An intelligent customer service scheduling method for a power business hall

Abstract

The present application relates to a kind of power business hall intelligent customer service scheduling method, belong to the technical field of digital data estimation in electricity before analysis classification.This method is first to establish user database, and the division of user, consultation question and consultation period is carried out;Then set 5 evaluation coefficients for evaluating user priority, when there is user consultation, the 5 evaluation coefficients of each user are calculated;Then the 5 evaluation coefficients of each user are standardized, so the first priority, the second priority and the third priority of each user are obtained.Finally, according to the three priorities of each user, the total priority of each user is synthesized, and then the corrected priority of each user is obtained according to the order from big to small, and the customer service is sequentially dispatched to the corresponding user.The method greatly improves the efficiency and accuracy of intelligent customer service response, and provides quantitative support and technical support for user demand rapid response and personalized service strategy formulation.

Description

Technical Field

[0001] This invention relates to a method for dispatching intelligent customer service in power business halls, belonging to the technical field of analysis-then-classification in electricity digital data calculation (G06F18 / 24). Background Technology

[0002] Currently, my country's electricity user groups exhibit distinctly diverse characteristics: residential users pay more attention to the convenience of electricity use and cost transparency, small and medium-sized enterprises focus on business processing efficiency and customized power supply solutions, while large industrial users have extremely high requirements for fault response speed and professional technical support. The service needs of different types of users vary significantly in terms of urgency.

[0003] However, traditional smart customer service in power service halls employs a service resource allocation model based on a single type of user inquiry, resulting in significant service homogenization and a series of resource mismatches: the service needs of high-value users are ignored, response efficiency for complex issues lags behind, while simple inquiries consume a large amount of human resources. According to industry statistics, the first-time resolution rate for user issues in traditional smart customer service in power service halls is less than 70%, and over 30% of users make repeat inquiries due to low service scheduling adaptability. Furthermore, large industrial users experience an average waiting time of up to 20 minutes for fault reporting, a duration far from meeting the stringent continuity requirements of industrial production. Summary of the Invention

[0004] The technical problem this invention aims to solve is: how to improve the efficiency of intelligent customer service in power business halls.

[0005] The technical solution proposed by this invention to solve the above-mentioned technical problems is: a method for intelligent customer service dispatching in power business halls, comprising the following steps:

[0006] Step 1: Establish a user database; the user database contains each user's electricity consumption (A1) in the past 12 months, payment timeliness rate (B) in the past 12 months, number of inquiries (C) in the past 12 months, cumulative payment years (D) number of months of inquiries in the past 12 months, number of inquiries without transfer in the past 12 months, average consultation duration in the past 12 months, number of special service inquiries in the past 12 months (O) number of emergency inquiries in the past 12 months (P) customized service contract status, and number of customized service inquiries in the past 12 months (S).

[0007] Users with more than 7 consultation months in the past 12 months are classified as frequent consultation users, users with 4 to 7 consultation months in the past 12 months are classified as secondary frequent consultation users, and users with less than 4 consultation months in the past 12 months are classified as occasional consultation users.

[0008] Consultations are categorized into three types: general consultation, expedited consultation, and emergency consultation.

[0009] The questions asked are categorized into three types: simple questions, general questions, and complex questions. When a user asks one question, it is considered a simple question. When a user asks more than one question and the questions belong to the same business section of the power business hall, it is considered a general question. When a user asks more than one question and the questions belong to different business sections of the power business hall, it is considered a complex question.

[0010] The peak consultation periods are defined as 9:00–12:00 and 14:00–18:00, the off-peak consultation periods are defined as 7:00–9:00, 12:00–14:00 and 18:00–20:00, and the low consultation period is defined as 22:00–7:00 the next day.

[0011] Step 2: Set five evaluation coefficients for prioritizing users, namely, value coefficient Q1, urgency coefficient Q2, preference coefficient Q3, demand complexity coefficient Q4, and service dependence coefficient Q5;

[0012] Step 2.1: When n users make inquiries, calculate the n value coefficients Q1 of the n users according to the following formula (1).

[0013] (1);

[0014] In formula (1), α1, α2, α3 and α4 are the first value weight, the second value weight, the third value weight and the fourth value weight, respectively, and all of them take values ​​in the range of (0,1) and α1+α2+α3+α4=1; A2 is the average electricity consumption in the past 12 months. For residential users, A2 is the average electricity consumption of the residential user's community in the past 12 months. For enterprise users, A2 is the average electricity consumption of the enterprise user's industry in the past 12 months.

[0015] Step 2.2: Calculate the n urgency coefficients Q2 for n users according to the following formula (2).

[0016] (2);

[0017] In equation (2), β1, β2, β3, and β4 are the first, second, third, and fourth emergency weights, respectively, all ranging from (0,1) and β1+β2+β3+β4=1; E is the consultation type factor, where E is 0.3 for ordinary consultations, 0.8 for expedited consultations, and 1 for urgent consultations; F is the user's waiting time; and J is the consultation fault range factor, where J is the consultation fault problem and the fault range is <5000m. 2 J is set to 0.2, and the user inquired about a fault within a 5000m radius. 2 ~10000m 2J is set to 0.6, and the user inquires about a fault problem with a fault range greater than 10000m. 2 J is set to 1 if the user is not inquiring about a fault, and J is set to 0; t is the consultation time coefficient, t is set to 1 if the user inquires during peak hours, t is set to 0.5 if the user inquires during off-peak hours, and t is set to 0 if the user inquires during low-peak hours.

[0018] Step 2.3: Calculate the n preference coefficients Q3 for the n users according to the following formula (3).

[0019] (3);

[0020] In equation (3), γ1, γ2, γ3 and γ4 are the first preference weight, the second preference weight, the third preference weight and the fourth preference weight, respectively, and their values ​​are all in the range of (0,1) and γ1+γ2+γ3+γ4=1; H is the consultation frequency factor, H is 1 for users who frequently consult, H is 0.7 for users who consult less frequently, and H is 0.3 for users who consult occasionally; I is the consultation duration factor, I is 0.5 for users whose average consultation duration in the past 12 months is ≥8min, and I is 0.8 for users whose average consultation duration in the past 12 months is <8min; G is the consultation fit rate, which is the ratio of the number of consultations without transfer in the past 12 months to the number of consultations in the past 12 months and mapped to [0,1].

[0021] Step 2.4: Calculate the complexity coefficient Q4 of the n requirements for n users using the following formula (4).

[0022] (4);

[0023] In equation (4), δ1, δ2, δ3 and δ4 are the first complexity weight, the second complexity weight, the third complexity weight and the fourth complexity weight, respectively, and their values ​​are all in the range of (0,1) and δ1+δ2+δ3+δ4=1; K is the consultation complexity factor, K is 0.3 for users consulting simple questions, K is 0.7 for users consulting general questions, and K is 1 for users consulting complex questions; L1 is the consultation language length; L2 is the average consultation language length of all users; M is the complexity question factor, M is 1 for users consulting complex questions, and 0 otherwise; N is the user's operation proficiency, which is the ratio of the user's average consultation time in the past 12 months to the average consultation time of all users in the past 12 months;

[0024] Step 2.5: Calculate the service dependency coefficients Q5 for n users using the following formula (5).

[0025] (5);

[0026] In equation (5), ε1, ε2, ε3 and ε4 are the first dependency weight, the second dependency weight, the third dependency weight and the fourth dependency weight, respectively, and their values ​​are all in the range of (0,1) and ε1+ε2+ε3+ε4=1; R is the customized service factor. When the user signs up for customized services, R is 1, and when the user does not sign up for customized services, R is 0.

[0027] Step 3: Obtain the historical maximum and minimum values ​​of all evaluation coefficients for each user from the user records of the power business hall, and standardize the five evaluation coefficients Q1 to Q5 for each user according to the following formula (6).

[0028] (6);

[0029] In equation (6), It is the standardized value of the i-th evaluation coefficient; It is the historical maximum value of the i-th evaluation coefficient; It is the historical minimum value of the i-th evaluation coefficient;

[0030] Step 4: Calculate the n first priority T1, second priority T2 and third priority T3 of the n users according to the following formula (7).

[0031] (7);

[0032] In equation (7), ω1, ω2 and ω3 are the first priority weight, the second priority weight and the third priority weight, respectively, and their values ​​are all in the range of (0,1) and ω1+ω2+ω3=1;

[0033] Step 5: Calculate the total priority U of n users using the following formula (8)

[0034] (8);

[0035] In equation (8), a, b and c are the first total priority weight, the second total priority weight and the third total priority weight, respectively, and their values ​​are all in the range of (0,1) and a +b +c = 1;

[0036] Step 6: Correct the n total priorities U of the n users according to the following formula (9) to obtain the n corrected priorities W of the n users.

[0037] (9);

[0038] In equation (9), V is a correction value. When a residential user consults during the peak consultation period, V is 0.2. When a business user consults during the peak consultation period, V is 0.15. When neither a residential user nor a business user consults during the peak consultation period, V is 0.

[0039] Customer service representatives are assigned to users whose correction priorities W are ranked from highest to lowest according to the order of n correction priorities W.

[0040] Furthermore, after step 6, the following steps are performed:

[0041] Step 7: After each user consultation ends, update the first to fourth value weights, the first to fourth urgency weights, the first to fourth preference weights, the first to fourth complexity weights, and the first to fourth dependency weights according to the formula (10);

[0042] (10);

[0043] In equation (10); It is the weight vector after the m-th update; α 1m-1 ~α 4m-1 It represents the first to fourth value weights after the (m-1)th update; β 1m-1 ~β 4m-1 These are the first to fourth urgent weights after the (m-1)th update; γ 1m-1 ~γ 4m-1 These are the first to fourth preference weights after the (m-1)th update; δ 1m-1 ~δ 4m-1 These are the first to fourth complexity weights after the (m-1)th update; ε 1m-1 ~ε 4m-1 These are the first to fourth dependency weights after the (m-1)th update; q is the correction priority for the m-th user whose consultation has ended; q is the rating given to customer service by the m-th user whose consultation has ended, 0≤q≤10; μ is the momentum factor, which is 0.9. λ is the learning rate, set to 0.05; λ is the update weight, set to 0.01; k is the weight index. It is the momentum vector after the m-th update. =0;

[0044] Update the first priority weight, second priority weight, and third priority weight according to formula (11).

[0045] (11);

[0046] In equation (11), , , These are the first priority weight, second priority weight, and third priority weight after the m-th update; , , , These are the standardized value coefficient, urgency coefficient, demand complexity coefficient, and service dependency coefficient for the user whose consultation has ended at the m-th consultation; η is the update factor, which is 0.2.

[0047] Step 8: After the m-th update in Step 7 is completed, the following occurs: When a user makes a consultation, replace the first to fourth value weights α1 to α4, the first to fourth urgency weights β1 to β4, the first to fourth preference weights γ1 to γ4, the first to fourth complexity weights δ1 to δ4, the first to fourth dependency weights ε1 to ε4, the first priority weight ω1, the second priority weight ω2, and the third priority weight ω3 in steps 2-6 with the first to fourth value weights α1 and α4 after the m-th update. 1m ~α 4m First to fourth emergency weights β 1m ~β 4m First to fourth preference weights γ 1m ~γ 4m The first to fourth complexity weights δ 1m ~δ 4m The first to fourth dependency weights ε 1m ~ε 4m First priority weight ω 1m Second priority weight ω 2m and the third priority weight ω 3m And calculate according to steps 2-6. users A new correction priority, and then based on A new correction priority is assigned in descending order, and customer service representatives are dispatched to users whose correction priorities correspond to the new priority.

[0048] The beneficial effects of this invention are as follows: By precisely segmenting users, inquiry questions, and inquiry times, five evaluation coefficients with unique characteristics are derived for each user; and based on these evaluation coefficients, the priority of each user is quantified, thus prioritizing customer service for users with higher modified priority levels. This invention achieves precise scheduling of power customer service resources through multi-dimensional evaluation coefficients and dynamic priority calculation, prioritizing the service rights of users with complex issues and high demands, effectively reducing waiting times for core users, optimizing resource allocation efficiency, improving user service experience, and significantly enhancing the intelligent customer service efficiency of power business halls. Detailed Implementation

[0049] Example 1

[0050] The intelligent customer service dispatching method for power business halls in this embodiment 1 includes the following steps:

[0051] Step 1: Establish a user database; the user database contains each user's electricity consumption (A1) in the past 12 months, payment timeliness rate (B) in the past 12 months, number of inquiries (C) in the past 12 months, cumulative payment years (D) number of months of inquiries in the past 12 months, number of inquiries without transfer in the past 12 months, average consultation duration in the past 12 months, number of special service inquiries in the past 12 months (O) number of emergency inquiries in the past 12 months (P) customized service contract status, and number of customized service inquiries in the past 12 months (S).

[0052] In addition to providing standardized basic electricity services such as electricity bill payment, account binding, and routine electricity business processing, power service halls also offer special services. Special services refer to users' needs for non-standardized, targeted services, such as special electricity guarantee consultation; customized services are exclusive services tailored to users after signing a contract. Special services and customized services are related, with customized services being the core sub-category of special services.

[0053] Users with more than 7 consultation months in the past 12 months are classified as frequent consultation users, users with 4 to 7 consultation months in the past 12 months are classified as secondary frequent consultation users, and users with less than 4 consultation months in the past 12 months are classified as occasional consultation users.

[0054] Consultations are categorized into three types: general consultation, expedited consultation, and emergency consultation.

[0055] The questions asked are categorized into three types: simple questions, general questions, and complex questions. When a user asks one question, it is considered a simple question. When a user asks more than one question and the questions belong to the same business segment of the power business hall, it is considered a general question. When a user asks more than one question and the questions belong to different business segments of the power business hall, it is considered a complex question.

[0056] The business segments of the power service hall are independent service modules divided according to the core functions, business attributes, and processing scenarios of electricity services. Each segment corresponds to a specific type of focused electricity service need and has its own exclusive business processing procedures, service standards, and professional handling standards. There is no direct functional connection between the segments, and they belong to the independent service scope of each other. Different business segments, that is, multiple questions consulted by users, correspond to two or more unrelated electricity service modules, requiring cross-segment professional knowledge and procedures to answer.

[0057] The peak consultation periods are defined as 9:00–12:00 and 14:00–18:00, the off-peak consultation periods are defined as 7:00–9:00, 12:00–14:00 and 18:00–20:00, and the low consultation period is defined as 22:00–7:00 the next day.

[0058] Step 2: Set five evaluation coefficients for prioritizing users, namely value coefficient Q1, urgency coefficient Q2, preference coefficient Q3, demand complexity coefficient Q4, and service dependence coefficient Q5.

[0059] In this embodiment, the value coefficient Q1 measures the long-term service priority of users, the urgency coefficient Q2 judges the urgency of users' consultation needs, the preference coefficient is used to match users' preferred service modes Q3, the demand complexity coefficient Q4 quantifies the difficulty of users' needs, and the service dependence coefficient Q5 quantifies the degree of users' dependence on customized / special services.

[0060] Step 2.1: When n users make inquiries, calculate the n value coefficients Q1 of the n users according to the following formula (1).

[0061] (1);

[0062] In equation (1), α1, α2, α3 and α4 are the first value weight, the second value weight, the third value weight and the fourth value weight, respectively. They are all determined by experience and can be satisfied as α1+α2+α3+α4=1. In this embodiment, α1=0.2, α2=0.1, α3=0.4 and α4=0.3. A2 is the average electricity consumption in the past 12 months. For residential users, A2 is the average electricity consumption of the residential user's community in the past 12 months. For enterprise users, A2 is the average electricity consumption of the enterprise user's industry in the past 12 months.

[0063] Step 2.2: Calculate the n urgency coefficients Q2 for n users according to the following formula (2).

[0064] (2);

[0065] In equation (2), β1, β2, β3, and β4 are the first, second, third, and fourth emergency weights, respectively, all determined empirically, satisfying β1+β2+β3+β4=1. In this embodiment, β1=0.3, β2=0.2, β3=0.3, and β4=0.2; E is the consultation type factor, with E being 0.3 for ordinary consultations, 0.8 for expedited consultations, and 1 for urgent consultations; F is the user's waiting time; J is the consultation fault range factor, where the user consults about a fault problem and the fault range is <5000m. 2 J is set to 0.2, and the user inquired about a fault within a 5000m radius. 2 ~10000m 2 J is set to 0.6, and the user inquires about a fault problem with a fault range greater than 10000m. 2J is set to 1 if the user is not inquiring about a fault, and J is set to 0; t is the consultation time coefficient, t is set to 1 if the user inquires during peak hours, t is set to 0.5 if the user inquires during off-peak hours, and t is set to 0 if the user inquires during low-peak hours.

[0066] Step 2.3: Calculate the n preference coefficients Q3 for the n users according to the following formula (3).

[0067] (3);

[0068] In equation (3), γ1, γ2, γ3 and γ4 are the first preference weight, the second preference weight, the third preference weight and the fourth preference weight, respectively. They are all determined by experience and satisfy γ1+γ2+γ3+γ4=1. In this embodiment, γ1=0.3, γ2=0.2, γ3=0.4 and γ4=0.1. H is the consultation frequency factor. H is 1 for users who frequently consult, 0.7 for users who consult less frequently, and 0.3 for users who consult occasionally. I is the consultation duration factor. I is 0.5 for users whose average consultation duration in the past 12 months is ≥8min, and 0.8 for users whose average consultation duration in the past 12 months is <8min. G is the consultation fit rate, which is the ratio of the number of consultations without transfer in the past 12 months to the number of consultations in the past 12 months and mapped to [0,1].

[0069] Step 2.4: Calculate the complexity coefficient Q4 of the n requirements for n users using the following formula (4).

[0070] (4);

[0071] In equation (4), δ1, δ2, δ3 and δ4 are the first complexity weight, the second complexity weight, the third complexity weight and the fourth complexity weight, respectively. They are all determined by experience and satisfy δ1+δ2+δ3+δ4=1. In this embodiment, δ1=0.4, δ2=0.2, δ3=0.1 and δ4=0.3. K is the consultation complexity factor. K is 0.3 for users consulting simple questions, 0.7 for users consulting general questions, and 1 for users consulting complex questions. L1 is the consultation language length. L2 is the average consultation language length of all users. M is the complexity question factor. M is 1 for users consulting complex questions, and 0 otherwise. N is the user's operation proficiency, which is the ratio of the user's average consultation time in the past 12 months to the average consultation time of all users in the past 12 months.

[0072] Step 2.5: Calculate the service dependency coefficients Q5 for n users using the following formula (5).

[0073] (5);

[0074] In equation (5), ε1, ε2, ε3 and ε4 are the first dependency weight, the second dependency weight, the third dependency weight and the fourth dependency weight, respectively. They are all determined by experience and satisfy ε1+ε2+ε3+ε4=1. In this embodiment, ε1=0.2, ε2=0.3, ε3=0.2 and ε4=0.3. R is the customization service factor. When the user signs up for the customization service, R is 1. When the user does not sign up for the customization service, R is 0.

[0075] Step 3: Obtain the historical maximum and minimum values ​​of all evaluation coefficients for each user from the user records of the power business hall, and standardize the five evaluation coefficients Q1 to Q5 for each user according to the following formula (6).

[0076] (6);

[0077] In equation (6), It is the standardized value of the i-th evaluation coefficient; It is the historical maximum value of the i-th evaluation coefficient; It is the historical minimum value of the i-th evaluation coefficient.

[0078] In this embodiment, standardization is used to avoid weight imbalance caused by differences in the range of different evaluation coefficients.

[0079] Step 4: Calculate the n first priority T1, second priority T2 and third priority T3 of the n users according to the following formula (7).

[0080] (7);

[0081] In equation (7), ω1, ω2 and ω3 are the first priority weight, the second priority weight and the third priority weight, respectively. They are all determined by experience and satisfy ω1+ω2+ω3=1. In this embodiment, ω1=0.3, ω2=0.4 and ω3=0.3.

[0082] Step 5: Calculate the total priority U of n users using the following formula (8)

[0083] (8);

[0084] In equation (8), a, b and c are the first total priority weight, the second total priority weight and the third total priority weight, respectively. They are all determined by experience and satisfy a + b + c = 1. In this embodiment, a = 0.4, b = 0.2 and c = 0.4.

[0085] Step 6: Correct the n total priorities U of the n users according to the following formula (9) to obtain the n corrected priorities W of the n users.

[0086] (9);

[0087] In equation (9), V is a correction value. When a residential user consults during the peak consultation period, V is 0.2. When a business user consults during the peak consultation period, V is 0.15. When neither a residential user nor a business user consults during the peak consultation period, V is 0.

[0088] Customer service representatives are assigned to users whose correction priorities W are ranked from highest to lowest according to the order of n correction priorities W.

[0089] Example 2

[0090] The intelligent customer service dispatching method for power business halls in Embodiment 2 is basically the same as that in Embodiment 1, except that the following steps are performed after step 6:

[0091] Step 7: After each user consultation ends, update the first to fourth value weights, the first to fourth urgency weights, the first to fourth preference weights, the first to fourth complexity weights, and the first to fourth dependency weights according to the formula (10);

[0092] (10);

[0093] In equation (10); It is the weight vector after the m-th update; α 1m-1 ~α 4m-1 It represents the first to fourth value weights after the (m-1)th update; β 1m-1 ~β 4m-1 These are the first to fourth urgent weights after the (m-1)th update; γ 1m-1 ~γ 4m-1 These are the first to fourth preference weights after the (m-1)th update; δ 1m-1 ~δ 4m-1 These are the first to fourth complexity weights after the (m-1)th update; ε 1m-1 ~ε 4m-1 These are the first to fourth dependency weights after the (m-1)th update; q is the correction priority for the m-th user whose consultation has ended; q is the rating given to customer service by the m-th user whose consultation has ended, 0≤q≤10; μ is the momentum factor, which is 0.9. λ is the learning rate, set to 0.05; λ is the update weight, set to 0.01; k is the weight index. It is the momentum vector after the m-th update. Take 0.

[0094] Update the first priority weight, second priority weight, and third priority weight according to formula (11).

[0095] (11);

[0096] In equation (11), , , These are the first priority weight, second priority weight, and third priority weight after the m-th update; , , , These are the standardized value coefficient, urgency coefficient, demand complexity coefficient, and service dependency coefficient for the user whose consultation has ended at the m-th consultation; η is the update factor, which is 0.2.

[0097] Step 8: After the m-th update in Step 7 is completed, the following occurs: When a user makes a consultation, replace the first to fourth value weights α1 to α4, the first to fourth urgency weights β1 to β4, the first to fourth preference weights γ1 to γ4, the first to fourth complexity weights δ1 to δ4, the first to fourth dependency weights ε1 to ε4, the first priority weight ω1, the second priority weight ω2, and the third priority weight ω3 in steps 2-6 with the first to fourth value weights α1 and α4 after the m-th update. 1m ~α 4m First to fourth emergency weights β 1m ~β 4m First to fourth preference weights γ 1m ~γ 4m The first to fourth complexity weights δ 1m ~δ 4m The first to fourth dependency weights ε 1m ~ε 4m First priority weight ω 1m Second priority weight ω 2m and the third priority weight ω 3m And calculate according to steps 2-6. users A new correction priority, then A new correction priority is assigned in descending order, and customer service representatives are dispatched to users whose correction priorities correspond to the new priority.

[0098] The above description is only a preferred embodiment of the present invention, but the present invention is not limited thereto. All equivalent substitutions or modifications made to the concepts and technical solutions of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for dispatching intelligent customer service in a power business hall, characterized in that... The steps include the following: Step 1: Establish a user database; the user database contains each user's electricity consumption (A1) in the past 12 months, payment timeliness rate (B) in the past 12 months, number of inquiries (C) in the past 12 months, cumulative payment years (D) number of months of inquiries in the past 12 months, number of inquiries without transfer in the past 12 months, average consultation duration in the past 12 months, number of special service inquiries in the past 12 months (O) number of emergency inquiries in the past 12 months (P) customized service contract status, and number of customized service inquiries in the past 12 months (S). Users with more than 7 consultation months in the past 12 months are classified as frequent consultation users, users with 4 to 7 consultation months in the past 12 months are classified as secondary frequent consultation users, and users with less than 4 consultation months in the past 12 months are classified as occasional consultation users. Consultations are categorized into three types: general consultation, expedited consultation, and emergency consultation. The questions asked in the consultation are categorized into three types: simple questions, general questions, and complex questions. When a user asks only one question, it is considered a simple question. When the number of questions a user asks is greater than 1 and belongs to the same business segment of the power business hall, it is recorded as a general inquiry. When the number of questions a user asks is greater than 1 and they belong to different business sectors of the power business hall, they are considered complex questions. The peak consultation periods are defined as 9:00–12:00 and 14:00–18:00, the off-peak consultation periods are defined as 7:00–9:00, 12:00–14:00 and 18:00–20:00, and the low consultation period is defined as 22:00–7:00 the next day. Step 2: Set five evaluation coefficients for prioritizing users, namely, value coefficient Q1, urgency coefficient Q2, preference coefficient Q3, demand complexity coefficient Q4, and service dependence coefficient Q5; Step 2.1: When n users make inquiries, calculate the n value coefficients Q1 of the n users according to the following formula (1). （1）； In formula (1), α1, α2, α3 and α4 are the first value weight, the second value weight, the third value weight and the fourth value weight, respectively, and all of them take values ​​in the range of (0,1) and α1+α2+α3+α4=1; A2 is the average electricity consumption in the past 12 months. For residential users, A2 is the average electricity consumption of the residential user's community in the past 12 months. For enterprise users, A2 is the average electricity consumption of the enterprise user's industry in the past 12 months. Step 2.2: Calculate the n urgency coefficients Q2 for n users according to the following formula (2). （2）； In equation (2), β1, β2, β3 and β4 are the first emergency weight, the second emergency weight, the third emergency weight and the fourth emergency weight, respectively, and their values ​​are all in the range of (0,1) and β1+β2+β3+β4=1; E is the consultation type factor, E is 0.3 for ordinary consultation, E is 0.8 for expedited consultation and E is 1 for emergency consultation; F represents the user's waiting time; J is the fault scope factor, indicating that the user is inquiring about a fault problem and the fault scope is <5000m. 2 J is set to 0.2, and the user inquired about a fault within a 5000m radius. 2 ~10000m 2 J is set to 0.6, and the user inquires about a fault problem with a fault range greater than 10000m. 2 If J is 1, then J is 0; if the user is not inquiring about a fault, then J is 0. t is the consultation time coefficient. When users consult during peak hours, t is 1; when users consult during off-peak hours, t is 0.5; and when users consult during low-peak hours, t is 0. Step 2.3: Calculate the n preference coefficients Q3 for the n users according to the following formula (3). （3）； In equation (3), γ1, γ2, γ3 and γ4 are the first preference weight, the second preference weight, the third preference weight and the fourth preference weight, respectively, and their values ​​are all in the range of (0,1) and γ1+γ2+γ3+γ4=1; H is the consultation frequency factor, H is 1 for users who frequently consult, H is 0.7 for users who consult less frequently, and H is 0.3 for users who consult occasionally; I is the consultation duration factor, I is 0.5 for users whose average consultation duration in the past 12 months is ≥8min, and I is 0.8 for users whose average consultation duration in the past 12 months is <8min; G is the consultation fit rate, which is the ratio of the number of consultations without transfer in the past 12 months to the number of consultations in the past 12 months and mapped to [0,1]. Step 2.4: Calculate the complexity coefficient Q4 of the n requirements for n users using the following formula (4). （4）； In equation (4), δ1, δ2, δ3 and δ4 are the first complexity weight, the second complexity weight, the third complexity weight and the fourth complexity weight, respectively, and their values ​​are all in the range of (0,1) and δ1+δ2+δ3+δ4=1; K is the consultation complexity factor, K is 0.3 for users consulting simple questions, K is 0.7 for users consulting general questions, and K is 1 for users consulting complex questions; L1 is the consultation language length; L2 is the average length of the consultation language for all users; M is the complexity factor; M is 1 if the user asks a complex question, otherwise it is 0. N is the user's operational proficiency, which is the ratio of the user's average consultation time in the past 12 months to the average consultation time of all users in the past 12 months. Step 2.5: Calculate the service dependency coefficients Q5 for n users using the following formula (5). （5）； In equation (5), ε1, ε2, ε3 and ε4 are the first dependency weight, the second dependency weight, the third dependency weight and the fourth dependency weight, respectively, and their values ​​are all in the range of (0,1) and ε1+ε2+ε3+ε4=1; R is the customized service factor. When the user signs up for customized services, R is 1, and when the user does not sign up for customized services, R is 0. Step 3: Obtain the historical maximum and minimum values ​​of all evaluation coefficients for each user from the user records of the power business hall, and standardize the five evaluation coefficients Q1 to Q5 for each user according to the following formula (6). （6）； In equation (6), It is the standardized value of the i-th evaluation coefficient; It is the historical maximum value of the i-th evaluation coefficient; It is the historical minimum value of the i-th evaluation coefficient; Step 4: Calculate the n first priority T1, second priority T2 and third priority T3 of the n users according to the following formula (7). （7）； In equation (7), ω1, ω2 and ω3 are the first priority weight, the second priority weight and the third priority weight, respectively, and their values ​​are all in the range of (0,1) and ω1+ω2+ω3=1; Step 5: Calculate the total priority U of n users using the following formula (8) （8）； In equation (8), a, b and c are the first total priority weight, the second total priority weight and the third total priority weight, respectively, and their values ​​are all in the range of (0,1) and a +b +c = 1; Step 6: Correct the n total priorities U of the n users according to the following formula (9) to obtain the n corrected priorities W of the n users. （9）； In equation (9), V is a correction value. When a residential user consults during the peak consultation period, V is 0.

2. When a business user consults during the peak consultation period, V is 0.

15. When neither a residential user nor a business user consults during the peak consultation period, V is 0. Customer service representatives are assigned to users whose correction priorities W are ranked from highest to lowest according to the n correction priorities W.

2. The intelligent customer service dispatching method for power business halls according to claim 1, characterized in that: After step 6, the following steps are performed: Step 7: After each user consultation ends, update the first to fourth value weights, the first to fourth urgency weights, the first to fourth preference weights, the first to fourth complexity weights, and the first to fourth dependency weights according to the formula (10); （10）； In equation (10); It is the weight vector after the m-th update; α 1m-1 ~α 4m-1 It represents the first to fourth value weights after the (m-1)th update; β 1m-1 ~β 4m-1 These are the first to fourth emergency weights after the (m-1)th update; γ 1m-1 ~γ 4m-1 These are the first to fourth preference weights after the (m-1)th update; δ 1m-1 ~δ 4m-1 These are the first to fourth complexity weights after the (m-1)th update; ε 1m-1 ~ε 4m-1 These are the first to fourth dependency weights after the (m-1)th update; q is the correction priority for the m-th user whose consultation has ended; q is the rating given to customer service by the m-th user whose consultation has ended, 0≤q≤10; μ is the momentum factor, which is set to 0.9; λ is the learning rate, set to 0.05; λ is the update weight, set to 0.01; k is the weight index. It is the momentum vector after the m-th update. =0; Update the first priority weight, second priority weight, and third priority weight according to formula (11). （11）； In equation (11), , , These are the first priority weight, second priority weight, and third priority weight after the m-th update; , , , These are the standardized value coefficient, urgency coefficient, demand complexity coefficient, and service dependence coefficient for the user whose consultation has ended at the m-th consultation. η is the update factor, which is set to 0.2; Step 8: After the m-th update in Step 7 is completed, the following occurs: When a user makes a consultation, replace the first to fourth value weights α1 to α4, the first to fourth urgency weights β1 to β4, the first to fourth preference weights γ1 to γ4, the first to fourth complexity weights δ1 to δ4, the first to fourth dependency weights ε1 to ε4, the first priority weight ω1, the second priority weight ω2, and the third priority weight ω3 in steps 2-6 with the first to fourth value weights α1 and α4 after the m-th update. 1m ~α 4m First to fourth emergency weights β 1m ~β 4m First to fourth preference weights γ 1m ~γ 4m The first to fourth complexity weights δ 1m ~δ 4m The first to fourth dependency weights ε 1m ~ε 4m First priority weight ω 1m Second priority weight ω 2m and the third priority weight ω 3m And calculate according to steps 2-6. users A new correction priority, and then based on A new correction priority is assigned in descending order, and customer service representatives are dispatched to users whose correction priorities correspond to the new priority.

Images