NLP-based intelligent customer service system and method

Abstract

The invention provides an intelligent customer service system and method based on NLP, and belongs to the technical field of data processing. The intelligent customer service system comprises a front-end interaction layer configured with a display module and used for receiving a text input by a user and displaying a reply result; the intention recognition layer is used for determining a user intention according to the input text received by the front-end interaction layer; the RPA execution layer is used for calling a corresponding preset automatic process according to the determined user intention, generating processing result data and sending the generated result data to the front-end interaction layer; the knowledge graph layer is used for constructing a knowledge graph according to historical operation and maintenance data of the photovoltaic power station; and the monitoring and optimizing layer is used for monitoring the running state of the intelligent customer service system. According to the intelligent customer service system provided by the invention, the processing efficiency can be remarkably improved, and user problems can be quickly responded and solved; the processing accuracy is greatly improved, the operation cost is effectively reduced, and the labor dependence and the potential error cost are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an intelligent customer service system and method based on NLP. Background Art

[0002] With the rapid growth of the global demand for clean energy, as an important renewable energy facility, the scale and quantity of photovoltaic power stations have shown an explosive expansion. Under this background, the operation and management of photovoltaic power stations are facing unprecedented challenges. As a key link in the communication between the power station and users, operation and maintenance personnel, etc., customer service work plays a crucial role in ensuring the efficient operation of the power station, promptly solving problems, and improving user satisfaction.

[0003] The traditional customer service mode of photovoltaic power stations mainly relies on manual processing. Staff need to handle a vast amount of complex consultations. On the one hand, the efficiency of manual processing is extremely low. Facing a large number of user consultations, customer service personnel often have difficulty in making accurate responses in a short time, especially during peak hours, and the response delay problem is more prominent. This not only reduces the user experience but also may cause some potential problems not to be solved in time, thus affecting the normal operation of the power station. On the other hand, the accuracy of manual processing is difficult to guarantee. The business of photovoltaic power stations involves complex technical knowledge and operation processes. When dealing with problems, customer service personnel are prone to make mistakes due to differences in personal knowledge reserves and understanding abilities. Summary of the Invention

[0004] The present invention provides an intelligent customer service system and method based on NLP to solve the problems of low efficiency and poor processing accuracy existing in the prior art.

[0005] To achieve the above object, on the one hand, an embodiment of the present invention provides an intelligent customer service system based on NLP. The intelligent customer service system includes: a front-end interaction layer configured with a display module for receiving user input text and displaying reply results; an intent recognition layer for determining a user intent according to the input text received by the front-end interaction layer; an RPA execution layer for calling a corresponding preset automated process according to the determined user intent, generating processing result data, and sending the generated result data to the front-end interaction layer; a knowledge graph layer for constructing a knowledge graph according to the historical operation and maintenance data of a photovoltaic power station; a monitoring and optimization layer for monitoring the running state of the intelligent customer service system.

[0006] Optionally, the intent recognition layer includes an intent recognition model constructed based on natural language processing technology. The intent recognition model analyzes the user input text through a preset deep learning algorithm to determine the user intent.

[0007] Optionally, the RPA execution layer includes: an intent matching unit, configured to perform intent matching and call a corresponding automation process after receiving the user intent information transmitted by the intent recognition layer; a task execution unit, configured to simulate manual operations in a corresponding preset business system according to the automation process and in accordance with preset operation steps; a feedback unit, configured to collate the result data after the operation is completed and send the collated data to the front-end interaction layer; and an exception handling unit, configured to send exception information to the monitoring and optimization layer and send an exception prompt to the front-end interaction layer when an exception occurs during the execution of the automation process.

[0008] Optionally, the knowledge graph layer is further configured to: provide semantic understanding support and intelligent decision-making support during semantic understanding by the intent recognition layer and execution of the automation process by the RPA execution layer, locate a faulty device based on the device relationships in the constructed knowledge graph when the monitoring and optimization layer detects a system exception, regularly collect new knowledge and information, and update the knowledge graph to ensure the timeliness and accuracy of the knowledge.

[0009] On the other hand, the present invention further provides an NLP-based intelligent customer service method applied to the above intelligent customer service system. The intelligent customer service method includes: receiving user input text, determining the user intent according to the input text through a preset intent recognition model; calling a corresponding automation process according to the user intent; executing the automation process and feeding back the execution result to the user.

[0010] Optionally, before calling a corresponding automation process according to the user intent, the intelligent customer service method further includes: extracting key features from the historical customer service record data of the photovoltaic power station; analyzing the extracted features using a process mining algorithm to determine common processes in the photovoltaic power station business; based on the determined processes, learning business rules and logic through a preset learning algorithm to construct a process model; generating a code template applicable to the photovoltaic power station customer service scenario according to the constructed process model and business rules; setting variable parameters in the code template according to different customer service scenarios; mapping actual business values to the variable parameters, generating an automation process through a code generation tool; testing the generated automation process and configuring the tested automation process into the intelligent customer service system.

[0011] Optionally, calling a corresponding automation process according to the user intent includes: calculating a similarity using natural language processing technology according to the determined user intent and the intent identifier corresponding to each automation process; if there is an intent identifier whose similarity to the determined user intent exceeds a preset threshold, calling the automation process corresponding to the intent identifier.

[0012] Optionally, performing the automated process and feeding back the execution result to the user includes: according to the key information in the input text, the automated process simulates manual operations in the corresponding preset business system according to the preset operation steps; sorting out the result data after the operations are completed and sending it to the front-end interaction layer.

[0013] Optionally, before calculating the similarity using natural language processing technology according to the determined user intent and the intent identifier of each automated process, the intelligent customer service method further includes: determining the intent identifier corresponding to each automated process.

[0014] Optionally, the intelligent customer service method further includes: evaluating the automated process according to user feedback; optimizing the parameters of the automated process according to the evaluation result.

[0015] An intelligent customer service system and method based on NLP provided by the present invention, with the front-end interaction layer, intent recognition layer, RPA execution layer, knowledge graph layer, and monitoring and optimization layer cooperating with each other by means of NLP and RPA technologies, significantly improves the processing efficiency, can quickly respond to and solve user problems; greatly improves the processing accuracy, reduces misjudgments by relying on the model and knowledge graph; effectively reduces the operation cost, reduces manual dependence and potential error costs; also has strong adaptability and optimization capabilities, and continuously adapts to business changes by updating the knowledge graph, training the model, and optimizing the process, providing efficient and stable customer service support for the operation and management of photovoltaic power stations. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the drawings: Figure 1 is a schematic structural diagram of the intelligent customer service system provided by an embodiment of the present invention; Figure 2 is a flowchart of the RPA execution layer provided by an embodiment of the present invention; Figure 3 is a flowchart of the knowledge graph layer provided by an embodiment of the present invention; Figure 4 is a flowchart of the monitoring and optimization layer provided by an embodiment of the present invention; Figure 5 is a flowchart of the intelligent customer service method provided by an embodiment of the present invention. Detailed Embodiments

[0017] The following will describe in detail the specific implementation manners of the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific implementation manners described herein are only for explaining and illustrating the embodiments of the present invention, and are not used to limit the embodiments of the present invention.

[0018] It should be noted that the acquisition, transmission, storage, use, processing, etc. of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, some existing industry solutions such as certain software, components, models, etc. may be mentioned. They should be regarded as exemplary, and their purpose is only to illustrate the feasibility in the implementation of the technical solution of this application, but it does not mean that the applicant has already or necessarily used this solution.

[0019] The business knowledge and operation processes of photovoltaic power stations are becoming increasingly complex with the continuous innovation of technology. New photovoltaic equipment and operation and maintenance technologies emerge in an endless stream, which requires customer service personnel to continuously learn and update their knowledge systems. However, in actual situations, personnel training often fails to keep up with the speed of technology updates. In addition, the cost of manual customer service is high, including personnel salaries, training costs, and office resources. As the scale of the power station expands, the customer service cost becomes an important part of the operating cost, bringing a heavy economic burden to the enterprise. Therefore, a more intelligent customer service system is developed.

[0020] In view of this problem, the present invention provides an NLP-based intelligent customer service system and method. The intelligent customer service system and method of the present invention, with the aid of NLP and RPA technologies, operate in coordination with the front-end interaction layer, intention recognition layer, RPA execution layer, knowledge graph layer, and monitoring and optimization layer, effectively solving the problems existing in the traditional photovoltaic power station customer service mode, such as low efficiency, poor accuracy, high cost, and difficulty in coping with complex business changes. At the same time, it also has strong adaptation and optimization capabilities, providing an innovative and efficient customer service solution for the operation and management of photovoltaic power stations.

[0021] The following will Figures 1 - 5 specifically describe the present invention.

[0022] As Figure 1 shown, the embodiments of the present invention provide an NLP-based intelligent customer service system. The intelligent customer service system includes: a front-end interaction layer configured with a display module for receiving user input text and presenting reply results; an intention recognition layer for determining the user intention according to the input text received by the front-end interaction layer; an RPA execution layer for calling a corresponding preset automated process according to the determined user intention, generating processed result data, and sending the generated result data to the front-end interaction layer; a knowledge graph layer for constructing a knowledge graph according to the historical operation and maintenance data of the photovoltaic power station; a monitoring and optimization layer for monitoring the running state of the intelligent customer service system.

[0023] The intelligent customer service system provided by the present invention is provided with five levels, namely, a front-end interaction layer, an intent recognition layer, an RPA execution layer, a knowledge graph layer, and a monitoring and optimization layer. The front-end interaction layer receives the user input text. The intent recognition layer determines the user intent according to the received input text. The RPA execution layer calls the corresponding preset automated process according to the determined user intent, generates processed result data, and sends the generated result data to the front-end interaction layer for display to the customer. Through the multi-level collaborative operation, the intelligent customer service system significantly improves the processing efficiency and can quickly respond to and solve user problems.

[0024] Preferably, the intent recognition layer includes an intent recognition model constructed based on natural language processing technology. The intent recognition model analyzes the user input text through a preset deep learning algorithm to determine the user intent.

[0025] In a preferred embodiment of the present invention, the intent recognition model constructed based on natural language processing technology can deeply analyze the user input text, accurately extract keywords in the text, and understand complex semantic structures, so as to accurately judge the user intent among numerous photovoltaic power station professional terms and diverse expressions. Compared with manual judgment based on experience and limited knowledge reserves, the possibility of misjudgment is greatly reduced. In scenarios such as fault repair and business consultation, the core of the problem can be accurately located to ensure that the subsequent processing process is in the correct direction, significantly improving the accuracy of problem processing. In addition, with the development of the photovoltaic power station industry, new equipment and new technologies emerge continuously, and the ways and contents of user consultation questions also change accordingly. Since the intent recognition model is based on a deep learning algorithm, it can continuously adjust the model parameters according to the new user interaction data collected by the monitoring and optimization layer. When new texts such as new business process consultations appear, the model can optimize its ability to recognize new intents by learning new data.

[0026] For example, when a new type of photovoltaic inverter is put into use and users consult about its unique operation problems, the model can accurately recognize such new intents by learning relevant new data, thereby promoting the entire intelligent customer service system to adapt to industry changes and continuously providing efficient support for the operation and management of photovoltaic power stations.

[0027] Such as Figure 2As shown, preferably, the RPA execution layer includes: an intent matching unit, configured to perform intent matching after receiving the user intent information transmitted by the intent recognition layer, and call the corresponding automated process; a task execution unit, configured to simulate manual operations in a corresponding preset business system according to the automated process and in accordance with preset operation steps; a feedback unit, configured to organize the result data after the operation is completed and send the organized data to the front-end interaction layer; and an exception handling unit, configured to send exception information to the monitoring and optimization layer and send an exception prompt to the front-end interaction layer when an exception occurs during the execution of the automated process.

[0028] In a preferred embodiment of the present invention, when the RPA execution layer receives the user intent information transmitted by the intent recognition layer, a set of intent matching mechanisms built into the intent matching unit will quickly determine the specific RPA process to be called based on the predefined correspondence between intent identifiers and automated processes. For example, if the intent recognition layer determines that the user intent is "query order status", the RPA execution layer will immediately call the corresponding order query automated process. After determining the called process, the RPA execution layer passes the key information in the user's question as parameters to the corresponding automated process. Taking the power generation data query intent as an example, information such as the time range and equipment number specified by the user will be accurately transmitted to the data query process. Subsequently, the automated process simulates manual operations in the corresponding business system according to the established operation steps. The RPA tool simulates entering a username and password for login; when querying data, according to the transmitted parameters, it accurately selects query conditions such as the time range and equipment number in the system interface and triggers the query operation. When the automated process completes the task operation in the business system, it will obtain the corresponding result data, and the feedback unit organizes and processes these result data, converting them into a format that is easy for users to understand. For the power generation data query result, the original data may be organized into a table form and data visualization processing may be performed as needed, such as generating bar charts, line charts, etc. Finally, the processed result is fed back to the front-end interaction layer for presentation to the user. In addition, during the process execution, various abnormal situations are inevitable, such as network interruption, system response timeout, operation failure, etc. At this time, the exception handling unit will record detailed error information and send an alarm to the monitoring and optimization layer for subsequent analysis and processing. At the same time, a friendly error prompt message will be fed back to the user, informing the user of the current problem and possible solutions.

[0029] Such as Figure 3As shown, preferably, the knowledge graph layer is further configured to: provide semantic understanding support and intelligent decision-making support during the semantic understanding in the intent recognition layer and the execution of the automated process in the RPA execution layer, and when the monitoring and optimization layer detects a system anomaly, locate the faulty device based on the device relationships in the constructed knowledge graph, regularly collect new knowledge and information, and update the knowledge graph to ensure the timeliness and accuracy of the knowledge.

[0030] In a preferred embodiment of the present invention, with the continuous progress of the technology in the photovoltaic power station industry, new device models, operation and maintenance technologies, and fault cases emerge continuously. The knowledge graph layer connects to multiple channels such as industry authoritative databases, device manufacturer information platforms, and actual operation and maintenance data records of power stations to obtain these new information in a timely manner. For example, when a new type of photovoltaic inverter is put into use, it has unique performance parameters and fault characteristics. After the knowledge graph layer collects this new information, it updates the relevant knowledge nodes and relationships about the inverter in the knowledge graph. This enables the intent recognition layer to understand user inquiries about the new type of inverter, the RPA execution layer to operate based on the latest knowledge when processing tasks involving the new type of inverter, and the monitoring and optimization layer to make more accurate judgments and decisions based on the latest knowledge when analyzing system operation data and handling anomalies, ensuring that the entire intelligent customer service system always adapts to the industry development and provides efficient and reliable services for the operation of photovoltaic power stations.

[0031] As Figure 4 shown, preferably, the monitoring and optimization layer is further configured to: collect the interaction data between the user and the intelligent customer service system, regularly train and optimize the intent recognition model, and update the preset automated process.

[0032] In a preferred embodiment of the present invention, the monitoring and optimization layer collects the interaction data between the user and the intelligent customer service system, and can obtain information such as the type of questions with the highest frequency of user inquiries, the processing duration under different intents, and the feedback evaluation of the user on the processing results. These data provide key basis for optimizing the processes in the RPA execution layer. At the same time, based on the collected data, the monitoring and optimization layer can discover the deviation of the intent recognition model in the judgment of certain specific user intents. For example, in the scenario of reporting a fault of photovoltaic power station equipment, if a large number of users feedback that the problems are not accurately understood, the monitoring and optimization layer can use these interaction data to conduct targeted training on the intent recognition model. Adjust the model parameters, optimize the feature extraction and semantic understanding capabilities of the text describing the equipment fault, so that the intent recognition layer can more accurately judge the user intent and provide more reliable intent information for the RPA execution layer, ensuring the correctness of the subsequent automated process calls.

[0033] As Figure 5 shown, an NLP-based intelligent customer service method provided in an embodiment of the present invention is applied to the above intelligent customer service system, and the intelligent customer service method includes: Step 101: Receive the user input text and determine the user intention according to the input text; Step 102: Call the corresponding automated process according to the user intention; Step 103: Execute the automated process and feedback the execution result to the user.

[0034] For example, when the user wants to query the order status, the intelligent customer service system will analyze the user intention according to the user input text, determine that the customer wants to query the order status, then the intelligent customer service system will call the automated process corresponding to querying the order status, execute this process, and finally feedback the executed order query interface to the user.

[0035] Preferably, the receiving the user input text and determining the user intention according to the input text includes: receiving the user input text and determining the user intention through a preset intention recognition model, wherein the user intention includes querying the order status, modifying account information, and power generation data query.

[0036] The front-end interaction layer in the intelligent customer service system provided by the present invention has a multi-modal interaction function. Whether the user inputs through voice or directly enters a question in the simple and intuitive text input box, it can be effectively collected. When the user uses voice input, the front-end interaction layer uses advanced speech recognition technology to quickly convert the user voice into text; if it is text input, the user input content is directly obtained. Subsequently, this text information is quickly transmitted to the intention recognition layer, and the graph recognition model in the intention recognition layer deeply analyzes the user input text to accurately recognize the user intention.

[0037] For example, when the user inputs "I want to know the power generation data in the power station", the intention recognition module analyzes keywords such as "power generation data" in the text, and combines the learning of knowledge in the photovoltaic power station field to determine that the user intention is power generation data query. Once the user intention is determined, the intention recognition layer will transmit the recognition result to the RPA execution layer, providing a basis for subsequent calling of the corresponding automated process and promoting the smooth progress of the entire intelligent customer service process.

[0038] Preferably, before invoking the corresponding automated process according to the user intention, the intelligent customer service method further includes: extracting key features from the historical customer service record data of the photovoltaic power station; using a process mining algorithm to analyze the extracted features to determine common processes in the photovoltaic power station business; based on the determined processes, learning business rules and logic through a preset learning algorithm to construct a process model; generating a code template applicable to the customer service scenario of the photovoltaic power station according to the constructed process model and business rules; setting variable parameters in the code template according to different customer service scenarios; mapping actual business values to the variable parameters, and generating an automated process through a code generation tool; testing the generated automated process, and configuring the tested automated process into the intelligent customer service system.

[0039] In a preferred embodiment of the present invention, before invoking the automated process, these processes need to be automatically generated for selection. During the generation of the automated process, key data will first be extracted from the historical data of the photovoltaic power station. These data can cover common operations in the photovoltaic power station business and extract features. Subsequently, a process mining algorithm (such as heuristic mining, genetic mining) is used to analyze these features to identify common business processes in the photovoltaic power station. Then, according to these processes, a machine learning algorithm (such as decision tree, random forest) is used to learn business rules and logic to construct a process model, defining the input, output, and execution conditions of each step, providing a basis for subsequent automated process generation. Subsequently, according to the process model and business rules, a code template applicable to the intelligent customer service scenario of the photovoltaic power station is created. The template contains common operation instructions, such as modifying information, querying data, querying bills, etc., facilitating subsequent rapid script generation. Subsequently, the code template is parameterized. According to different business scenarios and user requirements, the parameters in the script are dynamically adjusted. For example, in the process of querying the order status, according to the order information input by the user, the corresponding form fields are automatically filled. Subsequently, using a code generation algorithm, according to the process model and parameterized configuration, an automated process script can be automatically generated. Finally, the generated script is subjected to syntax checking and testing, and then configured into the system for use. This method of automatically generating automated processes can greatly improve the business response speed and reduce human consumption.

[0040] For example, for the construction of the order status query process, first, order data related to material procurement, equipment maintenance, etc. in the past six months of the photovoltaic power station will be collected, including order numbers, order placement times, supplier information, order processing progress records, and communication records of customer service and users regarding order inquiries. Data mining algorithms are used to extract key information from the order data. The genetic algorithm in process mining is used to analyze the extracted data features. Through the study of a large number of order processing records, the common process of order status query is sorted out. For example, the process: the user initiates an order status query request, the customer service obtains the user's order number, queries the order status in the order management system, organizes the order status information and feedbacks it to the user. Subsequently, based on the sorted out common process, a process model for order status query is constructed using the Bayesian network learning algorithm. The conditions and logical relationships of each step are clarified. For example, if the current order status is "shipped", then the next step is to obtain the logistics order number and the logistics query link. The model will automatically deduce the subsequent operations and feedback content based on the information at different stages of the order. When the user initiates an order status query, the order number provided by the user is mapped to the variable parameters of the code template. With the help of a code generation tool, a complete automated process script is generated. The script will establish a connection with the order management database, execute the query operation, and organize the query results into a format that is easy for users to understand.

[0041] Preferably, in order to optimize the automated process, the intelligent customer service method further includes: evaluating the automated process according to user feedback; optimizing the parameters of the automated process according to the evaluation results.

[0042] For example, after the user uses the order status query function, user feedback is collected through online questionnaires, customer service follow-up visits, etc. For example, the user feedback shows that the order status update in the query result is not timely, or the display format of the query result is not easy to understand. The system evaluates the automated process based on these feedbacks to analyze whether there is a delay in the data acquisition link or a problem in the result organization and display link. For example, if the evaluation finds that there is a data acquisition delay, it may be a problem with the execution efficiency of the database query statement, and the parameters of the query statement can be optimized, such as adding indexes, adjusting the order of query conditions, etc. If it is a display format problem, optimize the parameters of the result organization part, such as adjusting the date format, adding field descriptions, etc., to improve the user experience. After optimization, test again to ensure that the problem is solved, and then redeploy the optimized automated process to the intelligent customer service system. By collecting user feedback to evaluate the process and optimize the parameters, the service quality of the intelligent customer service system can be improved.

[0043] Preferably, the step of invoking the corresponding automation process according to the user intention includes: calculating the similarity using natural language processing technology based on the determined user intention and the intention identifier corresponding to each automation process; if there is an intention identifier whose similarity with the determined user intention exceeds a preset threshold, then invoking the automation process corresponding to this intention identifier.

[0044] More preferably, before calculating the similarity using natural language processing technology based on the determined user intention and the intention identifier of each automation process, the intelligent customer service method further includes: determining the intention identifier corresponding to each automation process.

[0045] In a preferred embodiment of the present invention, after determining the user intention, the intention matching unit first traverses the pre-constructed intention identifier system, which covers common user intentions in the field of photovoltaic power station services such as querying order status, modifying account information, and querying power generation data, and assigns a unique and clear identifier to each intention. Subsequently, the intention matching unit uses the text similarity algorithm in natural language processing technology to calculate the semantic similarity between the user intention and each intention identifier, and then determines the automation process.

[0046] For example, taking the user input "I want to check the power generation data of the inverter last month" as an example, after the intention recognition layer determines that the user intention is power generation data query and passes it to the RPA execution layer. The intention matching unit calculates the similarity between this intention and each intention identifier one by one. When the similarity with the intention identifier of "power generation data query" exceeds the preset threshold, such as 0.8, after being calculated by the cosine similarity algorithm, it is determined that the match is successful, and then the corresponding automation process for power generation data query is invoked.

[0047] Preferably, the step of executing the automation process and feeding back the execution result to the user includes: according to the key information in the input text, the automation process simulates manual operations in the corresponding preset business system according to the preset operation steps; sorts out the result data after the operations are completed and sends it to the front-end interaction layer.

[0048] For example, taking the user's query "Query the power generation data of PV panel A last month" as an example, the intent recognition layer determines that the user's intent is to query power generation data and passes it to the RPA execution layer. The task execution unit extracts key information such as "last month" and "PV panel A" from the user input text and passes them as parameters to the automated process for querying power generation data. This process follows the preset operation steps to simulate manual login to the data monitoring business system of the PV power station. In the system, through automated scripts, operations such as mouse clicks and keyboard inputs are simulated. In the specified data query interface, the query time range is accurately filled in as last month, and the device is precisely selected as PV panel A, and then the query instruction is triggered. When the automated process completes the operation in the business system and obtains the result data, the feedback unit takes over the work, organizes the result data, sends the organized result data to the front-end interaction layer, and finally the front-end interaction layer displays these results to the user in an intuitive form.

[0049] In summary, an intelligent customer service system and method provided by the present invention constructs an intelligent customer service system for PV power station operation and maintenance by integrating natural language processing (NLP), robotic process automation (RPA), and dynamic knowledge graph technology, significantly improving service efficiency and accuracy: the system can accurately parse professional terms and complex intents, automatically execute cross-platform data query and fault handling operations, reducing manual intervention; the dynamic update of the knowledge graph and the closed-loop feedback mechanism of the monitoring and optimization layer ensure that the system continuously adapts to changes in the operation and maintenance scenario, and at the same time support intelligent correlation positioning and root cause analysis of faulty equipment, ultimately realizing efficient, adaptive, and low-maintenance-cost intelligent operation and maintenance services.

[0050] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution, and the order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0051] In addition, the terms "system" and "network" are often used interchangeably in this article. The term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the front and rear associated objects.

[0052] It should be understood that in the embodiments of the present invention, "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0053] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0054] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0055] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed couplings or direct couplings or communication connections to each other can be indirect couplings or communication connections through some interfaces, devices, or units, and can also be electrical, mechanical, or other forms of connection.

[0056] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present invention.

[0057] In addition, the functional units in various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0058] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by hardware, or by firmware, or by a combination thereof. When implemented in software, the above functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. The computer-readable medium includes computer storage media and communication media, where the communication media includes any medium that facilitates the transfer of a computer program from one place to another. The storage media can be any available medium that can be accessed by a computer. By way of example but not limitation: the computer-readable medium can include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer. In addition, any connection can suitably be a computer-readable medium. For example, if the software is transmitted using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL) or wireless technologies such as infrared, radio and microwave from a website, server or other remote source, then the coaxial cable, fiber optic cable, twisted pair, DSL or wireless technologies such as infrared, wireless and microwave are included in the definition of the medium. As used in the present invention, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, where disk generally magnetically replicates data, while disc optically replicates data with a laser. The above combinations should also be included within the scope of protection of the computer-readable medium.

[0059] In summary, the above are only the preferred embodiments of the technical solution of the present invention, and are not intended to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An intelligent customer service system based on NLP, characterized in that: The intelligent customer service system comprises: The front-end interaction layer is equipped with a display module for receiving user input text and displaying reply results; The intention recognition layer is used to determine the user intention based on the input text received by the front-end interaction layer; The RPA execution layer is used to call the corresponding preset automation process according to the determined user intention, generate processing result data, and send the generated result data to the front-end interaction layer; The knowledge graph layer is used to build a knowledge graph based on the historical operation and maintenance data of photovoltaic power plants; The monitoring and optimization layer is used to monitor the operating status of the intelligent customer service system.

2. The intelligent customer service system according to claim 1, characterized in that: The intent recognition layer includes an intent recognition model built based on natural language processing technology. The intent recognition model analyzes the user input text through a preset deep learning algorithm to determine the user's intent.

3. The intelligent customer service system according to claim 1, characterized in that: The RPA execution layer includes: An intention matching unit, configured to perform intention matching and call a corresponding automation process after receiving the user intention information transmitted by the intention recognition layer; A task execution unit, configured to simulate manual operations in a corresponding preset business system according to the automation process and preset operation steps; A feedback unit, used to sort out the result data after the operation is completed, and send the sorted data to the front-end interaction layer; The exception handling unit is used to send the exception information to the monitoring and optimization layer and send the exception prompt to the front-end interaction layer when an exception occurs during the execution of the automated process.

4. The intelligent customer service system according to claim 1, characterized in that: The knowledge graph layer is also used to: Provide semantic understanding support and intelligent decision support during the semantic understanding of the intention recognition layer and the execution of the automation process by the RPA execution layer. When the monitoring and optimization layer detects a system anomaly, it locates the faulty device based on the device relationship in the constructed knowledge graph. Regularly collect new knowledge and information, update the knowledge graph, and ensure the timeliness and accuracy of knowledge.

5. An intelligent customer service method based on NLP, characterized in that: Applied to the intelligent customer service system described in claims 1-4, the intelligent customer service method comprises: Receive user input text and determine the user's intention through a preset intention recognition model; According to the user's intention, a corresponding automated process is called; Execute the automated process and feed back the execution result to the user.

6. The intelligent customer service method according to claim 5, characterized in that: Before calling the corresponding automated process according to the user intention, the intelligent customer service method further includes: Extract key features based on historical customer service record data of photovoltaic power stations; Using process mining algorithms, the extracted features are analyzed to identify common processes in photovoltaic power station business; Based on the determined process, the business rules and logic are learned through the preset learning algorithm to build a process model; Generate code templates suitable for photovoltaic power station customer service scenarios based on the constructed process model and business rules; According to different customer service scenarios, variable parameters are set in the code template; Mapping the actual business value to the variable parameter, and generating an automated process through a code generation tool; The generated automated process is tested, and the tested automated process is configured to the intelligent customer service system.

7. The intelligent customer service method according to claim 5, characterized in that: The calling of the corresponding automated process according to the user intention includes: Calculate similarity using natural language processing technology based on the determined user intent and the intent identifier corresponding to each of the automated processes; If there is an intent identifier whose similarity with the determined user intent exceeds a preset threshold, the automated process corresponding to the intent identifier is called.

8. The intelligent customer service method according to claim 5, characterized in that: The executing the automated process and feeding back the execution result to the user comprises: According to the key information in the input text, the automated process simulates manual operations in the corresponding preset business system according to preset operation steps; The result data after the operation is completed is sorted and sent to the front-end interaction layer.

9. The intelligent customer service method according to claim 7, characterized in that: Before calculating similarity based on the determined user intent and the intent identifier corresponding to each of the automated processes using natural language processing technology, the intelligent customer service method further includes: Determine the intent identifier corresponding to each of the automated processes.

10. The intelligent customer service method according to claim 6, characterized in that: The intelligent customer service method also includes: Evaluate the automated process based on user feedback; Based on the evaluated results, the parameters of the automated process are optimized.

Images