. The more repeated work orders, the lower the efficiency of work order processing. The service processing indicators are defined as follows:

[0032]

[0033] In the formula, C is the service satisfaction rate, N repeat is the number of repeated work orders, N repeat,mean is the reference value of the number of repeated work orders, α 41 is the service satisfaction rate correction coefficient, α 42 is the correction coefficient of repeated work order quantity, satisfying α 41 +α 42 =1.

[0034] As a preferred technical means: in step 4), comprising:

[0035] 401) Establish an indicator score matrix:

[0036] In the index score matrix R consisting of N power supply companies to be evaluated and M indicators to be evaluated, the entropy value of the mth evaluation indicator can be defined as follows:

[0037]

[0038] Where: And assume that when f nm = 0, f nm lnf nm =0; where r nm is the calculated score of the mth indicator of the nth power supply company;

[0039] 402) Index weight calculation

[0040] Based on the determination of weights by the improved entropy weight method, the weights of power supply service evaluation indicators are not affected by individual extreme cases, so that relatively reasonable indicator weights are obtained. The weight calculation expression is as follows:

[0041]

[0042]

[0043]

[0044] In the formula, ω m is the weight of the mth indicator, ω 1m is the weight required by the traditional entropy weight method, ω 2m is the modified weight, H m is the entropy value of the mth indicator, is the average value of all indicator entropies that are not 1; β is the coefficient.

[0045] As a preferred technical means: in step 5), the steps include:

[0046] 501) Establish a power supply service evaluation matrix;

[0047] In the power supply service evaluation of N power supply companies and M evaluation indicators, the power supply service evaluation matrix R is expressed as:

[0048]

[0049] 502) Establishing a weighted power supply service evaluation matrix G;

[0050] G=(g nm ) N×M

[0051] In the formula, g nm =ω m r nm .

[0052] 503) Calculate similarity

[0053] The highest score of each indicator is taken as the positive ideal solution, and the lowest score of each indicator is taken as the negative ideal solution. Thus, the ideal solution vector can be obtained. and in m∈{1,2,…,M};

[0054] Thus, we can get the distance similarity C between the index scores of each power supply company and the positive ideal solution and the negative ideal solution. n1 Similarity with cosine C n2 , the expression is as follows;

[0055]

[0056]

[0057]

[0058] In the formula, is the Euclidean distance between the index score of the nth power supply company and the positive ideal solution, is the Euclidean distance between the index score of the nth power supply company and the negative ideal solution, C n1 is the distance similarity of the nth power supply company, C n2 is the cosine similarity between the nth power supply company and the positive ideal solution, G n is the indicator score vector of the nth power supply company;

[0059] Combining distance similarity and cosine similarity, we can get the relative similarity of the power supply company. The expression is as follows:

[0060] S n =γ1C n1 +(1-γ1)C n2

[0061] In the formula, γ1 is the correction coefficient.

[0062] As a preferred technical means: in step 7), the fuzzy membership function of the index is taken as a ridge function;

[0063] For different businesses under the service area i, calculate the fuzzy relationship of each business and combine the weight W required by each business i , and obtain the fuzzy evaluation vector B of service field i i as follows:

[0064]

[0065] Then, according to the fuzzy evaluation vector of each service area and the weight vector W of each service area, the final fuzzy evaluation result C of the power supply company can be obtained:

[0066]

[0067] Where C is the fuzzy evaluation vector of the power supply company, B = [B1, B2, ..., B l ] T , l is the service area to be evaluated, c Vi Indicates the membership degree corresponding to different evaluation levels;

[0068] As a preferred technical means: β is 35.35.

[0069] Beneficial effects: This technical solution proposes the bottom-level evaluation indicators based on customer demands, responsibility attribution, abnormal level, and service processing, and establishes a complete evaluation indicator system based on the service business architecture of the power supply company. A weight determination method based on the improved entropy weight method is proposed to overcome the problem that the indicator weights calculated by the traditional entropy weight method are inconsistent with the actual importance when the overall score is relatively average. On the basis of the ideal point method, a similarity correction method based on cosine similarity is introduced to make the power supply service evaluation results closer to the actual standard. Based on the two-layer structure of the evaluation score, the score distribution type is obtained through a two-layer clustering model and the center score of each evaluation level under different types of distribution is determined. The evaluation indicator score is converted into an easy-to-understand evaluation level through a fuzzy evaluation method, and the power supply service quality of the power supply company is intuitively evaluated. This customer-side power supply service analysis method based on structured work order data can quickly and effectively evaluate the service level of the power supply company according to the work order structured data, which is of great significance to improving the customer-side power supply service level and accurately locating service weaknesses. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 It is a flow chart of the present invention.

[0071] Figure 2 This is a diagram of the power supply service business system of the present invention.

[0072] Figure 3 This is a statistical chart of the comprehensive score of power supply services according to an embodiment of the present invention.

[0073] Figure 4 This is a diagram showing the comprehensive analysis interface of power supply services based on the customer perspective of the present invention.

[0074] Figure 5 This is a diagram showing the power supply service analysis interface of the marketing service dimension of the present invention. DETAILED DESCRIPTION

[0075] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings of the specification;

[0076] like Figure 1 As shown, the present invention comprises the following steps:

[0077] Step 1: Input the structured data of 95598 work orders within a certain period of time, including the type of demand, work order classification, user satisfaction, whether the power supply company is responsible, etc.

[0078] Step 2: Comprehensively consider the number of user demands, service responsibilities of power supply companies, work order change levels, and power supply service quality. Based on structured data, establish a power supply service level evaluation index system. At the index level, propose four underlying indicators for different service businesses of power supply companies, including customer demand indicators, responsibility attribution indicators, work order change indicators, and service processing indicators. At the business level, propose a hierarchical division based on the power supply company's business; at the field level, divide the power supply company's service business into fields. The specific indicator expressions are as follows:

[0079] a. Customer demand indicator R1:

[0080]

[0081]

[0082] In the formula, S i is the score of the i-th type of work order, N i N is the number of work orders of the i-th category of the power supply company per million households. i,min is the reference value of the number of work orders of the i-th category, α 1i is the correction coefficient of the i-th type of work order, 0<α 1i ≤1 and

[0083] b. Responsibility attribution indicators

[0084]

[0085] Where N ZR N is the number of responsible work orders in the power supply company. ZR,min is the reference value of the number of responsible work orders, and N is the total number of work orders for this business.

[0086] c. Work order change indicators

[0087]

[0088] N warn =N -1 (1+G mean )(1+δ)

[0089] Where N warn is the threshold of abnormal quantity, N -1 is the number of work orders last month, G mean is the average work order growth rate, δ is the confidence coefficient of work order change, which is set to 10% in this paper based on expert experience.

[0090] d. Service processing indicators

[0091]

[0092] In the formula, C is the service satisfaction rate, N repeat is the number of repeated work orders, N repeat,mean is the reference value of the number of repeated work orders, α 41 is the service satisfaction rate correction coefficient, α 42 is the correction coefficient of repeated work order quantity, satisfying α 41 +α 42 =1.

[0093] Step 3: Calculate the index value of each power supply service business under each evaluation index, and form an evaluation decision matrix R of the power supply service level according to the four indicators;

[0094]

[0095] Step 4, 4) Establish an objective weighting model for power supply service level evaluation based on the improved entropy weight method to determine the indicator weights at the indicator layer, business layer, and field layer; and obtain the entropy value and weight of each sub-business in the marketing, operation and inspection, and infrastructure business fields according to the objective weighting model. The sub-businesses include business environment and electricity charges;

[0096]

[0097]

[0098]

[0099] In the formula, ω m is the weight of the mth indicator, ω 1m is the weight required by the traditional entropy weight method, ω 2m is the modified weight, H m is the entropy value of the mth indicator, It is the average value of all index entropies that are not 1. After engineering verification, when β is 35.35, the weight value obtained under typical circumstances is more objective and reasonable.

[0100] Step 5: Calculate and rank the scores of different power supply companies based on the improved TOPSIS method

[0101]

[0102]

[0103]

[0104] In the formula, is the Euclidean distance between the index score of the nth power supply company and the positive ideal solution, is the Euclidean distance between the index score of the nth power supply company and the negative ideal solution, C n1 is the distance similarity of the nth power supply company, C n2is the cosine similarity between the nth power supply company and the positive ideal solution, G n is the indicator score vector of the nth power supply company.

[0105] Combining distance similarity and cosine similarity, we can get the relative similarity of the power supply company. The expression is as follows:

[0106] S n =γ1C n1 +(1-γ1)C n2

[0107] In the formula, γ1 is the correction coefficient.

[0108] Step 6: Perform double-layer clustering based on the power supply service score to determine the cluster center of the score distribution

[0109] Step 7: Confirm the fuzzy membership function according to the cluster center and calculate the power supply service level membership matrix.

[0110]

[0111] In the formula, Indicates the index value r ip For state V j The membership degree, matrix A i The nth row in represents the single factor evaluation result of the nth indicator, which is a fuzzy subset of V.

[0112] Step 8: Determine the final level V of the power supply company's service level based on the maximum membership principle.

[0113] In order to further understand the present invention, the following data related to 95,598 work orders in Zhejiang Province are taken as the research object, and the evaluation object is the power supply companies in eleven cities in Zhejiang Province. The work order data in September 2020 are used as the evaluation method verification, involving a total of 66,249 work orders to be evaluated and more than 200,000 historical work order data.

[0114] The structure of the electricity supply service business is as follows: Figure 2 As shown, there are 12 service areas and 34 specific businesses. The third-level scoring takes the power restoration business for overdue payments as an example, and the second-level scoring takes the electricity fee and electricity price area as an example to calculate the final scoring results for each city.

[0115] Table 1 Indicator scores for the power restoration service for overdue payments

[0116]

[0117] According to the data in Table 1, Power Supply Company 3 scored 0.097 in this business, which is a relatively weak performance. Its main problem is that it did not reasonably meet customer demands and had a large number of work orders. In terms of responsibility attribution, the responsibility attribution index value of most power supply companies is 1, so most of the work order responsibilities for this business should not be on the power supply company, but Power Supply Company 3 still has a lot of service responsibilities. Therefore, Power Supply Company 3 needs to focus on the business of restoring power for arrears, so as to improve the service level and reduce the number of work orders; at the same time, it is necessary to analyze the reasons for the work orders of the power supply company and reduce the number of responsible work orders as much as possible. Power Supply Company 8 has an overall good performance in the business of restoring power for arrears, but the work order change index is not good. Compared with other power supply companies, its work order number has abnormal fluctuations, which may have service risks and need to be paid attention to.

[0118] The weights of four businesses in the field of electricity charges and electricity prices, including the restoration of power supply for arrears, are calculated by using the improved entropy weight method, and their scores are calculated by the TOPSIS method. The entropy value and weight calculation results are shown in Table 2, and the indicator score calculation results are shown in Table 3.

[0119] Table 2 Entropy values ​​and weights of various businesses in the field of electricity charges and prices

[0120]

[0121] Table 3 Scores of various business indicators in the field of electricity charges and prices

[0122]

[0123] As shown in Table 2, in the electricity fee and electricity price business, the entropy value of the electricity fee account is the smallest, which is 0.750, indicating that this indicator provides the most information for the service evaluation of the power supply company, so this indicator has the highest weight of 0.457 in the field of electricity fee and electricity price; at the same time, the meter reading business has the highest information entropy of 0.923, indicating that the distribution of each power supply company in this business is relatively even, so it has little impact on the evaluation of the electricity fee and electricity price field, so its weight is also small after calculation, which is 0.142. As shown in Table 3, in the field of electricity fee and electricity price, the power supply companies with better performance are Power Supply Company 8, Power Supply Company 9, and Power Supply Company 2. The worst performing power supply company was Power Supply Company 3, which scored 0.032 for its fee collection business, 0.162 for its meter reading business, 0.097 for its power restoration business for overdue payments, and 0.069 for its electricity bill accounting business, all of which were at the bottom of the province and needed attention. The next worst performing power supply company was Power Supply Company 11, which performed relatively well in meter reading and power restoration business, with a fee collection business score of 0.230 and an electricity bill accounting business score of 0.037, and needs to improve on these two businesses.

[0124] The scores of each power supply company are calculated for a total of 12 power supply service areas to form a secondary indicator matrix for power supply service evaluation. The entropy value and weight of each service area are calculated according to the improved entropy weight method. The results are shown in Table 4.

[0125] Table 4 Entropy value and weight of each service field

[0126]

[0127] It can be seen from Table 4 that among all the electricity consumption fields, the electricity inspection field has the smallest entropy value of 0.757, which means that the score distribution of each power supply company in this field is relatively irregular, that is, it provides the most evaluation information, so it has the highest weight of 0.163; on the other hand, in the field of business environment, most power supply companies have similar scores, and its entropy value is also the smallest among all service fields, which is 0.946. Therefore, its weight in the evaluation of power supply service level is also the lowest. After calculation, the weight of this field is 0.037.

[0128] Finally, a comprehensive evaluation will be conducted on each power supply company to obtain a comprehensive score for power supply services. Figure 3 shown.

[0129] Depend on Figure 3 It can be seen that the power supply company ZS has the highest power supply service level, with a comprehensive evaluation result of 0.794; followed by the power supply company LS, with a comprehensive evaluation result of 0.581.

[0130] The calculated domain scores are used as data input for double-layer clustering. The upper layer clustering can obtain the optimal number of clusters of 3 according to the DBI index calculation, and the lower layer can be divided into 4 categories according to the evaluation level. After clustering, the centers of each evaluation level can be obtained as shown in Table 5.

[0131] Table 5. Level center cluster distribution table

[0132]

[0133] According to the scores of each power supply company in different fields, the membership degree of its corresponding evaluation level is obtained, and the weight obtained by the improved entropy weight method is calculated to finally obtain the power supply service level of the power supply company, as shown in Table 6.

[0134] Table 6 Service level membership of power supply companies

[0135]

[0136] According to calculation, in September, there were 3 power supply companies in the province with excellent evaluation grades, namely Power Supply Company 7, Power Supply Company 8, and Power Supply Company 10; 1 power supply company with good evaluation grade, namely Power Supply Company 2; the rest of the power supply companies were rated as medium, and there was no power supply company with poor evaluation grade. Through the fuzzy comprehensive evaluation method proposed in this paper, the service level of power supply companies can be rated and the service level can be intuitively evaluated.

[0137] Step 9. Based on the result data of steps 4), 5), and 8), use visualization to display the comprehensive analysis results of power supply services from the customer perspective and the analysis results of power supply services from the marketing service dimension. Display them in multiple dimensions, including time dimension, geographical dimension, and appeal dimension, accurately locate the shortcomings of marketing services, and provide reminders and warnings to facilitate the management of potential power supply service problems, and effectively support the quantitative analysis and evaluation of the service levels of companies in various cities.

[0138] For specific function interface, see Figure 4 , Figure 5 As shown. Taking the monthly data of 95598 work orders as input, on the basis of the original statistical analysis, the overall service level of each power supply company is evaluated through this patented method, and the scores of various cities are intuitively displayed through radar charts and other means. In terms of time, the power supply companies are evaluated and analyzed on a monthly basis to provide a reference for the periodic service evaluation of power supply companies; in terms of geographical dimensions, the advantages and disadvantages of various cities are analyzed through horizontal comparisons across the province, taking cities as units, which is conducive to mutual learning and improvement; in terms of category dimensions, the scores of various indicators of specific service businesses are accurately located, and each business is accurately scored so that each power supply company can analyze and trace the source.

[0139] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting from any point of view, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention, and any reference numerals in the claims should not be regarded as limiting the claims involved.

Claims

1. A power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation, characterized in that The following steps are involved: 1) Input the structured data of 95598 work orders within a certain period of time; Including types of demands, work order classification, user satisfaction, and whether the power supply company is responsible; 2) Comprehensively consider the number of user demands, service responsibilities of power supply companies, work order change levels, and power supply service quality, and establish a power supply service level evaluation index system based on structured data; 3) Calculate the index value of each power supply service business under various evaluation indicators, and form an evaluation decision matrix of the power supply service level based on multiple indicators; 4) Based on the improved entropy weight method, an objective weighting model for power supply service level evaluation is established, and the entropy value and weight of each sub-business in the fields of marketing, operation and inspection, and infrastructure are obtained according to the objective weighting model. The sub-business includes business environment and electricity charges; 5) Calculate and rank the scores of different power supply companies based on the improved TOPSIS method; 6) Perform double-layer clustering based on the power supply service score to determine the cluster center of the score distribution; 7) Confirm its fuzzy membership function and calculate the power supply service level membership matrix; In the formula, Indicates the index value r ip For state V j The membership degree, matrix A i The nth row in represents the single factor evaluation result of the nth indicator, which is a fuzzy subset of V; 8) Determine the service level of the power supply company based on the principle of maximum affiliation and timely identify power supply service risk issues; 9) Based on the result data of step 4), step 5), and step 8), use visualization to display the comprehensive analysis results of power supply services from the customer perspective and the power supply service analysis results from the marketing service dimension. Display them in multiple dimensions, including time dimension, geographical dimension, and appeal dimension, accurately locate the shortcomings of marketing services, and provide reminders and warnings to facilitate the management of potential power supply service problems; In step 4), including: 401) Establish an indicator score matrix: In the index score matrix R consisting of N power supply companies to be evaluated and M indicators to be evaluated, the entropy value of the mth evaluation indicator can be defined as follows: Where: And assume that when f nm = 0, f nm lnf nm =0; where r nm is the calculated score of the mth indicator of the nth power supply company; 402) Index weight calculation Based on the determination of weights by the improved entropy weight method, the weights of power supply service evaluation indicators are not affected by individual extreme cases, so that relatively reasonable indicator weights are obtained. The weight calculation expression is as follows: In the formula, ω m is the weight of the mth indicator, ω 1m is the weight required by the traditional entropy weight method, ω 2m is the modified weight, H m is the entropy value of the mth indicator, is the average value of all index entropies that are not 1; β is the coefficient; In step 5), the steps include: 501) Establish a power supply service evaluation matrix; In the power supply service evaluation of N power supply companies and M evaluation indicators, the power supply service evaluation matrix R is expressed as: 502) Establishing a weighted power supply service evaluation matrix G; G=(g nm ) N×M In the formula, g nm =ω m r nm ; 503) Calculate similarity The highest score of each indicator is taken as the positive ideal solution, and the lowest score of each indicator is taken as the negative ideal solution. Thus, the ideal solution vector can be obtained. in m∈{1,2,…,M}; Thus, we can get the distance similarity C between the index scores of each power supply company and the positive ideal solution and the negative ideal solution. n1 Similarity with cosine C n2 , the expression is as follows; In the formula, is the Euclidean distance between the index score of the nth power supply company and the positive ideal solution, is the Euclidean distance between the index score of the nth power supply company and the negative ideal solution, C n1 is the distance similarity of the nth power supply company, C n2 is the cosine similarity between the nth power supply company and the positive ideal solution, G n is the indicator score vector of the nth power supply company; Combining distance similarity and cosine similarity, we can get the relative similarity of the power supply company. The expression is as follows: S n =γ1C n1 +(1-γ1)C n2 In the formula, γ1 is the correction coefficient.

2. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 1 is characterized by: In step 2), in the power supply service level evaluation index system, the index layer adopts multiple underlying indicators for different service businesses of the power supply company, including customer demand indicators, responsibility attribution indicators, work order change indicators, and service processing indicators; at the business layer, the hierarchical division is based on the power supply company's business; The service businesses of power supply companies are divided into fields at the field level; in step 4), an objective weighting model for evaluating power supply service levels is established based on the improved entropy weight method to determine the indicator weights at the indicator layer, business layer, and field layer.

3. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 2 is characterized by: When using customer demand indicators to evaluate the power supply service level, based on the 95598 work orders, they are divided into three categories: complaints, opinions, and service application work orders. The customer demand indicators are defined as: In the formula, S i is the score of the i-th type of work order, Ni is the number of i-th type of work orders per million households of the power supply company, and N i,min is the reference value of the number of work orders of the i-th category, α 1i is the correction coefficient of the i-th type of work order, 0<α 1i ≤1 and 4. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 2 is characterized by: When using the responsibility attribution index to evaluate the power supply service level, the causes of all customer complaints are divided into three categories from the perspective of responsibility attribution: customer, power supply enterprise, and indirect personnel responsibility; and the responsibility of the power supply enterprise will cause a worse user experience than the other two reasons, and it is also the direction that the power supply enterprise can optimize; therefore, the smaller the proportion of responsibility in all work orders is at the power supply enterprise, the better the performance of the power supply company. The responsibility attribution index is defined as: Where N ZR N is the number of responsible work orders in the power supply company. ZR,min is the reference value of the number of responsible work orders, and N is the total number of work orders for this business.

5. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 2 is characterized by: When using the work order change index to evaluate the power supply service level, based on the fact that the work order traffic volume is a nonlinear time series data affected by external subjective factors including power load and weather, the number of work order changes is 95,598, which is an abnormal growth of work orders outside the normal fluctuation. When the number of work order changes is too large, it indicates that the power supply company has abnormal conditions in this business area, resulting in a decline in the power supply service level; based on the average change, the normal range of the work order quantity is determined, and the work order change index is defined as: N warn =N -1 (1+G mean (1+d) Where N warn is the threshold of abnormal quantity, N -1 is the number of work orders last month, G mean is the average work order growth rate, and δ is the confidence coefficient of work order changes.

6. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 5 is characterized by: When using service processing indicators to evaluate the power supply service level, the processing results based on the power supply service application are an important part of the power supply service evaluation and an important basis for a good customer experience; among them, the service satisfaction rate of customer demands and the number of repeated work orders are important indicators reflecting the service level of the power supply company. The service satisfaction rate is the customer's feedback on the work order processing results. A high satisfaction rate indicates that customers are generally satisfied with the processing results; the number of repeated work orders reflects that the same event has not been resolved in a timely manner. The more repeated work orders, the lower the efficiency of work order processing. The service processing indicators are defined as follows: In the formula, C is the service satisfaction rate, N repeat is the number of repeated work orders, N repeat,mean is the reference value of the number of repeated work orders, α 41 is the service satisfaction rate correction coefficient, α 42 is the correction coefficient of repeated work order quantity, satisfying α 41 +α 42 =1.

7. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 6 is characterized by: In step 7), the fuzzy membership function of the index is taken as a ridge function; For different businesses under the service area i, calculate the fuzzy relationship of each business and combine the weight W required by each business i , and obtain the fuzzy evaluation vector B of service field i i as follows: Then, according to the fuzzy evaluation vector of each service area and the weight vector W of each service area, the final fuzzy evaluation result C of the power supply company can be obtained: Where C is the fuzzy evaluation vector of the power supply company, B = [B1, B2, ..., B l ] T , l is the service area to be evaluated, Indicates the degree of membership corresponding to different evaluation levels.

8. The power supply service analysis method based on double-layer clustering and fuzzy comprehensive evaluation according to claim 1 is characterized by: β is 35.35.

Images