Electricity charge inquiry appeal analysis system and method for customers preferred to inquire electricity charge

Abstract

The invention discloses an electric charge inquiry appeal analysis system and method for customers who preferentially inquire electric charge, and relates to the technical field of power supply customer services. Compared with a previous service system, the system solves the problems that an existing system is limited in intelligent degree, complex appeals of clients cannot be fully understood, personalized reply precision is not high, user experience is affected, part of systems are lack of monitoring and feedback mechanisms for real-time behaviors of the users, demand changes and feedback opinions of the users cannot be captured in time, and user experience is affected. And system optimization is not timely enough. According to the method, customer demands are analyzed on the basis of demands word segmentation, demands standardization, keyword matching and other operations, further statistical analysis is performed on the basis of an analysis result, and corresponding service analysis and suggestions are intelligently generated; customer service demands are mined by analyzing demands of various crowds, and intelligent services are assisted to realize customer service precision; and process automation and real-time monitoring of the system are realized based on the monitoring platform, so that the feedback efficiency of the system is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply customer service, and in particular to an electricity bill inquiry demand analysis system and method for customers who prefer to check their electricity bills. Background Art

[0002] Today, we are in the Internet information age. Affected by the dual factors of economic development and living standards, power customers have higher and higher expectations for power supply services, resulting in a situation of diversified customer demands. From the essence of customer complaints, multiple factors determine that power supply services have the characteristics of imperfection. Although customer complaints are inevitable, as an important part of customer feedback information in high-quality services, how to efficiently, reasonably and scientifically use the demand information has become a problem that all power grid companies must face in the "Internet +" era.

[0003] The existing systems have limited intelligence and are unable to fully understand the complex demands of customers. The accuracy of personalized responses is also low, which affects the user experience. In addition, some systems lack monitoring and feedback mechanisms for real-time user behavior and are unable to promptly capture changes in user needs and feedback, resulting in untimely system optimization.

[0004] In order to solve the above problems, the present invention proposes an electricity bill inquiry demand analysis system and method for customers who prefer to check their electricity bills. Summary of the invention

[0005] The purpose of the present invention is to provide a system and method for analyzing electricity bill inquiry demands for customers who prefer to check electricity bills to solve the problems raised in the background technology:

[0006] The existing systems have limited intelligence and are unable to fully understand the complex demands of customers. The accuracy of personalized responses is also low, which affects the user experience. In addition, some systems lack monitoring and feedback mechanisms for real-time user behavior and are unable to promptly capture changes in user needs and feedback, resulting in untimely system optimization.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] An electricity bill inquiry demand analysis system for customers who prefer to check their electricity bills, comprising:

[0009] Customer demand collection module: used to collect electricity bill inquiry demand data of customers who prefer to check electricity bills and perform data source classification, data cleaning and data fusion processing;

[0010] Appeal word segmentation processing module: used to obtain the processed text data, segment the text data, and obtain the word frequency of the segmented words;

[0011] Keyword classification module: used to select keywords related to electricity charges from the word frequency, classify the keywords, and obtain the keyword set corresponding to the demand;

[0012] Keyword matching module: used to build a rule model, use the obtained demand keyword set to match, obtain the corresponding demand, and train and optimize the model based on the training set;

[0013] Result statistical analysis module: used to perform statistical analysis on the model output results;

[0014] Result visualization module: used to visualize the statistical results using visualization tools;

[0015] Automatic monitoring module: used to connect the modules of the development program in series using ETL tools, upload them to the monitoring platform for automatic execution and monitoring;

[0016] The customer demand collection module includes a customer demand collection unit and a customer demand preprocessing module, which are respectively used for collecting and preprocessing customer demands.

[0017] A method for analyzing the electricity bill inquiry demands of customers who prefer to check their electricity bills comprises the following steps:

[0018] S1: Collect electricity bill inquiry demand data from customers who prefer to check electricity bills and perform data source classification, data cleaning and data fusion processing;

[0019] S2: Obtain the processed text data, segment the text data, and obtain the frequency of the segmented words;

[0020] S3: Select keywords related to electricity charges from the word frequency, classify the keywords, and obtain a keyword set corresponding to the demand;

[0021] S4: Build a rule model, use the obtained demand keyword set to match, obtain the corresponding demand, and train and optimize the model based on the training set;

[0022] S5: Statistical analysis of model output results;

[0023] S6: Use visualization tools to visualize the statistical results.

[0024] Preferably, the electricity bill inquiry request includes electricity bill, electricity consumption, arrears, balance, meter reading, electricity price, steps and payment.

[0025] Preferably, in S4, feature extraction is performed based on the extracted demand keyword set to be analyzed to obtain word vectors, and similarity calculation is performed based on the optimized similarity model and the word vectors of the keyword set stored in the database:

[0026]

[0027] Among them, S(A,B) represents the calculated similarity; β is the correction parameter; A and B represent the keyword set to be analyzed and the keyword set stored in the database respectively; a i represents the i-th demand keyword vector in the demand keyword set to be analyzed; b j represents the jth demand keyword vector of the demand keyword set stored in the database; n is the number of word vectors in the keyword set to be analyzed; m is the number of word vectors of the keyword set stored in the database;

[0028] The correction parameters are obtained based on the optimization of the improved ant colony algorithm, and the improved ant colony algorithm is specifically as follows:

[0029] Set the original pheromone concentration, each ant collects the pheromone concentration on all paths, and judges according to the probability formula. The probability p of the vth ant choosing the uth path in the tth iteration is uv (t) is calculated as follows:

[0030]

[0031] Among them, A v is the path chosen by the vth ant; ρ u is the pheromone concentration on the u-th path; u is the expected degree of the path; α is the heuristic operator; γ is the expected operator;

[0032]

[0033] Among them, ρ u (t+1) and ρ u (t) are the paths before and after the tth iteration, ε0∈[0,1] is the volatilization rate of pheromone concentration; Δρ u is the increment of pheromone on the uth path;

[0034]

[0035] Where I is the information common function; K(x) is the inverse of the error function;

[0036] The iterative formulas of the heuristic and expectation operators are:

[0037] α(t)=σ1α(t-1)

[0038] γ(t)=σ2γ(t-1)

[0039] Among them, σ1 and σ2 are both iteration factors, satisfying σ1>1 and σ2>1;

[0040] If the maximum number of iterations is reached, the next step is entered; if the maximum number of iterations is not reached, the probability calculation formula is returned;

[0041] Based on the genetic algorithm, the operators after crossover and mutation are filled in the ant colony algorithm, and the crossover operator is used as the key operator, and the mutation operator is used as the auxiliary operator. At the same time, the ants are given particle properties, among which the particle speed determines the flying distance and direction of the particles, so that the solution space is continuously searched with the optimal particles, as follows:

[0042] Each time the ants iterate, they track and update themselves in a unified way, taking crossover and mutation measures according to the genetic algorithm:

[0043] v v,t+1 =c1v v,t +c2(p b -x v,t )+c2(g b -x v,t )

[0044] Among them, v v,k+1 and v v,k They represent the speed of the vth ant before and after the tth iteration; c1, c2 and c3 represent different paths; p b is the individual extreme value; g b is the population extreme value; x v,t is the current value of the ant at the tth iteration;

[0045] Each time the ant iterates, it tracks the individual extreme value p b and population extreme value g b It uniformly updates itself to find the optimal position and calculates the final shortest path based on the optimal position.

[0046] Preferably, in S5, the customer's demand attention, main demand and demand correlation are analyzed according to the model output result;

[0047] The appeal attention is used to evaluate the degree of customer attention to the appeal, as follows:

[0048]

[0049] The main demand analysis represents the first demand of the customer consultation, and is used to evaluate the demand that the customer is most concerned about;

[0050] The demand association is based on the association rule Apriori algorithm to perform association analysis on multiple demand samples to analyze the attention of multiple demands and the correlation between demands, as follows:

[0051]

[0052]

[0053] Then generate intelligent service analysis and suggestions based on the analysis results.

[0054] Compared with the prior art, the present invention provides a system and method for analyzing electricity bill inquiry demands for customers who prefer to check electricity bills, which has the following beneficial effects:

[0055] The present invention analyzes corresponding demands based on operations such as demand segmentation, demand standardization and keyword matching, and performs further statistical analysis based on the analysis results to intelligently generate corresponding service analysis and suggestions; the present invention mines customer service needs by analyzing the demands of various groups of people, and assists intelligent services to achieve precise customer service; it also realizes system process automation and real-time monitoring based on the monitoring platform, greatly improving the feedback efficiency of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a system block diagram mentioned in Embodiment 1 of the present invention;

[0057] Figure 2 This is a flow chart of the method mentioned in Example 2 of the present invention;

[0058] Figure 3 The visualization report result diagram mentioned in Example 2 of the present invention;

[0059] Figure 4 The appeal attention graph mentioned in Embodiment 3 of the present invention;

[0060] Figure 5 The main demand distribution diagram mentioned in Example 3 of the present invention;

[0061] Figure 6 This is a distribution diagram of the number of customer demands mentioned in Example 3 of the present invention;

[0062] Figure 7 The appeal attention graph mentioned in Embodiment 4 of the present invention;

[0063] Figure 8 This is the main demand distribution diagram mentioned in Example 4 of the present invention.

[0064] Meaning of the symbols in the figure:

[0065] 100. Customer demand collection module; 110. Customer demand collection unit; 120. Customer demand preprocessing module; 200. Demand segmentation processing module; 300. Keyword classification module; 400. Keyword matching module; 500. Result statistical analysis module; 600. Result visualization module; 700. Automated monitoring module. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0067] The present invention analyzes the corresponding demands based on demand segmentation, demand standardization and keyword matching, and performs further statistical analysis based on the analysis results to intelligently generate corresponding service analysis and suggestions; the present invention mines customer service needs by analyzing the demands of various groups of people, and assists intelligent services to achieve precise customer service; the present invention also realizes system process automation and real-time monitoring based on the monitoring platform, greatly improving the feedback efficiency of the system. Specifically, it includes the following contents.

[0068] Embodiment 1:

[0069] See also Figure 1 The present invention provides an electricity bill inquiry demand analysis system for customers who prefer to check electricity bills, comprising:

[0070] Customer demand collection module 100: used to collect electricity bill inquiry demand data of customers who prefer to check electricity bills and perform data source classification, data cleaning and data fusion processing;

[0071] The appeal word segmentation processing module 200 is used to obtain the processed text data, segment the text data, and obtain the word frequency of the segmentation;

[0072] Keyword classification module 300: used to select keywords related to electricity charges from the word frequency, and classify the keywords to obtain a keyword set corresponding to the demand;

[0073] Keyword matching module 400: used to build a rule model, use the obtained demand keyword set to match, obtain the corresponding demand, and train and optimize the model based on the training set;

[0074] Result statistical analysis module 500: used to perform statistical analysis on the model output results;

[0075] Result visualization module 600: used to visualize the statistical results using visualization tools;

[0076] Automatic monitoring module 700: used to connect the development program in series using ETL tools, upload it to the monitoring platform, and perform automatic execution and monitoring;

[0077] The customer demand collection module 100 includes a customer demand collection unit 110 and a customer demand preprocessing module 120, which are respectively used for collecting and preprocessing customer demands.

[0078] Embodiment 2:

[0079] See also Figure 2-3 The present invention provides an electricity bill inquiry demand analysis method for customers who prefer to check electricity bills, comprising the following steps:

[0080] S1: Collect the electricity bill inquiry demand data of customers who prefer to check electricity bills and classify the data sources, clean the data and perform data fusion processing; the details are as follows:

[0081] Collect electricity bill inquiry demand data from customers who prefer to check their electricity bills. The electricity bill inquiry demand data includes electricity bills, electricity consumption, arrears, balance, meter reading, electricity price, tiered and payment, etc. Perform data source classification, data cleaning and data fusion processing on the collected electricity bill inquiry demand data, and classify and pre-process them according to different customers and different needs.

[0082] S2: Get the processed text data, segment the text data, and obtain the frequency of the segmented words; the details are as follows:

[0083] Get a certain amount of text data, use the jieba Chinese word segmentation library to segment the text data, and get the word frequency. The details are as follows:

[0084] First, make sure that the jieba library has been installed. Then load the obtained text data into Python. Assuming that you already have a text file, you can use Python's built-in function to read the text file. Then, use the jieba library to segment the text data. Finally, count the word frequency after segmentation. This can be achieved using the Counter class in Python's collections library. Through the above steps, you can get the processed text data, and use the jieba library to segment the words, and finally get the word frequency information after segmentation.

[0085] S3: Select keywords related to electricity charges from the word frequency, classify the keywords, and obtain a keyword set corresponding to the demand; the details are as follows:

[0086] First, select keywords related to "electricity bills" and so on from the word frequency. You can determine the relevant keywords by looking at the words containing "electricity bills" and their contexts. Next, classify the selected keywords to identify the keyword set corresponding to the appeal. According to your specific needs and the content of the text data, you can classify the keywords into different categories, such as "query", "payment", "adjustment", etc. Finally, based on the classified keyword set, you can further process the relevant text data and take corresponding processing measures for different appeals. Through the above steps, you can get the keyword set related to "electricity bills" and classify the keywords, so that you can better understand the electricity bill-related appeals involved in the text data.

[0087] S4: Build a rule model, use the obtained demand keyword set to match, obtain the corresponding demand, and train and optimize the model based on the training set; the details are as follows:

[0088] Based on the extracted demand keyword set to be analyzed, feature extraction is performed to obtain word vectors, and similarity calculation can be performed based on the optimized similarity model built based on the gbase development program and the word vectors of the keyword set stored in the database:

[0089]

[0090] Among them, S(A,B) represents the calculated similarity; β is the correction parameter; A and B represent the keyword set to be analyzed and the keyword set stored in the database respectively; a i represents the i-th demand keyword vector in the demand keyword set to be analyzed; b j represents the jth demand keyword vector of the demand keyword set stored in the database; n is the number of word vectors in the keyword set to be analyzed; m is the number of word vectors of the keyword set stored in the database;

[0091] The correction parameters are obtained based on the improved ant colony algorithm. The improved ant colony algorithm is as follows:

[0092] Set the original pheromone concentration, each ant collects the pheromone concentration on all paths, and judges according to the probability formula. The probability p of the vth ant choosing the uth path in the tth iteration is uv (t) is calculated as follows:

[0093]

[0094] Among them, A v is the path chosen by the vth ant; ρ u is the pheromone concentration on the u-th path; u is the expected degree of the path; α is the heuristic operator; γ is the expected operator;

[0095]

[0096] Among them, ρ u (t+1) and ρ u (t) are the paths before and after the tth iteration, ε0∈[0,1] is the volatilization rate of pheromone concentration; Δρ u is the increment of pheromone on the uth path;

[0097]

[0098] Where I is the information common function; K(x) is the inverse of the error function;

[0099] The iterative formulas of the heuristic and expectation operators are:

[0100] α(t)=σ1α(t-1)

[0101] γ(t)=σ2γ(t-1)

[0102] Among them, σ1 and σ2 are both iteration factors, satisfying σ1>1 and σ2>1;

[0103] If the maximum number of iterations is reached, the next step is entered; if the maximum number of iterations is not reached, the probability calculation formula is returned;

[0104] Based on the genetic algorithm, the operators after crossover and mutation are filled in the ant colony algorithm, and the crossover operator is used as the key operator, and the mutation operator is used as the auxiliary operator. At the same time, the ants are given particle properties, among which the particle speed determines the flying distance and direction of the particles, so that the solution space is continuously searched with the optimal particles, as follows:

[0105] Each time the ants iterate, they track and update themselves in a unified way, taking crossover and mutation measures according to the genetic algorithm:

[0106] v v,t+1 =c1v v,t +c2(p b -x v,t )+c2(g b -x v,t )

[0107] Among them, v v,k+1 and v v,k They represent the speed of the vth ant before and after the tth iteration; c1, c2 and c3 represent different paths; p b is the individual extreme value; g b is the population extreme value; x v,t is the current value of the ant at the tth iteration;

[0108] Each time the ant iterates, it tracks the individual extreme value p b and population extreme value g b It uniformly updates itself to find the optimal position and calculates the final shortest path based on the optimal position.

[0109] Traditional ant colony algorithms are prone to local optimal situations. Improvements based on genetic algorithms can enrich the ant population when the ant colony algorithm reaches a local optimal point and stagnates, thereby preventing the algorithm from maturing too early. This can greatly enhance the genetic algorithm's search capabilities and improve the matching accuracy of the similarity model.

[0110] S5: Perform statistical analysis on the model output results; the details are as follows:

[0111] Comprehensively analyze the model output results based on the customer's demand attention, main demand and demand correlation:

[0112] In order to evaluate the degree of customer attention to demands, the demand attention index is defined as follows:

[0113]

[0114] Among them, the larger the appeal attention index value is, the higher the attention is;

[0115] The main demand refers to the first demand of the customer consultation, which is used to assess the customer's priority demand.

[0116] First, the number of customer consultation requests is analyzed, and then the association rule Apriori algorithm is used to perform association analysis on multiple request samples to analyze the attention of multiple requests and the correlation between the requests. The formula is as follows:

[0117]

[0118]

[0119] Finally, intelligent service analysis and recommendations are generated based on the analysis results.

[0120] S6: Use visualization tools to visualize the statistical results. The details are as follows:

[0121] Based on visualization tools, such as tables, bar charts, etc., you can refer to Figure 3 The visualization results can realize the intuitive display of electricity bill inquiry demand data for customers who prefer to check their electricity bills.

[0122] Embodiment 3:

[0123] From the 12,000 customers who prefer to check their electricity bills in a certain area, 400 manual service work orders for residential electricity bill inquiries in October 2023 were randomly selected, and customer demands were analyzed through recordings. The customer demands were analyzed from three aspects: customer demand attention, main demand, and demand relevance. The demand classification and meaning are detailed in Table 1:

[0124] Table 1: Classification and meaning of electricity bill inquiry service requirements

[0125]

[0126]

[0127] Analysis of appeal attention:

[0128] Statistics on the attention paid to various demands and drawing bar charts can be used for reference Figure 4After analysis, firstly, the residential customers who prefer to check electricity bills in a certain area are mainly concerned about meter readings, arrears, electricity bills and other information, among which meter readings are concerned about 66%, arrears are concerned about 41%, and electricity bills are concerned about 28%. Secondly, customers pay less attention to balances and payments.

[0129] Main appeal analysis:

[0130] Statistical sample main demands distribution, and draw a bar chart for reference Figure 5 After analysis, it was found that residential customers of a certain area who preferred to check their electricity bills gave priority to consulting three types of demands: meter reading numbers, electricity bills, and arrears. Among them, the samples for consulting meter reading numbers accounted for 42%, electricity bills accounted for 24%, and arrears accounted for 22%. The samples of these three types of demands accounted for 88%, while demands such as balances and payments were rarely consulted in priority.

[0131] Demand correlation analysis:

[0132] First, analyze the number of customer consultation requests, count the number of sample customer requests and draw a pie chart for reference Figure 6 ,After analysis, it was found that the customers of a company in a certain region preferred to check their electricity ,bills with more demands, that is, they made multiple demands at one time, among which 65% of the customers made 2 or more demands, and 34% of the customers made 3 or more demands.

[0133] The analysis results show that the demands with strong correlation are mainly concentrated in the four categories of demands: meter reading, arrears, electricity price, and tiered demands. For example, the correlation between "meter reading" and "arrears" and "electricity price" is 49% and 31% respectively, as shown in Table 2:

[0134] Table 2 Correlation of electricity bill inquiry requests from residential customers of a company in a certain region

[0135]

[0136] It can be seen from the table that when customers check the display data, they tend to inquire about the information such as arrears and electricity prices at the same time, while the correlation between electricity charges, electricity consumption, balance and other demands is relatively low.

[0137] Intelligent service analysis and suggestions:

[0138] Analyze the current intelligent service strategy based on the results of customer demand analysis. See Table 3 for details:

[0139] Table 3 Intelligent service strategy for electricity bill query scenario

[0140]

[0141]

[0142] The table shows that there are 7 problems in the intelligent service in 5 types of demands such as meter reading and arrears checking, including "redundant broadcast content" and "broadcast content does not meet customer needs". 7 optimization suggestions have been put forward based on the analysis results, including "simplifying broadcast content" and "adjusting broadcast strategy", as shown in Table 4.

[0143] Table 4: Optimization suggestions and business feedback on intelligent services for residential customers of a company in a certain region

[0144]

[0145]

[0146]

[0147] Embodiment 4:

[0148] 360 manual service work orders were randomly selected from the electric vehicle electricity bill inquiry work orders of a certain regional company in August, September and October 2023, and customer demand analysis was conducted through recording. The customer demands were analyzed from three aspects: customer demand attention, main demand, and demand correlation.

[0149] Analysis of appeal attention:

[0150] Count the attention paid to various demands and draw a bar chart such as Figure 7 After analysis, it was found that customers were mainly concerned about electricity bills and peak and valley electricity prices, with a concern rate of 27% each, while customers paid less attention to arrears, balances, meter readings, etc.

[0151] Main appeal analysis

[0152] Statistical sample main demands distribution, and draw a bar chart such as Figure 8 After analysis, customers prioritized consulting on three types of demands, namely electricity bills, peak and valley electricity prices, and electricity consumption, accounting for 80%, while balances and arrears were less prioritized.

[0153] Demand correlation analysis:

[0154] The correlation analysis of multiple appeal samples shows that the three types of appeals are highly correlated, namely, peak-valley time period, peak-valley electricity price, and whether there is a peak-valley electricity meter. The correlation between peak-valley time period and peak-valley electricity price is 83%. The correlation between electricity fee, electricity consumption, and bill is highly correlated, and the correlation between electricity fee and electricity consumption is 60%. See Table 5 for details:

[0155] Table 5 Correlation of electricity bill inquiry requests by electric vehicle customers of a company in a certain region

[0156]

[0157]

[0158] Intelligent service analysis and suggestions:

[0159] Based on the analysis of customer demand results, we analyzed the current intelligent service situation and found that the intelligent service has the following two problems:

[0160] Problem 1: The content of the report may not meet customer needs. When customers inquire about their electricity bills, the intelligent system will first report the electricity bill, including five items of information: electricity fee, electricity consumption, cycle, arrears, and balance. However, analysis found that electric vehicle customers of a company in a certain region mainly care about two types of information: peak and valley electricity prices and electricity fees.

[0161] Optimization suggestions: It is recommended to differentiate customer types and add "peak and valley electricity prices" reporting content for electric vehicle customers in the electricity price checking scenario.

[0162] Question 2: Smart customer service does not yet provide electric vehicle service scenarios. Analysis shows that after customers say "electric vehicle" or enter the electric vehicle Internet of Vehicles service through the traditional button 7, they are all directly transferred to manual service.

[0163] Optimization suggestion: It is recommended to add new intelligent service scenarios for electric vehicles.

[0164] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An electricity bill inquiry demand analysis system for customers who prefer to check electricity bills, characterized in that: include: Customer demand collection module (100): used to collect electricity bill inquiry demand data of customers who prefer to check electricity bills and perform data source classification, data cleaning and data fusion processing; The appeal word segmentation processing module (200) is used to obtain the processed text data, segment the text data, and obtain the word frequency of the segmentation; Keyword classification module (300): used to select keywords related to electricity charges from the word frequency, and classify the keywords to obtain a keyword set corresponding to the demand; Keyword matching module (400): used to construct a rule model, use the obtained demand keyword set to match, obtain the corresponding demand, and train and optimize the model based on the training set; Result statistical analysis module (500): used to perform statistical analysis on the model output results; Result visualization module (600): used to visualize the statistical results using visualization tools; Automatic monitoring module (700): used to connect the modules of the development program in series using ETL tools, upload them to the monitoring platform, and perform automatic execution and monitoring; The customer demand collection module (100) comprises a customer demand collection unit (110) and a customer demand preprocessing module (120), which are respectively used for collecting and preprocessing customer demands.

2. A method for analyzing electricity bill inquiry demands for customers who prefer to check electricity bills as claimed in claim 1, characterized in that: The steps include: S1: Collect electricity bill inquiry demand data from customers who prefer to check electricity bills and perform data source classification, data cleaning and data fusion processing; S2: Obtain the processed text data, segment the text data, and obtain the frequency of the segmented words; S3: Select keywords related to electricity charges from the word frequency, classify the keywords, and obtain a keyword set corresponding to the demand; S4: Build a rule model, use the obtained demand keyword set to match, obtain the corresponding demand, and train and optimize the model based on the training set; S5: Statistical analysis of model output results; S6: Use visualization tools to visualize the statistical results.

3. The method for analyzing the electricity bill inquiry demands of customers who prefer to check their electricity bills according to claim 2, characterized in that: The electricity bill inquiry request includes electricity bill, electricity consumption, arrears, balance, meter reading, electricity price, steps and payment.

4. The method for analyzing the electricity bill inquiry demands of customers who prefer to check their electricity bills according to claim 2, characterized in that: In S4, feature extraction is performed based on the extracted demand keyword set to be analyzed to obtain word vectors, and similarity calculation is performed based on the optimized similarity model and the word vectors of the keyword set stored in the database: Among them, S(A,B) represents the calculated similarity; β is the correction parameter; A and B represent the keyword set to be analyzed and the keyword set stored in the database respectively; a i represents the i-th demand keyword vector in the demand keyword set to be analyzed; b j represents the jth demand keyword vector of the demand keyword set stored in the database; n is the number of word vectors in the keyword set to be analyzed; m is the number of word vectors of the keyword set stored in the database; The correction parameters are obtained based on the optimization of the improved ant colony algorithm, and the improved ant colony algorithm is specifically as follows: Set the original pheromone concentration, each ant collects the pheromone concentration on all paths, and judges according to the probability formula. The probability p of the vth ant choosing the uth path in the tth iteration is uv (t) is calculated as follows: Among them, A v is the path chosen by the vth ant; ρ u is the pheromone concentration on the u-th path; u is the expected degree of the path; α is the heuristic operator; γ is the expected operator; Among them, ρ u (t+1) and ρ u (t) are the paths before and after the tth iteration, ε0∈[0,1] is the volatilization rate of pheromone concentration; Δρ u is the increment of pheromone on the uth path; Where I is the information common function; K(x) is the inverse of the error function; The iterative formulas of the heuristic and expectation operators are: α(t)=σ1α(t-1) γ(t)=σ2γ(t-1) Among them, σ1 and σ2 are both iteration factors, satisfying σ1>1 and σ2>1; If the maximum number of iterations is reached, the next step is entered; if the maximum number of iterations is not reached, the probability calculation formula is returned; Based on the genetic algorithm, the operators after crossover and mutation are filled in the ant colony algorithm, and the crossover operator is used as the key operator, and the mutation operator is used as the auxiliary operator. At the same time, the ants are given particle properties, among which the particle speed determines the flying distance and direction of the particles, so that the solution space is continuously searched with the optimal particles, as follows: Each time the ants iterate, they track and update themselves in a unified way, taking crossover and mutation measures according to the genetic algorithm: v v,t+1 =c1v v,t +c2(p b -x v,t )+c2(g b -x v,t ) Among them, v v,k+1 and v v,k They represent the speed of the vth ant before and after the tth iteration; c1, c2 and c3 represent different paths; p b is the individual extreme value; g b is the population extreme value; x v,t is the current value of the ant at the tth iteration; Each time the ant iterates, it tracks the individual extreme value p b and population extreme value g b It uniformly updates itself to find the optimal position and calculates the final shortest path based on the optimal position.

5. The method for analyzing the electricity bill inquiry demands of customers who prefer to check their electricity bills according to claim 2, characterized in that: In said S5, the customer's demand attention, main demand and demand correlation are analyzed according to the model output result; The appeal attention is used to evaluate the degree of customer attention to the appeal, as follows: The main demand analysis represents the first demand of the customer consultation, and is used to evaluate the demand that the customer is most concerned about; The demand association is based on the association rule Apriori algorithm to perform association analysis on multiple demand samples to analyze the attention of multiple demands and the correlation between demands, as follows: Then generate intelligent service analysis and suggestions based on the analysis results.

Images