Power line repair progress pushing method based on machine learning

Abstract

The invention discloses an electric power first-aid repair progress pushing method based on machine learning, and belongs to the field of electric power, and the electric power first-aid repair progress pushing method based on machine learning comprises the steps: 1, building an electric power first-aid repair progress short message sending logic; 2, establishing a power failure sensitive customer database; step 3, presetting a power line repair short message template; and step 4, intelligently pushing the emergency repair progress short message. According to the method, customers who make calls in recent two years to reflect the power failure problem are collected, key information is extracted from telephone records through ASR and NLP, the information is integrated with information in a PMS 3.0 and marketing 2.0 system to construct a power failure sensitive customer database, an RPA tool is used for being connected with the power failure sensitive database, a power failure area is compared with a customer address, customer data are extracted, and the power failure sensitive customer database is established. Power failure sensitive clients in a power failure range are screened out, and the RPA tool automatically calls a short message sending platform and sends a generated short message to the corresponding client; and the pushing of the power repair progress of the power system is more accurate and efficient.

Description

Technical Field

[0001] The present invention relates to the electric power field, and in particular to a method for pushing electric power emergency repair progress based on machine learning. Background Art

[0002] The power system is an indispensable part of people's lives. Electric energy is closely related to people's lives. The power system needs maintenance in daily life, and the maintenance of the power system often requires power outages for inspection and maintenance. Power outages will affect the lives of surrounding residents and make their lives inconvenient. Even if the surrounding residents are informed of the power outage in advance, manual notices or broadcasting are generally used for communication, but some people may not be aware of the power outage, the communication efficiency is low, the communication level is poor, and the power outage time will be affected by the maintenance situation. The power outage time may be extended or shortened. Residents do not know when the maintenance will be completed and how long the power will be supplied. Therefore, manual communication is used to inform the progress of power emergency repairs. However, this method is manpower-consuming, inefficient, and slow to inform. Residents who have not been notified are often dissatisfied with the power supply company and may even complain to the power supply company. At the same time, customers who have experienced power outages many times are often more sensitive to power outages, and are more likely to be dissatisfied with the power supply company, and even complain to the power supply company. Therefore, the efficiency of manual notification of the progress of power emergency repairs is low, customer satisfaction is low, and the labor cost is high, which affects the image and reputation of the power supply company. Summary of the invention

[0003] In order to solve the deficiencies in the prior art, the present invention provides a method for pushing power repair progress based on machine learning, comprising the following steps:

[0004] Step 1: Establish a database of customers who are sensitive to power outages;

[0005] Step 2: Establish the logic of sending SMS messages about power repair progress;

[0006] Step 3: Preset the power emergency repair SMS template;

[0007] Step 4: Intelligent SMS push of repair progress.

[0008] Preferably, the step 1 comprises:

[0009] Step 1.1: Collect the customers who called to report power outages in the past two years, use the ASR engine to convert the phone recordings into text and store them in the database;

[0010] Step 1.2: Then use NLP to extract key information from the text converted by the ASR engine in the database;

[0011] Step 1.3: Integrate the information from the customer call with the information in the PMS 3.0 and Marketing 2.0 systems;

[0012] Step 1.4: Build a power outage sensitive customer database and store the integrated information in the power outage sensitive customer database.

[0013] Preferably, the step 1.2 first preprocesses the text converted by the ASR engine through NLP, first performs word segmentation and removes stop words, and then uses the named entity recognition model, intent recognition technology and time series analysis in NLP to extract key information including: customer address, account number, customer demand, number of power outages and power outage tolerance, and screens customers who meet the logic of sending emergency repair SMS.

[0014] Preferably, in step 1.3, the customer address and account number extracted from the database are matched with the customer data in the PMS3.0 and Marketing 2.0 systems, and by matching the customer's historical data, the line and substation information involved in the customer is queried, and the customer's historical power outage records are integrated to analyze the number of calls made and the customer's sensitivity.

[0015] Preferably, in step 1.4, a data model is designed for the power outage-sensitive customer database, including the customer's basic information, power outage records, power outage tolerance, and number of phone calls fields, and the integrated customer information is stored in the power outage-sensitive customer database. In subsequent calls, the power outage-sensitive customer database is updated in real time.

[0016] Preferably, the logic of sending the SMS message of power repair progress in step 2 specifically includes:

[0017] For customers in the power outage-sensitive customer database, the cause and scope of the power outage will be informed to them within 5 minutes after the power outage is discovered. During the emergency repair work, information on the progress of the emergency repair will be pushed to customers at intervals of 30 minutes. At the same time, customers will be informed of the time of power restoration within 5 minutes after the emergency repair is completed.

[0018] Preferably, step 3 analyzes key information in the database of power outage-sensitive customers, designs different scripts according to different scenarios based on customers' tolerance for power outages, number of power outages, duration of power outages, and different time periods of power outages, including before, during, and after power restoration, and eventually combines different types of scripts into a variety of power outage SMS templates.

[0019] Preferably, step 4 comprises:

[0020] Step 4.1: Use the RPA tool to connect to the outage-sensitive database, execute SQL queries to extract customer data by comparing the outage area with the customer address, and store the extracted customer data in the RPA variable;

[0021] Step 4.2: Use string matching in RPA to compare the address and account number information with the power outage area, filter out customers within the power outage area, and store the comparison results in a table for easy processing;

[0022] Step 4.3: Use RPA combined with customer data to intelligently send SMS messages on power repair progress.

[0023] Preferably, the step 4.3 comprises:

[0024] Step 4.31: The RPA tool replaces the placeholders in the preset power emergency repair SMS template, replacing the key information of name, address and account number;

[0025] Step 4.32: Use the RPA tool to automatically call the SMS sending platform and send the generated SMS to the corresponding customers;

[0026] Step 4.33: 5 minutes before the SMS is sent, the system automatically detects whether the customer’s power supply has been restored;

[0027] Step 4.34: For customers whose power has not been restored, send power repair SMS messages again as a second notification to calm their emotions.

[0028] Preferably, the step 4.33 obtains the power status via the power restoration detection interface and API connection or obtains the power status of the power outage area through monitoring, and then detects the power restoration situation.

[0029] The beneficial effect of the present invention is that compared with the prior art, the present invention collects customers who have called to report power outage problems in the past two years, and analyzes the telephone recordings through ASR and NLP to extract key information including: customer address, account number, customer demands, number of power outages and power outage tolerance, thereby obtaining power outage-sensitive customer information, and integrating the information with the information in the PMS3.0 and Marketing 2.0 systems to build a power outage-sensitive customer database. Sending power outage SMS notifications for customer information in the power outage-sensitive customer database is more efficient, disseminates faster, and is more targeted to avoid customer complaints, saving labor costs.

[0030] Moreover, the present invention obtains the time of informing customers of the cause and scope of power outage within 5 minutes after discovering the power outage for customers in the power outage sensitive database through multiple experiments, which can better soothe customers' emotions, and during the emergency repair work, push information about the progress of the emergency repair to customers at intervals of 30 minutes, and inform customers of the time of power restoration within 5 minutes after the emergency repair is completed. Both can better soothe power outage sensitive customers, avoid dissatisfaction of power outage sensitive customers with the power supply company, and reduce the complaint rate. At the same time, the progress of power emergency repair is sent via text messages, which has the highest customer acceptance, the fastest information acquisition, and the ability to understand the power outage situation more quickly.

[0031] At the same time, the present invention uses RPA tools and database connections, executes SQL queries to compare the power outage area with the customer address to extract customer data, and stores the extracted customer data in RPA variables; then uses string matching to compare information such as the address with the power outage area, screens out customers within the power outage range, and stores the comparison results in a table for easy processing; then uses RPA in combination with customer data to dynamically replace the placeholders in the preset power emergency repair SMS template, replacing information such as the name, address, and account number; the RPA tool automatically calls the SMS sending platform to send the generated SMS to the corresponding customer; 5 minutes before the SMS is sent, the system automatically detects whether the customer's power supply has been restored. For customers whose power has not been restored, the power emergency repair SMS message is sent again for a second notification to calm the customer's emotions, and the entire process is intelligently operated, saving labor costs, improving SMS push efficiency, and transmitting efficiency to a high degree and wide range, so that the push of the power emergency repair progress of the power system is more accurate and more efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of a method for pushing progress of power emergency repairs based on machine learning in the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0034] The present invention provides a method for pushing the progress of power emergency repair based on machine learning, and the present invention comprises the following steps:

[0035] Step 1: Establish a database of customers who are sensitive to power outages;

[0036] Step 2: Establish the logic of sending SMS messages about power repair progress;

[0037] Step 3: Preset the power emergency repair SMS template;

[0038] Step 4: Intelligent SMS push of repair progress.

[0039] In addition, steps 1 to 3 are not strictly performed in chronological order and can be performed simultaneously.

[0040] The step 1 specifically includes:

[0041] Step 1.1: Collect the customers who called to report power outages in the past two years, and use the ASR (automatic speech recognition technology) engine to convert the phone recordings into text and store them in the database;

[0042] Among them, the ASR engine can use GoogleSpeech-to-Text, iFlytek, Baidu ASR, etc.

[0043] Step 1.2: Then use NLP (natural language processing) to extract key information from the text converted by the ASR engine in the database. First, NLP preprocesses the text, performs word segmentation, removes stop words, etc., to ensure that key information is easy to extract. Then use the named entity recognition model in NLP to identify the customer's address and account number. NLP's intent recognition technology identifies the customer's demands and determines whether the customer is dissatisfied, and then obtains the customer's tolerance for power outages. Then, NLP's time series extracts the number of power outages, and then through NLP, you can obtain information including: customer address, account number, customer demands, number of power outages, and power outage tolerance;

[0044] Step 1.3: Integrate the information in the customer's call with the information in PMS3.0 (State Grid's lean equipment asset management system) and Marketing 2.0 system. First, match the customer address and account number extracted from the database with the customer data in PMS3.0 and Marketing 2.0 systems. By matching the customer's historical data, query the line, substation and other information involved in the customer. At the same time, integrate the customer's historical power outage records and analyze the number of calls and the customer's sensitivity.

[0045] Step 1.4: Build a power outage-sensitive customer database and store the integrated information in the power outage-sensitive customer database; first, design a data model for the power outage-sensitive customer database, which includes fields such as customer basic information, power outage records, power outage tolerance, and number of calls made, and store customer data in the power outage-sensitive customer database. In subsequent calls, update the power outage-sensitive customer database in real time, clean and maintain the database regularly, and ensure the accuracy and timeliness of the data.

[0046] The step 2 is to comprehensively consider the tolerance of different customers to the number of text messages and the demand for emergency repair progress information. In order to avoid customer dissatisfaction, the customers in the power outage sensitive customer database should be informed of the cause and scope of the power outage within 5 minutes after the power outage is discovered, which can better calm customer emotions. During the emergency repair work, information on the progress of the emergency repair should be pushed to customers at intervals of 30 minutes, and the time of power restoration should be informed to customers within 5 minutes after the emergency repair is completed. The sending of these text messages can improve customer recognition of the power supply company and avoid complaints from power outage sensitive customers. The customer recognition rate reaches 85%, which is the text message push logic with the highest customer satisfaction.

[0047] The step 3 analyzes the key information of the power outage-sensitive customer database, and designs different scripts according to different scenarios based on the customer's tolerance for power outages, the number of power outages, the duration of power outages, and different time periods of power outages, including before, during, and after the power is restored. Based on the permutations and combinations of different types of scripts, 3×3×3×4×A(4,4)=2592 power outage SMS templates can be finally combined, which can better appease customers. The combination of various types of SMS can avoid customer aesthetic fatigue.

[0048] The step 4 comprises:

[0049] Step 4.1: Use the RPA (Robotic Process Automation) tool to connect to the outage-sensitive database, execute a SQL (database query and programming language) query to compare the outage area with the customer address to extract customer data, and store the extracted customer data in the RPA variable;

[0050] Step 4.2: Use string matching in RPA to compare information such as addresses with the power outage area, filter out customers within the power outage area, and store the comparison results in a table for easy processing;

[0051] Step 4.3: Use RPA combined with customer data to intelligently send SMS messages about power repair progress.

[0052] The step 4.3 includes:

[0053] Step 4.31: The RPA tool replaces the placeholders in the preset power emergency repair SMS template, replacing the key information of name, address and account number;

[0054] Step 4.32: Use the RPA tool to automatically call the SMS sending platform and send the generated SMS to the corresponding customers;

[0055] Step 4.33: 5 minutes before the SMS is sent, the system automatically detects whether the customer’s power supply has been restored;

[0056] Step 4.34: For customers whose power has not been restored, send power repair SMS messages again as a second notification to calm their emotions.

[0057] The step 4.33 can obtain the power status through the power restoration detection interface and API connection or obtain the power status of the power outage area through monitoring, and then detect the power restoration situation.

[0058] The beneficial effect of the present invention is that compared with the prior art, the present invention collects customers who have called to report power outage problems in the past two years, and analyzes the telephone recordings through ASR and NLP to extract key information including: customer address, account number, customer demands, number of power outages and power outage tolerance, thereby obtaining power outage-sensitive customer information, and integrating the information with the information in the PMS3.0 and Marketing 2.0 systems to build a power outage-sensitive customer database. Sending power outage SMS notifications for customer information in the power outage-sensitive customer database is more efficient, disseminates faster, and is more targeted to avoid customer complaints, saving labor costs.

[0059] Moreover, the present invention obtains the time of informing customers of the cause and scope of power outage within 5 minutes after discovering the power outage for customers in the power outage sensitive database through multiple experiments, which can better soothe customers' emotions, and during the emergency repair work, push information about the progress of the emergency repair to customers at intervals of 30 minutes, and inform customers of the time of power restoration within 5 minutes after the emergency repair is completed. Both can better soothe power outage sensitive customers, avoid dissatisfaction of power outage sensitive customers with the power supply company, and reduce the complaint rate. At the same time, the progress of power emergency repair is sent via text messages, which has the highest customer acceptance, the fastest information acquisition, and the ability to understand the power outage situation more quickly.

[0060] At the same time, the present invention uses RPA tools and database connections, executes SQL queries to compare the power outage area with the customer address to extract customer data, and stores the extracted customer data in RPA variables; then uses string matching to compare information such as the address with the power outage area, screens out customers within the power outage range, and stores the comparison results in a table for easy processing; then uses RPA in combination with customer data to dynamically replace the placeholders in the preset power emergency repair SMS template, replacing information such as the name, address, and account number; the RPA tool automatically calls the SMS sending platform to send the generated SMS to the corresponding customer; 5 minutes before the SMS is sent, the system automatically detects whether the customer's power supply has been restored. For customers whose power has not been restored, the power emergency repair SMS message is sent again for a second notification to calm the customer's emotions, and the entire process is intelligently operated, saving labor costs, improving SMS push efficiency, and transmitting efficiency to a high degree and wide range, so that the push of the power emergency repair progress of the power system is more accurate and more efficient.

[0061] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for pushing power repair progress based on machine learning, comprising the following steps: Step 1: Establish a database of customers who are sensitive to power outages; Step 2: Establish the logic of sending SMS messages about power repair progress; Step 3: Preset the power emergency repair SMS template; Step 4: Intelligent SMS push of repair progress.

2. The method for pushing the progress of power repair based on machine learning according to claim 1, characterized in that: The step 1 comprises: Step 1.1: Collect the customers who called to report power outages in the past two years, use the ASR engine to convert the phone recordings into text and store them in the database; Step 1.2: Then use NLP to extract key information from the text converted by the ASR engine in the database; Step 1.3: Integrate the information from the customer call with the information in the PMS 3.0 and Marketing 2.0 systems; Step 1.4: Build a power outage sensitive customer database and store the integrated information in the power outage sensitive customer database.

3. The method for pushing the progress of power repair based on machine learning according to claim 2 is characterized in that: The step 1.2 first pre-processes the text converted by the ASR engine through NLP, first performs word segmentation and removes stop words, and then uses the named entity recognition model, intent recognition technology and time series analysis in NLP to extract key information including: customer address, account number, customer demand, number of power outages and power outage tolerance, and screens customers who meet the logic of sending emergency repair SMS.

4. The method for pushing the progress of power repair based on machine learning according to claim 2 is characterized in that: In step 1.3, the customer address and account number extracted from the database are matched with the customer data in the PMS3.0 and Marketing 2.0 systems. By matching the customer's historical data, the line and substation information involved in the customer is queried, and the customer's historical power outage records are integrated to analyze the number of calls made and the customer's sensitivity.

5. The method for pushing the progress of power repair based on machine learning according to claim 2 is characterized in that: In step 1.4, a data model is designed for the power outage sensitive customer database, which includes the customer's basic information, power outage records, power outage tolerance, and number of phone calls. The integrated customer information is stored in the power outage sensitive customer database, and in subsequent calls, the power outage sensitive customer database is updated in real time.

6. The method for pushing power repair progress based on machine learning according to claim 1, characterized in that: The logic of sending the power repair progress SMS in step 2 specifically includes: For customers in the power outage-sensitive customer database, the cause and scope of the power outage will be informed to them within 5 minutes after the power outage is discovered. During the emergency repair work, information on the progress of the emergency repair will be pushed to customers at intervals of 30 minutes. At the same time, customers will be informed of the time of power restoration within 5 minutes after the emergency repair is completed.

7. The method for pushing power repair progress based on machine learning according to claim 1, characterized in that: The step 3 analyzes the key information of the power outage-sensitive customer database, and designs different scripts according to different scenarios based on the customer's tolerance for power outages, the number of power outages, the duration of power outages, and different power outage time periods, including before, during, and after the power is restored. Based on the arrangement and combination of different types of scripts, a variety of power outage SMS templates can be finally combined.

8. The method for pushing power repair progress based on machine learning according to claim 1, characterized in that: The step 4 comprises: Step 4.1: Use the RPA tool to connect to the outage-sensitive database, execute SQL queries to extract customer data by comparing the outage area with the customer address, and store the extracted customer data in the RPA variable; Step 4.2: Use string matching in RPA to compare the address and account number information with the power outage area, filter out customers within the power outage area, and store the comparison results in a table for easy processing; Step 4.3: Use RPA combined with customer data to intelligently send SMS messages on power repair progress.

9. The method for pushing the progress of power repair based on machine learning according to claim 8, characterized in that: The step 4.3 comprises: Step 4.31: The RPA tool replaces the placeholders in the preset power emergency repair SMS template, replacing the key information of name, address and account number; Step 4.32: Use the RPA tool to automatically call the SMS sending platform and send the generated SMS to the corresponding customers; Step 4.33: 5 minutes before the SMS is sent, the system automatically detects whether the customer’s power supply has been restored; Step 4.34: For customers whose power has not been restored, send power repair SMS messages again as a second notification to calm their emotions.

10. The method for pushing power repair progress based on machine learning according to claim 9, characterized in that: The step 4.33 obtains the power status via the power restoration detection interface and API connection or obtains the power status of the power outage area through monitoring, and then detects the power restoration situation.

Images