Comprehensive energy intelligent session design method and device based on large language model

Abstract

The invention discloses an integrated energy intelligent session design method and device based on a large language model, and the method comprises the steps: transmitting a session text and an intention recognition prompt word to the large language model, so as to recognize a session intention; the intention recognition prompt word comprises a plurality of intentions, and each intention is determined according to a business scene demand provided by each business party; determining a corresponding interface and a parameter extraction requirement cue word corresponding to a parameter required by the interface according to the session intention, and transmitting the parameter extraction requirement cue word to a large language model to extract the parameter; interface and parameter extraction requirement cues are generated by corresponding business parties; and calling the service system interface according to the parameters. According to the method, the intelligent session is split into intention recognition and parameter extraction, each stage corresponds to a respective cue word, the cue word in the intention recognition stage can be generated by a general team, and the cue word and an interface in the parameter extraction stage are independently developed by a corresponding service party, so that redundancy and disorder of the cue words are reduced, and the session efficiency and quality are improved.

Description

Technical Field

[0001] The present application relates to the field of integrated energy technology, and in particular to an integrated energy intelligent conversation design method and device based on a large language model. Background Art

[0002] With the rapid development of artificial intelligence technology, large language models have brought new changes to intelligent conversations with their powerful language understanding and generation capabilities. Currently, many companies are connecting to large language models for intelligent conversation functions.

[0003] In the integrated energy scenario, various energy businesses can interact through the intelligent conversation system. At present, when combining the large language model with various energy business systems of the company, most of the business requirements of all energy business systems in the integrated energy are written in a large prompt word, and a conversation is initiated to the large language model through a single request. In the above conversation method, multiple teams need to write the same set of prompt words together, which makes collaboration difficult, and makes the prompt words redundant and confusing, reducing the efficiency and quality of intelligent conversations.

[0004] To sum up, how to reduce the redundancy and confusion of prompt words used in intelligent conversations in integrated energy scenarios and improve the efficiency and quality of intelligent conversations in integrated energy scenarios are technical problems that need to be urgently solved by technical personnel in this field. Summary of the invention

[0005] In view of this, the purpose of this application is to provide a method and device for designing integrated energy intelligent conversations based on a large language model, which is used to reduce the redundancy and confusion of prompt words used in intelligent conversations in integrated energy scenarios, and improve the efficiency and quality of intelligent conversations in integrated energy scenarios.

[0006] In order to achieve the above objectives, this application provides the following technical solutions:

[0007] A comprehensive energy intelligent conversation design method based on a large language model, comprising:

[0008] Receive a conversation text input by a user, pass the conversation text and the intention recognition prompt words in the integrated energy scenario to a large language model, and the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario;

[0009] Determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, pass the parameter extraction requirement prompt words to the large language model, and the large language model extracts the parameters required by the business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one-to-one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one-to-one, and the business system interface and the corresponding parameter extraction requirement prompt words are generated by the business party of the corresponding energy business system;

[0010] The business system interface is called according to the parameters required by the energy business system interface to obtain a session result.

[0011] Optionally, after calling the energy business system interface according to the parameters required by the energy business system interface to obtain a session result, the method further includes:

[0012] Acquire a session result processing prompt word corresponding to the user session intention; the session result processing prompt word includes a processing method for the session result corresponding to the user session intention, and the session result processing prompt word is generated by a business party of the corresponding energy business system;

[0013] The conversation result and the conversation result processing prompt word are passed to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word to obtain a processing result.

[0014] Optionally, after the conversation result and the conversation result processing prompt word are transmitted to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word to obtain the processing result, the method further includes:

[0015] The processing result is displayed.

[0016] Optionally, when the parameter extraction requirement prompt word is passed to the large language model, the method further includes:

[0017] Passing the conversation text to the large language model;

[0018] When the conversation result and the conversation result processing prompt word are transmitted to the large language model, the method further includes:

[0019] The conversation text is passed to the large language model.

[0020] Optionally, each of the intents in the intent recognition prompt word includes an intent name and an intent description.

[0021] Optionally, each of the intentions in the intention recognition prompt word further includes a priority;

[0022] When the large language model identifies multiple user conversation intentions, the method further includes:

[0023] The user session intent with the highest priority is determined as the final user session intent.

[0024] Optionally, each of the intents corresponds to a parameter extraction module, and the parameter extraction module includes a business system interface corresponding to the intent and parameters required by the business system interface.

[0025] A comprehensive energy intelligent conversation design device based on a large language model, comprising:

[0026] A receiving module is used to receive a conversation text input by a user, and pass the conversation text and the intention recognition prompt words in the integrated energy scenario to the large language model, so that the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario;

[0027] A first determination module is used to determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, and pass the parameter extraction requirement prompt words to the large language model, so that the large language model extracts the parameters required by the energy business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one-to-one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one-to-one, and the business system interface and the corresponding parameter extraction requirement prompt words are generated by the business party of the corresponding energy business system;

[0028] The calling module is used to call the energy business system interface according to the parameters required by the energy business system interface to obtain a session result.

[0029] A comprehensive energy intelligent conversation design device based on a large language model, comprising:

[0030] Memory for storing computer programs;

[0031] A processor is used to implement the steps of the comprehensive energy intelligent conversation design method based on a large language model as described in any of the above items when executing the computer program.

[0032] A readable storage medium having a computer program stored therein, wherein the computer program, when executed by a processor, implements the steps of the comprehensive energy intelligent conversation design method based on a large language model as described in any one of the above items.

[0033] The present application provides a method and device for designing an integrated energy intelligent conversation based on a large language model, wherein the method comprises: receiving a conversation text input by a user, passing the conversation text and an intention recognition prompt word in an integrated energy scenario to a large language model, and the large language model identifying the user conversation intention according to the intention recognition prompt word; the intention recognition prompt word contains multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario; determining the corresponding energy business system interface and the parameter extraction requirement prompt word corresponding to the parameters required by the energy business system interface according to the user conversation intention, passing the parameter extraction requirement prompt word to the large language model, and the large language model extracting the parameters required by the energy business system interface according to the parameter extraction requirement prompt word; each intention in the intention recognition prompt word corresponds to the energy business system interface one-to-one, and the energy business system interface corresponds to the parameter extraction requirement prompt word corresponding to the required parameters of the energy business system interface one-to-one, and the energy business system interface and the corresponding parameter extraction requirement prompt word are generated by the corresponding business party; calling the business system interface according to the parameters required by the business system interface to obtain the conversation result.

[0034] The above-mentioned technical scheme disclosed in the present application not only splits the integrated energy intelligent conversation process into intent recognition and parameter extraction, with each stage corresponding to its own prompt words, but also the intent recognition prompt words used in the integrated energy scenario in the intent recognition stage can be developed and generated by a general team, and the energy business system interface corresponding to each intent in the intent recognition prompt words and the corresponding parameter extraction prompt words are independently developed by the business parties of the corresponding energy business systems in the integrated energy, without the need to bring together the prompt words for all energy business requirements and have multiple teams write the same set of prompt words together, thereby reducing the bloatedness, redundancy and confusion of the prompt words used in intelligent conversations in integrated energy scenarios, improving the efficiency and quality of intelligent conversations in integrated energy scenarios, and enabling the business parties of each energy business system to develop their own prompt words respectively, thereby improving development collaboration capabilities, reducing the complexity of team collaboration, and effectively integrating multiple business processes. Moreover, even if new energy services are connected, it is only necessary to determine the intent based on the energy business scenario requirements of the new energy services and write the intent into the intention recognition prompt words under the comprehensive energy scenario. The business party corresponding to the new energy services will generate corresponding parameter extraction requirement prompt words and store them in the intelligent conversation system without redesigning the entire process, reducing development time delays and complexity and improving development efficiency.

[0035] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flowchart of a method for designing an integrated energy intelligent conversation based on a large language model provided in an embodiment of the present application;

[0037] Figure 2 A flowchart of intention recognition in a comprehensive energy scenario provided in an embodiment of the present application;

[0038] Figure 3 A flowchart for extracting parameters required for the photovoltaic daily report interface provided in the embodiment of the present application;

[0039] Figure 4 A flowchart of another integrated energy intelligent conversation design method based on a large language model provided in an embodiment of the present application;

[0040] Figure 5 A flowchart of actual business development and intelligent conversation applicable to integrated energy scenarios provided in the embodiments of the present application;

[0041] Figure 6 A flowchart of calling the business system interface corresponding to the photovoltaic operation daily query intention and determining the corresponding session result processing prompt word provided in the embodiment of the present application;

[0042] Figure 7 A schematic diagram of the structure of a comprehensive energy intelligent conversation design device based on a large language model provided in an embodiment of the present application;

[0043] Figure 8 A schematic diagram of the structure of a comprehensive energy intelligent conversation design device based on a large language model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0044] With the rapid development of artificial intelligence technology, large language models have brought new changes to intelligent conversations with their powerful language understanding and generation capabilities. At present, many companies are connecting to large language models for intelligent conversation functions. In comprehensive energy management, various energy (such as photovoltaics, HVAC, boilers, etc.) businesses usually need to interact through intelligent conversation systems.

[0045] At present, when combining the big language model with various energy business systems, most of them write the business requirements of all energy business systems in the integrated energy into a big prompt word, and initiate a session to the big language model through a request. The specific implementation process is: the user enters the query text through the corresponding intelligent conversation system; the intelligent conversation system collects the text entered by the user and the prompt words of all business requirements together to form a big prompt word and send it to the big language model; the big language model parses the prompt word and generates the corresponding system query results.

[0046] In the comprehensive energy scenario, there are many business scenarios such as photovoltaic, energy storage, HVAC, boilers, etc. Different business scenarios have their own unique business processes and business data. All business requirements are encapsulated in one prompt word. When there are more and more business scenarios and the business requirements are increasingly different, the above method has the following problems:

[0047] 1. Complexity: There are more and more prompt words, which are bloated and lack a unified processing framework. As a result, the intelligent conversation system reacts complexly after the user inputs, making it difficult to meet the needs in real time.

[0048] 2. Chaos: Various business prompts may affect each other, causing online problems and affecting the quality of intelligent conversations;

[0049] 3. Low collaboration efficiency: Multiple teams write the same set of prompts together, which makes collaboration difficult;

[0050] 4. Poor scalability: When new energy services are connected, prompts and processes need to be redesigned, causing delays in development time.

[0051] To this end, the present application provides a method and device for designing integrated energy intelligent conversations based on a large language model, which is used to reduce the redundancy and confusion of prompt words used in intelligent conversations in integrated energy scenarios, and improve the efficiency and quality of intelligent conversations in integrated energy scenarios.

[0052] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0053] See also Figure 1 , which shows a flowchart of a comprehensive energy intelligent conversation design method based on a large language model provided in an embodiment of the present application. A comprehensive energy intelligent conversation design method based on a large language model provided in an embodiment of the present application may include:

[0054] S11: Receive the conversation text input by the user, pass the conversation text and the intention recognition prompt words in the integrated energy scenario to the large language model, and the large language model recognizes the user conversation intention based on the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined based on the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario.

[0055] It should be noted that the execution subject of the method provided in the embodiment of the present application may be an intelligent conversation system. Moreover, the comprehensive energy scenario mentioned in the embodiment of the present application may include multiple energy business systems, such as photovoltaic business system, energy storage business system, HVAC business system, etc.

[0056] Before conducting an intelligent conversation, the business parties of each energy business system in the integrated energy scenario (specifically, the business teams corresponding to each energy business system, such as the photovoltaic team corresponding to the photovoltaic business system, the energy storage team corresponding to the energy storage business system, the HVAC team corresponding to the HVAC business system, etc.) can send or inform or submit the energy business scenario requirements in their respective energy business systems to the general team, and the general team will convert the business scenario requirements into independent intentions through communication, semantic recognition, etc., and write the converted intentions into the intention recognition prompt words under the integrated energy scenario. That is, before the intelligent conversation, the general team can convert the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario into independent intentions, and write the converted intentions into the intention recognition prompt words, so as to realize the pre-construction of the intention recognition prompt words under the integrated energy scenario. Among them, the pre-constructed intention recognition prompt words under the integrated energy scenario contain multiple intentions, each of which is converted from the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario, so that the large language model can determine the user conversation intention based on the intention recognition prompt words under the integrated energy scenario. It should be noted that the pre-built intent recognition prompt words in the integrated energy scenario can be stored in the intelligent conversation system, so as to facilitate intelligent conversations suitable for the integrated energy scenario based on the intent recognition prompt words in the integrated energy scenario. In addition, while the general team converts the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario into intents, it can also determine the correspondence between intents and business parties, so that each intent can be informed to the corresponding business party respectively, so that the business party can independently develop and deploy the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the energy business system interface.

[0057] Exemplarily, for a comprehensive energy scenario that includes energy business systems such as photovoltaics, energy storage, and HVAC, the photovoltaic business party can provide the corresponding photovoltaic business scenario requirements (such as photovoltaic operation daily query requirements, photovoltaic revenue query requirements, photovoltaic alarm query requirements, etc.) to the general team, the energy storage business party can provide the corresponding energy storage business scenario requirements (such as energy storage revenue query, energy storage power query, etc.) to the general team, and the HVAC business party can provide the corresponding HVAC business scenario requirements (such as terminal temperature query, intelligent control level query, etc.) to the general team. The general team will convert the energy business scenario requirements provided by the business parties of each energy business system in the comprehensive energy scenario into independent intents, and write the converted intents into the intent recognition prompt words under the comprehensive energy scenario.

[0058] In one embodiment, each intention in the intention recognition prompt words in the comprehensive energy scenario may include an intention name and an intention description.

[0059] In the embodiment of the present application, each intention included in the pre-constructed intention recognition prompt words in the comprehensive energy scenario can include an intention name and an intention description, wherein the intention description is a description of the corresponding intention name, which is used to describe under what circumstances the corresponding intention name is hit, so as to facilitate the large language model to use semantic recognition and other methods to identify and match the intention name corresponding to the conversation text input by the user based on the conversation text input by the user and the intention description, and identify the corresponding intention name as the user conversation intention. Exemplarily, for the intention of photovoltaic operation daily query, it can include the intention name: photovoltaic operation daily query, and the intention description: the user needs to obtain the photovoltaic operation daily report.

[0060] When conducting an intelligent conversation, the user can input a conversation text (i.e., the question the user wants to ask) into the corresponding intelligent conversation system. Then, the intelligent conversation system can pass the conversation text input by the user and the pre-built intention recognition prompt words in the integrated energy scenario to the large language model, and the large language model recognizes the user conversation intention based on the intention recognition prompt words and the conversation text input by the user. If the large language model recognizes the user conversation intention based on the intention recognition prompt words in the integrated energy scenario and the conversation text input by the user, it indicates that the conversation text input by the user is related to the integrated energy scenario. Specifically, the large language model can match the intention description of each intention in the intention recognition prompt words in the integrated energy scenario and the conversation text input by the user, and recognize the intention name corresponding to the intention description that matches the conversation text input by the user as the user conversation intention. After the large language model identifies the user's conversation intention, the intelligent conversation system can obtain the user's conversation intention corresponding to the conversation text input by the user, and the intelligent conversation system can display and / or send the user's conversation intention to the general team and the business party of the corresponding energy business system according to the user's needs, so as to facilitate the acquisition of the user's conversation intention identified by the large language model. Sending the user's conversation intention to the general team also facilitates the general team to edit and modify the intention recognition prompt words according to the specific situation. If the large language model cannot identify the user's conversation intention according to the intention recognition prompt words in the integrated energy scenario, it means that the conversation text input by the user is irrelevant to the integrated energy scenario. At this time, the large language model can directly identify the user's intention based on the conversation text input by the user, and can display the user's intention or reply according to the user's intention, etc., or the large language model can feedback the prompt that the user's conversation intention cannot be correctly identified, so that the user can enter the conversation text according to the prompt. Alternatively, after receiving the conversation text input by the user, it can be determined whether the conversation text input by the user is related to the integrated energy scenario based on the pre-constructed integrated energy scenario knowledge base, wherein the integrated energy scenario knowledge base can be constructed by the general team and / or the business parties of each energy business system in the integrated energy scenario according to the composition of the integrated energy scenario, the industry terms corresponding to each energy business system in the integrated energy scenario, etc. When judging, it can be specifically determined whether the conversation text input by the user contains at least one keyword in the pre-constructed integrated energy scenario knowledge base. If so, it is determined that the conversation text input by the user is related to the integrated energy scenario. If not, it is determined that the conversation text input by the user is not related to the integrated energy scenario. When it is determined that the conversation text input by the user is related to the integrated energy scenario, the conversation text input by the user and the intention recognition prompt words under the integrated energy scenario can be passed to the large language model, and the large language model recognizes the user's conversation intention based on the intention recognition prompt words under the integrated energy scenario.When it is determined that the conversation text input by the user is not related to the integrated energy scenario, a conversation text input prompt can be fed back to the user based on a pre-built integrated energy scenario knowledge base, so that the user can enter the conversation text according to the conversation text input prompt. Alternatively, the conversation text input by the user can be passed to a large language model, which directly recognizes the user's intention based on the conversation text input by the user, and displays the user's intention or responds according to the user's intention.

[0061] For example, in a comprehensive energy scenario, Figure 2 As shown, it shows a flowchart of intent recognition in an integrated energy scenario provided in an embodiment of the present application. The intelligent conversation system passes the conversation text input by the user and the intention recognition prompt word in the integrated energy scenario to the large language model, and the large language model recognizes that the user's conversation intention is an alarm query, a daily query of photovoltaic operation, a photovoltaic income query, an energy storage income query, etc.

[0062] In addition, in one embodiment, each intention in the intention recognition prompt word in the comprehensive energy scenario may also include a priority;

[0063] When a large language model identifies multiple user conversation intents, it may also include:

[0064] The user session intent with the highest priority is determined as the final user session intent.

[0065] Specifically, when the general team converts the energy business scenario requirements provided by the business parties of each energy business system in the comprehensive energy scenario into intentions, it can also set priorities for the converted intentions based on the needs and / or specific circumstances of the business parties.

[0066] On the basis of the above, when the large language model identifies multiple user conversation intentions based on the conversation text and intention recognition prompts in the integrated energy scenario, the priority of each user conversation intention can be obtained, and the user conversation intention with the highest priority can be determined as the final user conversation intention, so as to improve the reliability of intent recognition.

[0067] Exemplarily, when the large language model identifies the three user conversation intentions of photovoltaic profit query, energy storage profit query, and photovoltaic and energy storage profit query based on the intention recognition prompt words in the comprehensive energy scenario and the conversation text input by the user, among which the priority of the photovoltaic and energy storage profit query intention is greater than the priority of the photovoltaic profit query and energy storage profit query intentions, then the photovoltaic and energy storage profit query intention can be determined as the final user conversation intention.

[0068] Of course, the above process of determining the final user conversation intention can also be implemented by a large language model.

[0069] S12: Determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, pass the parameter extraction requirement prompt words to the large language model, and the large language model extracts the parameters required for the energy business system interface according to the parameter extraction requirement prompt words; there is a one-to-one correspondence between each intention in the intention recognition prompt words and the energy business system interface, and there is a one-to-one correspondence between the energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface, and the energy business system interface and the corresponding parameter extraction requirement prompt words are generated by the business party of the corresponding energy business system.

[0070] In addition, before the intelligent conversation, the general team can inform the corresponding business party of the converted intent after converting the energy business scenario requirements provided by the business parties of each energy business system in the comprehensive energy scenario into the corresponding intent. After obtaining the corresponding intent, the corresponding business party can independently develop and deploy the energy business system interface corresponding to the intent and the parameters required by the energy business system interface (for example, time, enterprise, user information, etc.), and can determine the extraction requirements of the parameters required by the energy business system interface (for example, extraction format, etc.), and can write the extraction requirements of the parameters required by the energy business system interface into the parameter extraction requirement prompt words, so that the large language model can extract the parameters required by the business system interface according to the parameter extraction requirements. Among them, each intention in the intent recognition prompt words under the comprehensive energy scenario can correspond one-to-one with the energy business system interface, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface can also correspond one-to-one, so as to reduce the redundancy and confusion of the prompt words used in the intelligent conversation in the comprehensive energy scenario, and facilitate the improvement of the efficiency of the intelligent conversation based on the one-to-one corresponding energy business system interface. The energy business system interface is used for the intelligent conversation system to call in order to obtain the conversation result corresponding to the intent (that is, the interface result obtained after calling the corresponding energy business system interface, or the expected return result behind the corresponding intent), and the parameters required by the energy business system interface are the parameters required for calling the energy business system interface. That is, the business parties of each energy business system in the comprehensive energy scenario can independently develop and deploy the energy business system interface corresponding to the corresponding intent and the parameter extraction requirements corresponding to the parameters required by the energy business system interface, without the need for multiple teams to write the same set of prompt words together, so as to reduce the complexity of team collaboration, improve the overall development efficiency, reduce the redundancy and bloat of the prompt words, and avoid the prompt words of all energy business systems from being mixed together and affecting each other, reducing the confusion of the prompt words. Among them, the corresponding energy business system interface generated by each business party and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface can be stored in the intelligent conversation system.

[0071] After the large language model recognizes the user's conversation intention, the intelligent conversation system can determine the corresponding energy business system interface and the parameter extraction prompt words corresponding to the parameters required by the energy business system interface according to the user's conversation intention, and can pass the parameter extraction prompt words to the large language model so that the large language model can extract the parameters required by the energy business system interface according to the parameter extraction requirement prompt words. Among them, the parameter extraction requirement prompt words contain the extraction requirements of the parameters required by the energy business system interface, and the large language model can extract the parameters required by the energy business system interface according to the extraction requirements of the parameters required by the energy business system interface contained in the parameter extraction requirement prompt words.

[0072] For example, see Figure 3 , which shows a flow chart for extracting the parameters required for the photovoltaic daily report interface provided in an embodiment of the present application. When the large language model recognizes that the user's conversation intention is to query the photovoltaic operation daily report, the intelligent conversation system can determine that this query needs to call the photovoltaic daily report interface. The parameters required by the interface are the enterprise id and date. The date is a certain day and has a fixed format requirement. The parameter prompt words required for the photovoltaic daily report interface generated by the business party (also referred to as the photovoltaic daily report parameter extraction prompt words) are determined as follows: the date to be queried is extracted according to the chat context, and the format is YYY-MM-DD. The parameters extracted by the large language model according to the parameter prompt words required for the photovoltaic daily report interface are: enterprise id, date.

[0073] S13: calling the energy business system interface according to the parameters required by the energy business system interface to obtain a session result.

[0074] After obtaining the required parameters of the corresponding energy business system interface extracted by the large language model, the intelligent conversation system can call the energy business system interface according to the required parameters of the energy business system interface to obtain the conversation result (or interface result) to respond to the conversation text input by the user. Among them, after obtaining the conversation result, the intelligent conversation system can display the conversation result to facilitate the user to obtain the conversation result. It should be noted that when conducting an intelligent conversation in a comprehensive energy scenario, it is possible to realize energy knowledge questions and answers, query energy system data, predict the operating parameters of the corresponding energy business system (such as power generation, power consumption, etc.), or obtain the operation plan of the energy business system, so as to control the operation of the energy business system according to the operation plan of the energy business system. Specifically, when the user session intention is identified as an energy knowledge question and answer intention or an energy system data query intention, the corresponding energy business system can be accessed by calling the energy business system interface to obtain the energy knowledge question and answer results or the energy system data query results; when the user session intention is identified as an intention to predict the operating parameters of the corresponding energy business system or to obtain the operating plan of the energy business system, the corresponding energy business system interface can be called and the corresponding energy business system interface can be used to obtain the corresponding session results (such as predicting the operating parameters of the corresponding energy business system, obtaining the operating plan of the energy business system, etc.).

[0075] As can be seen from the above, the embodiment of the present application divides the comprehensive energy intelligent conversation into two core stages: intent recognition and parameter extraction. Each stage has its own corresponding prompt words, and the intentions contained in the intent recognition prompt words in the comprehensive energy scenario used in the intent recognition stage can be generated in advance by the general team according to the energy business scenario requirements provided by the business parties of each energy business system in the comprehensive energy scenario. The parameter extraction prompt words in the parameter extraction stage are independently developed and generated by the business parties of the corresponding energy business systems in the comprehensive energy scenario. That is, for the generation of prompt words, a general-divided architecture is adopted (that is, the intent prompt words in the comprehensive energy scenario are developed and generated by the general team, and the parameter extraction prompt words are independently developed and generated by each business party) to achieve the engineering design of the prompt words used for the intelligent conversation in the comprehensive energy scenario, without the need to concentrate the prompt words for all business requirements together and have multiple teams write the same set of prompt words together, thereby reducing the redundancy and confusion of the prompt words, so as to improve the efficiency of the intelligent conversation, and avoid affecting the quality of the intelligent conversation due to the mutual influence between the prompt words. Through the above process, it can be achieved that when a new energy business is connected, the energy business scenario requirements provided by the new energy business party can be converted into corresponding intentions, and the converted intentions can be written into the intention recognition prompt words under the comprehensive energy scenario. The new business party can develop and deploy the corresponding energy business system interface and the parameter extraction prompt words corresponding to the parameters required by the energy business system interface according to the converted intentions, and save them in the intelligent conversation system without redesigning prompts and processes, thereby reducing development time delays and reducing development complexity.

[0076] Through the above process, for integrated energy intelligent conversations, the embodiment of the present application provides a standardized prompt word organization method, and specifically proposes a systematic method to organize and manage prompt words for different energy business scenarios under integrated energy scenarios to ensure clarity and consistency under the requirements of multiple energy business scenarios. This method can reduce the redundancy and confusion of prompt words and improve the response efficiency of the system. It also provides an efficient development collaboration mechanism, provides a friendly workflow for multi-team collaboration, reduces interference and errors caused by inconsistent prompt word design during multi-person collaboration, and improves overall development efficiency. It also provides a unified intelligent conversation management framework: a scalable conversation management framework is developed that can effectively integrate multiple energy business processes, ensure the smooth transmission and processing of information, and reduce the complexity of team collaboration.

[0077] The above-mentioned technical scheme disclosed in the embodiment of the present application not only splits the integrated energy intelligent conversation process into intent recognition and parameter extraction, with each stage corresponding to its own prompt words, but also the intent recognition prompt words used in the integrated energy scenario in the intent recognition stage can be developed and generated by a general team, and the energy business system interface corresponding to each intent in the intent recognition prompt words and the corresponding parameter extraction prompt words are independently developed by the business parties of the corresponding energy business systems in the integrated energy, without the need to bring together the prompt words for all energy business requirements and have multiple teams write the same set of prompt words together, thereby reducing the bloatedness, redundancy and confusion of the prompt words used in intelligent conversations in integrated energy scenarios, improving the efficiency and quality of intelligent conversations in integrated energy scenarios, and enabling the business parties of each energy business system to develop their own prompt words respectively, thereby improving development collaboration capabilities, reducing the complexity of team collaboration, and effectively integrating multiple business processes. Moreover, even if new energy services are connected, it is only necessary to determine the intent based on the energy business scenario requirements of the new energy services and write the intent into the intention recognition prompt words under the comprehensive energy scenario. The business party corresponding to the new energy services will generate parameter extraction requirement prompt words, etc. and store them in the intelligent conversation system without redesigning the entire process, reducing development time delays and complexity and improving development efficiency.

[0078] See also Figure 4 and Figure 5 ,in, Figure 4 A flowchart of another integrated energy intelligent conversation design method based on a large language model provided in an embodiment of the present application is shown. Figure 5 The actual business development and intelligent conversation flow chart applicable to the integrated energy scenario provided by the embodiment of the present application is shown, wherein: Figure 5 Taking the comprehensive energy including photovoltaic, energy storage and HVAC business scenarios as an example, the present application provides a method for realizing a conversation in a phased manner based on a large language model. After calling the energy business system interface according to the parameters required by the energy business system interface to obtain the conversation result, it can also include:

[0079] Obtaining a conversation result processing prompt word corresponding to the user conversation intention; the conversation result processing prompt word includes a processing method for the conversation result corresponding to the user conversation intention, and the conversation result processing prompt word is generated by the business party of the corresponding energy business system;

[0080] The conversation result and the conversation result processing prompt word are passed to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word to obtain the processing result.

[0081] In an embodiment of the present application, the business parties of each energy business system in the comprehensive energy scenario can independently develop and deploy the energy business system interface corresponding to the intent, the parameters required by the energy business system interface, and generate the prompt words for parameter extraction requirements at the same time or after that, and can also set the processing method for the expected return result behind the corresponding intent, that is, the processing method for the session result corresponding to the corresponding intent can be set, and the session result processing prompt words can be written and generated according to the processing method. In other words, the business parties corresponding to each intent can independently write and generate the session result processing prompt words corresponding to the corresponding intent, so as to reduce the complexity of team collaboration, improve the overall development efficiency, reduce the redundancy and bloat of the prompt words, and avoid the prompt words of all energy business systems being mixed together and affecting each other, reducing the confusion of the prompt words. Among them, the session result processing prompt words corresponding to the corresponding intent generated by each business party can be stored in the intelligent conversation system. For example, for a photovoltaic energy business system, the corresponding business party hopes to perform trend analysis on the photovoltaic power generation data after querying the photovoltaic power generation data and then output the analysis results. The conversation result processing prompt words generated by the corresponding business party corresponding to the photovoltaic power generation data query intention can include the photovoltaic power generation data trend analysis method, so that the large language model can perform corresponding processing on the corresponding conversation results according to the processing method contained in the conversation result processing prompt words.

[0082] On the basis of the above, after the intelligent conversation system calls the energy business system interface according to the energy business system interface to obtain the conversation result, it can determine the conversation result processing prompt word corresponding to the user's conversation intention, and can pass the conversation result and the corresponding conversation result processing prompt word to the large language model, for example, the conversation result and the corresponding conversation result processing prompt word (or the processing method in the conversation result processing prompt word) can be converted into prompt words and passed to the large language model. The large language model processes the conversation result according to the included processing method to obtain the corresponding processing result.

[0083] For example, see Figure 6, which shows a flowchart of calling the business system interface corresponding to the photovoltaic operation daily query intention and determining the corresponding session result processing prompt words provided by the embodiment of the present application. After extracting the two parameters of enterprise id and date required by the photovoltaic daily interface, the intelligent conversation system calls the photovoltaic daily interface according to these two parameters, returns the conversation result, and then determines the corresponding conversation result processing prompt words: according to business requirements, tell the large language model how to organize the copy and return it to the user, such as: the photovoltaic operation daily yesterday is: power generation 1000KWH, power generation efficiency 85%. Afterwards, the intelligent conversation system can send the conversation result and the corresponding conversation result processing prompt words to the large language model, so that the large language model organizes the copy in the conversation result according to the way of organizing the copy contained in the corresponding conversation result processing prompt words, and obtains the copy organization result.

[0084] From the above, it can be seen that the embodiment of the present application divides the integrated energy intelligent conversation into three core processes of intent recognition, parameter extraction and conversation result processing. Each process has its own corresponding prompt words, and the general team pre-generates the intent recognition prompt words in the integrated energy scenario, and the business parties of each energy business system in the integrated energy scenario respectively generate their corresponding energy business system interfaces, parameter extraction requirement prompt words and conversation result processing prompt words, that is, for the generation of prompt words used in the intelligent conversation in the integrated energy scenario, a general-division-division architecture is adopted (that is, the intent prompt words in the integrated energy scenario are developed and generated by the general team, and each parameter extraction prompt word is independently developed and generated by the business party of each energy business system in the integrated energy scenario, and each conversation result processing prompt word is also independently developed and generated by the business party of each energy business system in the integrated energy scenario) to realize the engineering design of the prompt words used for the intelligent conversation in the integrated energy scenario, without the need to concentrate the prompt words of all business requirements together and have multiple teams write the same set of prompt words together, thereby reducing the redundancy and confusion of the prompt words, so as to improve the efficiency of the intelligent conversation, and avoid affecting the quality of the intelligent conversation due to the mutual influence between the prompt words. Furthermore, through the above process, when a new energy business is connected, the energy business scenario requirements provided by the corresponding business party can be converted into corresponding intentions, and the converted intentions can be written into the intention recognition prompt words under the comprehensive energy scenario. The corresponding business party can develop and deploy the corresponding energy business system interface, parameter extraction prompt words corresponding to the parameters required by the energy business system interface, and conversation result processing prompt words according to the converted intentions, and save them in the intelligent conversation system without redesigning prompts and processes, thereby reducing development time delays and reducing development complexity.

[0085] From the above, it can be seen that the embodiment of the present application has designed a set of prompt word engineering solutions for comprehensive energy intelligent conversations based on multi-energy business scenarios. Through the three core processes of intent recognition, parameter extraction and conversation result processing, it supports the efficient access and integration of multiple energy business processes, improves query efficiency, development collaboration capabilities and system maintainability.

[0086] The embodiment of the present application provides a comprehensive energy intelligent conversation design method based on a large language model. After the conversation result and the conversation result processing prompt word are transmitted to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word, after obtaining the processing result, the method may also include:

[0087] Display the processing results.

[0088] In the embodiment of the present application, after the conversation result and the conversation result processing prompt are passed to the large language model, the large language model processes the conversation result according to the conversation result processing prompt, and after obtaining the processing result, the processing result can also be displayed so that the user can obtain the corresponding processing result in time. Of course, the processing result can also be displayed by the large language model.

[0089] The comprehensive energy intelligent conversation design method based on a large language model provided in an embodiment of the present application may further include, when the parameter extraction requirement prompt words are passed to the large language model:

[0090] Pass the conversation text to the big language model;

[0091] When the conversation result and the conversation result processing prompt words are transmitted to the large language model, the following may also be included:

[0092] Pass the conversation text to the big language model.

[0093] In the embodiment of the present application, in order to improve the quality of intelligent conversation, when the intelligent conversation system passes the parameter extraction requirement prompt words to the large language model, it can also pass the conversation text input by the user to the large language model, that is, the conversation text input by the user and the parameter extraction requirement prompt words can be passed to the large language model together. In addition, when the intelligent conversation system passes the conversation result and the conversation result processing prompt words to the large language model, it can also pass the conversation text input by the user to the large language model, that is, the conversation text input by the user, the conversation result and the conversation result processing prompt words can be passed to the large language model together, for example: the conversation text input by the user, the conversation result and the processing method of the conversation result can be converted into prompt words and passed to the large language model.

[0094] By passing the conversation text input by the user to the large language model not only in the intention recognition stage, but also in the parameter extraction stage and the conversation result processing stage, the large language model can continue the corresponding intelligent conversation process, thereby improving the quality of the intelligent conversation.

[0095] An embodiment of the present application provides a comprehensive energy intelligent conversation design method based on a large language model, wherein each intent corresponds to a parameter extraction module, and the parameter extraction module includes a business system interface corresponding to the intent and parameters required by the business system interface.

[0096] In an embodiment of the present application, each intent in the intent recognition prompt word under the comprehensive energy scenario uniquely corresponds to a parameter extraction module, that is, there is a one-to-one correspondence between the intent and the parameter extraction module, and each parameter extraction module contains the energy business system interface corresponding to the corresponding intent and the parameters required for the energy business system interface, so as to extract the parameters required for the energy business system interface based on the parameter extraction module. Specifically, after each business party develops and deploys the energy business system interface corresponding to the corresponding intent and the parameters required for the energy business system interface, it can form this information into a parameter extraction module and store it in the intelligent conversation system, so as to generate the parameter extraction requirement prompt word according to the corresponding parameter extraction module and extract the corresponding parameters based on the prompt word by the large language model, and facilitate the intelligent conversation system to determine the interface to be called based on the parameter extraction module.

[0097] The present application also provides a comprehensive energy intelligent conversation design device based on a large language model, see Figure 7 , which shows a schematic diagram of the structure of a comprehensive energy intelligent conversation design device based on a large language model provided in an embodiment of the present application, which may include:

[0098] The receiving module 71 is used to receive the conversation text input by the user, and pass the conversation text and the intention recognition prompt words in the integrated energy scenario to the large language model, and the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario;

[0099] The first determination module 72 is used to determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, and pass the parameter extraction requirement prompt words to the large language model, so that the large language model extracts the parameters required by the business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one by one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one by one, and the energy business system interface and the corresponding parameter extraction requirement prompt words are generated by the business party of the corresponding energy business system;

[0100] The calling module 73 is used to call the energy business system interface according to the parameters required by the energy business system interface to obtain the session result.

[0101] The embodiment of the present application provides a comprehensive energy intelligent conversation design device based on a large language model, which may also include:

[0102] The acquisition module is used to obtain the session result processing prompt words corresponding to the user's session intention after calling the energy business system interface according to the parameters required by the energy business system interface to obtain the session result; the session result processing prompt words include the processing method of the session result corresponding to the user's session intention, and the session result processing prompt words are generated by the business party of the corresponding energy business system;

[0103] The first transmission module is used to transmit the conversation result and the conversation result processing prompt word to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word to obtain a processing result.

[0104] The embodiment of the present application provides a comprehensive energy intelligent conversation design device based on a large language model, which may also include:

[0105] The display module is used to pass the conversation result and the conversation result processing prompt word to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word, and after obtaining the processing result, the processing result is displayed.

[0106] The embodiment of the present application provides a comprehensive energy intelligent conversation design device based on a large language model, which may also include:

[0107] A second transfer module is used to transfer the conversation text to the large language model when transferring the parameter extraction requirement prompt word to the large language model;

[0108] The third transmission module is used to transmit the conversation text to the large language model when transmitting the conversation result and the conversation result processing prompt word to the large language model.

[0109] An embodiment of the present application provides a comprehensive energy intelligent conversation design device based on a large language model, and each intent in the intent recognition prompt word can include an intent name and an intent description.

[0110] In an integrated energy intelligent conversation design device based on a large language model provided in an embodiment of the present application, each intention in the intention recognition prompt word may also include a priority;

[0111] The comprehensive energy intelligent conversation design device based on the large language model may also include:

[0112] The second determining module is used to determine the user session intention with the highest priority as the final user session intention.

[0113] An embodiment of the present application provides an integrated energy intelligent conversation design device based on a large language model, wherein each intent corresponds to a parameter extraction module, and the parameter extraction module includes a business system interface corresponding to the intent and parameters required by the business system interface.

[0114] The present application also provides a comprehensive energy intelligent conversation design device based on a large language model, see Figure 8 , which shows a structural schematic diagram of a comprehensive energy intelligent conversation design device based on a large language model provided in an embodiment of the present application, which may include:

[0115] A memory 81, used for storing computer programs;

[0116] The processor 82, when used to execute the computer program stored in the memory 81, can implement the following steps:

[0117] Receive the conversation text input by the user, pass the conversation text and the intention recognition prompt words in the comprehensive energy scenario to the large language model, and the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the comprehensive energy scenario; determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, pass the parameter extraction requirement prompt words to the large language model, and the large language model extracts the parameters required by the energy business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one-to-one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one-to-one, and the energy business system interface and the corresponding parameter extraction requirement prompt words are generated by the corresponding business party; call the business system interface according to the parameters required by the business system interface to obtain the conversation result.

[0118] The present application also provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the following steps can be implemented:

[0119] Receive the conversation text input by the user, pass the conversation text and the intention recognition prompt words in the comprehensive energy scenario to the large language model, and the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the comprehensive energy scenario; determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, pass the parameter extraction requirement prompt words to the large language model, and the large language model extracts the parameters required by the energy business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one-to-one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one-to-one, and the energy business system interface and the corresponding parameter extraction requirement prompt words are generated by the corresponding business party; call the business system interface according to the parameters required by the business system interface to obtain the conversation result.

[0120] For the description of the relevant parts of a comprehensive energy intelligent conversation design device, equipment and readable storage medium based on a large language model provided in an embodiment of the present application, please refer to the detailed description of the corresponding parts of an intelligent conversation method provided in an embodiment of the present application, which will not be repeated here.

[0121] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0122] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0123] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0124] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0125] In this application, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0126] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A comprehensive energy intelligent conversation design method based on a large language model, characterized in that: include: Receive a conversation text input by a user, pass the conversation text and the intention recognition prompt words in the integrated energy scenario to a large language model, and the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario; Determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, pass the parameter extraction requirement prompt words to the large language model, and the large language model extracts the parameters required by the energy business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one-to-one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one-to-one, and the energy business system interface and the corresponding parameter extraction requirement prompt words are generated by the business party of the corresponding energy business system; The energy business system interface is called according to the parameters required by the energy business system interface to obtain a session result.

2. The method for designing integrated energy intelligent conversation based on a large language model according to claim 1 is characterized in that: After calling the energy business system interface according to the parameters required by the energy business system interface to obtain a session result, the method further includes: Acquire a session result processing prompt word corresponding to the user session intention; the session result processing prompt word includes a processing method for the session result corresponding to the user session intention, and the session result processing prompt word is generated by a business party of the corresponding energy business system; The conversation result and the conversation result processing prompt word are passed to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word to obtain a processing result.

3. The method for designing integrated energy intelligent conversation based on a large language model according to claim 2 is characterized in that: After the conversation result and the conversation result processing prompt word are transmitted to the large language model, and the large language model processes the conversation result according to the conversation result processing prompt word to obtain the processing result, the method further includes: The processing result is displayed.

4. The method for designing integrated energy intelligent conversation based on a large language model according to claim 2 is characterized in that: When the parameter extraction requirement prompt word is passed to the large language model, it also includes: Passing the conversation text to the large language model; When the conversation result and the conversation result processing prompt word are transmitted to the large language model, the method further includes: The conversation text is passed to the large language model.

5. The method for designing integrated energy intelligent conversation based on a large language model according to claim 1 is characterized in that: Each of the intentions in the intention recognition prompt word includes an intention name and an intention description.

6. The method for designing integrated energy intelligent conversation based on a large language model according to claim 5 is characterized in that: Each of the intentions in the intention recognition prompt word also includes a priority; When the large language model identifies multiple user conversation intentions, the method further includes: The user session intent with the highest priority is determined as the final user session intent.

7. The method for designing integrated energy intelligent conversation based on a large language model according to claim 1, characterized in that: Each of the intents corresponds to a parameter extraction module, and the parameter extraction module includes the business system interface corresponding to the intent and the parameters required by the business system interface.

8. A comprehensive energy intelligent conversation design device based on a large language model, characterized in that: include: A receiving module is used to receive a conversation text input by a user, and pass the conversation text and the intention recognition prompt words in the integrated energy scenario to the large language model, so that the large language model recognizes the user conversation intention according to the intention recognition prompt words; the intention recognition prompt words contain multiple intentions, and each intention is determined according to the energy business scenario requirements provided by the business parties of each energy business system in the integrated energy scenario; A first determination module is used to determine the corresponding energy business system interface and the parameter extraction requirement prompt words corresponding to the parameters required by the energy business system interface according to the user conversation intention, and pass the parameter extraction requirement prompt words to the large language model, so that the large language model extracts the parameters required by the business system interface according to the parameter extraction requirement prompt words; each intention in the intention recognition prompt words corresponds to the energy business system interface one-to-one, and the energy business system interface and the parameter extraction requirement prompt words corresponding to the required parameters correspond one-to-one, and the business system interface and the corresponding parameter extraction requirement prompt words are generated by the corresponding business party; The calling module is used to call the energy business system interface according to the parameters required by the energy business system interface to obtain a session result.

9. A comprehensive energy intelligent conversation design device based on a large language model, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the method for designing integrated energy intelligent conversation based on a large language model as described in any one of claims 1 to 7 when executing the computer program.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the comprehensive energy intelligent conversation design method based on a large language model as described in any one of claims 1 to 7 are implemented.

Images