Scheme generation method and device based on large model and electronic equipment

Abstract

The invention provides a scheme generation method and device based on a large model, electronic equipment and a readable storage medium, and relates to the technical field of artificial intelligence such as natural language processing, deep learning and large models. The scheme generation method based on the large model comprises the steps of obtaining first scheme description information; acquiring multiple pieces of second scheme description information from a target scheme library according to the first scheme description information; using a large model to generate parameter adjustment information according to the first scheme description information and the plurality of second scheme description information; and obtaining a target scheme according to the parameter adjustment information. The generation efficiency and the generation accuracy of the target scheme can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, and particularly to artificial intelligence technologies such as natural language processing, deep learning, and large models. A method, apparatus, electronic device, and readable storage medium for generating a solution based on a large model are provided. Background Art

[0002] During the production process, system adjustment process, or fault recovery process, it is often necessary to generate corresponding solutions to adjust a large number of parameters existing in different processes, so as to achieve the purposes of improving production efficiency, reducing costs, accelerating the fault recovery speed, etc. Existing technologies usually generate solutions including parameter adjustment information based on manual experience, and there are problems such as low generation efficiency and low generation accuracy. Therefore, how to generate solutions faster and more accurately has become an urgent technical problem to be solved. Summary of the Invention

[0003] According to a first aspect of the present disclosure, a method for generating a solution based on a large model is provided, including: obtaining first solution description information; obtaining a plurality of second solution description information from a target solution library according to the first solution description information; using a large model to generate parameter adjustment information according to the first solution description information and the plurality of second solution description information; and obtaining a target solution according to the parameter adjustment information.

[0004] According to a second aspect of the present disclosure, a device for generating a solution based on a large model is provided, including: an obtaining unit for obtaining first solution description information; a retrieval unit for obtaining a plurality of second solution description information from a target solution library according to the first solution description information; a processing unit for using a large model to generate parameter adjustment information according to the first solution description information and the plurality of second solution description information; and a generating unit for obtaining a target solution according to the parameter adjustment information.

[0005] According to a third aspect of the present disclosure, an electronic device is provided, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0006] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method as described above.

[0007] According to a fifth aspect of the present disclosure, a computer program product is provided, including a computer program, and the computer program implements the method as described above when executed by a processor.

[0008] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The drawings are used to better understand the present solution and do not constitute a limitation to the present disclosure. Among them:

[0010] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure;

[0011] Figure 2 is a schematic diagram according to the second embodiment of the present disclosure;

[0012] Figure 3 is a schematic diagram according to the third embodiment of the present disclosure;

[0013] Figure 4 is a schematic diagram according to the fourth embodiment of the present disclosure;

[0014] Figure 5 is a block diagram of an electronic device for implementing the large model-based solution generation method of the embodiments of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, descriptions of well-known functions and mechanisms are omitted for clarity and conciseness.

[0016] Figure 1 is a schematic diagram according to the first embodiment of the present disclosure. As Figure 1 shown, the large model-based solution generation method of this embodiment specifically includes the following steps:

[0017] S101. Obtain first solution description information;

[0018] S102. According to the first solution description information, obtain multiple second solution description information from the target solution library;

[0019] S103. Use a large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information;

[0020] S104. Obtain a target solution according to the parameter adjustment information.

[0021] The method for generating a solution based on a large model in this embodiment utilizes the text generation ability of the large model and the target solution library, generates parameter adjustment information according to the obtained first solution description information, and then obtains the target solution corresponding to the first solution description information based on the parameter adjustment information. Since only the first solution description information needs to be obtained during the generation process of the target solution, this embodiment can improve the generation efficiency of the target solution. Moreover, by using the large model to generate the parameter adjustment information, the generation accuracy of the target solution can also be improved.

[0022] In this embodiment, the large model can be a large language model (LLM) specifically used for performing natural language processing tasks, or a multi-modal large model.

[0023] When this embodiment executes S101, the content input at the input end can be used as the first solution description information, that is, the manually input content can be used as the first solution description information; or the corresponding environmental information, working condition information, production target information, equipment information, etc. can be obtained as the first solution description information.

[0024] The target solution generated in this embodiment can be a production solution (such as a weaving production solution), a fault recovery solution (such as a power supply fault recovery solution), a system adjustment solution (such as a refrigeration system adjustment solution), etc. That is, this embodiment can generate target solutions corresponding to different scenarios according to the first solution description information corresponding to different scenarios.

[0025] For example, this embodiment can generate a weaving production solution according to the first solution description information corresponding to the weaving production scenario (such as including weaving production target information, fabric material information, loom model information, etc.).

[0026] This embodiment can also generate a power supply fault recovery solution according to the first solution description information corresponding to the power supply fault recovery scenario (such as including working condition information such as the number of main transformers, busbars, downstream feeders, and power-loss users in a substation).

[0027] This embodiment can further generate a refrigeration system adjustment solution according to the first solution description information corresponding to the refrigeration system adjustment scenario (such as including environmental information such as the current temperature and current humidity, and model information of each refrigeration device).

[0028] That is to say, the first solution description information obtained in this embodiment is information used to describe the environment, working conditions, production targets, etc. corresponding to the target solution to be generated.

[0029] After this embodiment executes S101 to obtain the first solution description information, it executes S102 to obtain multiple second solution description information from the target solution library according to the obtained first solution description information.

[0030] In this embodiment, the target solution library includes multiple historical solution description information and the corresponding second embedding vectors of each historical solution description information; further, the target solution library in this embodiment may also include the historical parameter adjustment information corresponding to each historical solution description information.

[0031] When this embodiment executes S102 to obtain multiple second solution description information from the target solution library according to the first solution description information, the implementation method that can be adopted is: obtain the first embedding vector corresponding to the first solution description information; calculate the vector similarity between the first embedding vector and multiple second embedding vectors in the target solution library respectively; obtain the historical solution description information corresponding to the second embedding vector whose vector similarity meets the first preset requirement from the target solution library as the second solution description information.

[0032] That is to say, this embodiment uses the vector retrieval method to obtain multiple second solution description information similar to the first solution description information from the target solution library, which can improve the accuracy of the obtained second solution description information, and further improve the accuracy of the parameter adjustment information generated based on the second solution description information.

[0033] Among them, when this embodiment executes S102 to obtain the historical solution description information corresponding to the second embedding vector whose vector similarity meets the first preset requirement, the second embedding vectors with the top N vector similarities can be used as the second embedding vectors whose vector similarities meet the first preset requirement, and N is a positive integer greater than or equal to 2.

[0034] Under normal circumstances, different historical solution description information included in the target solution library corresponds to the same solution generation scenario, that is, the solution generation method based on the large model provided in this embodiment can be applied only to the fault recovery scenario, or only for the production scenario, or only for the system adjustment scenario, so there is no need to select the target solution library.

[0035] In addition, in order to expand the applicable scope of the solution generation method based on the large model provided in this embodiment, this embodiment can also set multiple candidate solution libraries, and different candidate solution libraries correspond to different solution generation scenarios.

[0036] Therefore, when this embodiment executes S102, it may also include the following content: obtain the scenario information corresponding to the first solution description information, and the scenario information is one of the fault recovery scenario, production scenario, system adjustment scenario, etc.; use the candidate solution library corresponding to the obtained scenario information as the target solution library.

[0037] That is to say, in this embodiment, the target scenario library can also be selected from multiple candidate scenario libraries according to the scenario information corresponding to the first scenario description information, avoiding unnecessary calculation of the vector similarity between description information (for example, calculating the vector similarity between the first scenario description information corresponding to the fault recovery scenario and the historical scenario description information corresponding to the production scenario), which can improve the acquisition efficiency and accuracy of the second scenario description information.

[0038] It can be understood that the historical parameter adjustment information corresponding to the historical scenario description information included in the target scenario library can be the initial parameter adjustment information corresponding to the historical scenario description information, or the updated parameter adjustment information corresponding to the historical scenario description information.

[0039] For example, if the scenario generation scenario corresponding to the target scenario library is a fault recovery scenario, the target scenario library includes different historical scenario description information corresponding to different fault types. For example, the historical scenarios corresponding to fault type A include Scenario 1 and Scenario 2. The historical scenario description information corresponding to Scenario 1 is the first description information, and the historical scenario description information corresponding to Scenario 2 is the second description information. The historical scenarios corresponding to fault type B include Scenario 3 and Scenario 4. The historical scenario description information corresponding to Scenario 3 is the third description information, and the historical scenario description information corresponding to Scenario 4 is the fourth description information.

[0040] For fault type A, if the evaluation score of Scenario 1 is higher than that of Scenario 2, then in this embodiment, the initial parameter adjustment information corresponding to the first description information is used as the historical parameter adjustment information corresponding to both the first description information and the second description information. For fault type B, if the evaluation score of Scenario 4 is higher than that of Scenario 3, then in this embodiment, the initial parameter adjustment information corresponding to the fourth description information is used as the historical parameter adjustment information corresponding to both the third description information and the fourth description information.

[0041] That is to say, the historical parameter adjustment information stored in the target scenario library in this embodiment is the optimal parameter adjustment information corresponding to different fault types, different production types, or different adjustment types.

[0042] After this embodiment executes S102 to obtain multiple second scenario description information from the target scenario library, it executes S103 to use a large model to generate parameter adjustment information according to the first scenario description information and the multiple second scenario description information.

[0043] In this embodiment, the generated parameter adjustment information is data for specifically adjusting relevant equipment or technical parameters during the fault recovery process, system adjustment process, or production process.

[0044] In this embodiment, when using a large model to generate parameter adjustment information according to the first solution description information and multiple second solution description information in S103, the implementation method that can be adopted is as follows: concatenate the first solution description information and multiple second solution description information into a prompt text (i.e., prompt); input the prompt text into the large model, and obtain the parameter adjustment information according to the output result of the large model.

[0045] Specifically, when the large model in this embodiment generates parameter adjustment information according to the concatenated prompt text, first, the large model judges the similarity degree between the first solution description information and multiple second solution description information according to the prompt text, then selects a second solution description information that is closest to the first solution description information from multiple second solution description information, and finally generates parameter adjustment information according to the selected second solution description information.

[0046] It can be understood that when the large model in this embodiment generates parameter adjustment information according to the second solution description information, it can call the target solution library to use the historical parameter adjustment information corresponding to the selected second solution description information in the target solution library as the generated parameter adjustment information; it can also directly generate parameter adjustment information according to the selected second solution description information.

[0047] That is to say, after completing the retrieval of the second solution description information in this embodiment, retrieval enhancement is performed based on the large model, so that all semantic information of the text in the solution description information is fully utilized in the process of generating parameter adjustment information, and the accuracy of the generated parameter adjustment information can be improved.

[0048] In addition, when concatenating the first solution description information and multiple second solution description information into a prompt text in S103 in this embodiment, the following content can also be included: obtain the judgment priority information of the solution description information; concatenate the first solution description information, multiple second solution description information and judgment priority information into a prompt text.

[0049] That is to say, in this embodiment, the flexibility in generating parameter adjustment information can also be improved by adjusting the prompt text input into the large model so that the priority information for judging the similarity degree between solution description information is included in the prompt text.

[0050] For example, if the description information of the first solution is "There are 2 main transformers, 4 busbars, 3 outgoing feeders that can be automatically switched to off-site feeders by the lower-level feeders, 4 direct supply users, and 0 incoming lines for important users" in the faulty substation, and if the description information 1 of the second solution is "There are 3 main transformers, 3 busbars, 3 outgoing feeders that can be automatically switched to off-site feeders by the lower-level feeders, 0 direct supply users, and 0 incoming lines for important users", and if the description information 2 of the second solution is "There are 3 main transformers, 4 busbars, 3 outgoing feeders that can be automatically switched to off-site feeders by the lower-level feeders, 0 direct supply users, and 1 incoming line for important users".

[0051] When this embodiment executes S103, in addition to splicing the above information into the prompt text, it can also add "When judging whether the description information of the solutions is similar, give priority to whether the number of main transformers and the number of incoming lines for important users are the same" to the spliced prompt text, so that when the large model makes a judgment, it selects the description information 1 of the second solution as the solution description information that is more similar to the description information of the first solution.

[0052] In addition, when this embodiment executes S103, the large model can also output the thinking process during the generation of the parameter adjustment information, such as which information is used for comparison and which information is emphasized during the comparison process, so as to improve the credibility of the generated parameter adjustment information.

[0053] In different scenario of generating solutions, the parameter adjustment information generated by the large model will also be different.

[0054] For example, in the scenario of power supply fault recovery, the parameter adjustment information generated by the large model can include whether to allow operating non-remote control switches (allow parameter is true, disallow parameter is false), the weight parameter of the restored users, the line overload threshold parameter, and whether to allow busbar power restoration (allow parameter is true, disallow parameter is false); in the scenario of weaving production, the parameter adjustment information generated by the large model can include the steam output parameter, the ambient temperature parameter, etc.; in the scenario of refrigeration system adjustment, the parameter adjustment information generated by the large model can include the refrigeration temperature parameter, the water pump frequency parameter, etc.

[0055] After this embodiment executes S103 to generate the parameter adjustment information, it executes S104 to obtain the target solution according to the generated parameter adjustment information.

[0056] When this embodiment executes S104, in addition to using the parameter adjustment information, it can further use the description information of the first solution to obtain the target solution, that is, the target solution includes both the description information of the first solution and the parameter adjustment information.

[0057] After obtaining the target solution in this embodiment, the parameter adjustment information in the target solution can be used to perform fault recovery, system adjustment, or production. For example, the target solution can be sent to a faulty power plant for power supply fault recovery, or the target solution can be sent to a refrigeration system for refrigeration adjustment, or the target solution can be sent to a weaving system for weaving, etc.

[0058] Figure 2 is a schematic diagram according to the second embodiment of the present disclosure. As Figure 2 shown, after obtaining the target solution in this embodiment, the following content may further be included:

[0059] S201. Simulate the target solution to obtain a solution simulation result;

[0060] S202. In response to determining that the solution simulation result meets the second preset requirement, store the target solution in the target solution library.

[0061] That is to say, after obtaining the target solution in this embodiment, it is also possible to determine whether the obtained target solution is stored in the target solution library by means of solution simulation, so as to improve the richness of the information included in the target solution library and make the target solution library more and more perfect.

[0062] When executing S201 in this embodiment, a target simulator can be used to simulate the target solution, so as to obtain a solution simulation result according to the output result of the target simulator; the solution simulation process in this embodiment is to simulate fault recovery, system adjustment, production process, etc. according to the parameter adjustment information in the target solution.

[0063] When executing S201 in this embodiment to simulate the target solution and obtain a solution simulation result, it is also possible to first obtain the scenario information corresponding to the target solution (for example, obtain it according to the first solution description information in the target solution), then determine the simulator corresponding to the scenario information as the target simulator, and finally use the determined target simulator to simulate the target solution, and obtain a solution simulation result according to the output result of the target simulator.

[0064] In this embodiment, simulators corresponding to different scenario information can be preset, such as a simulator corresponding to a fault recovery scenario, a simulator corresponding to a production scenario, a simulator corresponding to a system adjustment scenario, etc., so as to select a target simulator according to the scenario information corresponding to the target solution and improve the accuracy of the obtained solution simulation result.

[0065] When it is determined in S202 that the simulation result of the solution meets the second preset requirement in this embodiment, the evaluation index of the simulation result of the solution can be obtained first, and then when it is determined that the obtained evaluation index is improved compared with the preset evaluation index, it is determined that the simulation result of the solution meets the second preset requirement.

[0066] That is to say, this embodiment determines whether to store the simulation result of the solution in the target solution library according to the evaluation index of the simulation result of the solution, which can improve the quality of the target solution stored in the target solution library.

[0067] It can be understood that if the target solution only includes parameter adjustment information, when this embodiment executes S202, it is also necessary to further obtain the first solution description information, and then store the first solution description information and the parameter adjustment information together in the target solution library.

[0068] Figure 3 It is a schematic diagram according to the third embodiment of the present disclosure. Figure 3 The flowchart of the solution generation method based on the large model is shown: S301, obtain the first solution description information. The first solution description information obtained in this embodiment can be the current working condition information of the faulty substation corresponding to the power supply fault recovery scenario; S302, according to the first solution description information, obtain multiple second solution description information from the target solution library. The second solution description information obtained in this embodiment can be the historical working condition information of the faulty substation corresponding to the power supply fault recovery scenario; S303, use the large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information; S304, obtain the target solution according to the parameter adjustment information. In this embodiment, the first solution description information can also be further obtained to obtain the target solution; S305, simulate the target solution to obtain the simulation result of the solution; S306, in response to determining that the simulation result of the solution meets the second preset requirement, store the target solution in the target solution library. In this embodiment, the embedding vector corresponding to the first solution description information can also be further obtained, and then the embedding vector and the target solution are stored in the target solution library together; wherein, the first solution description information stored in the target solution library is the historical solution description information in the subsequent use process, the parameter adjustment information is the historical parameter adjustment information in the subsequent use process, and the embedding vector is the second embedding vector in the subsequent use process.

[0069] Figure 4 It is a schematic diagram according to the fourth embodiment of the present disclosure. As Figure 4 shown, the solution generation device 400 based on the large model in this embodiment includes:

[0070] The acquisition unit 401 is used to acquire the first solution description information;

[0071] The retrieval unit 402 is configured to obtain multiple second solution description information from the target solution library according to the first solution description information;

[0072] The processing unit 403 is configured to use a large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information;

[0073] The generation unit 404 is configured to obtain a target solution according to the parameter adjustment information.

[0074] In this embodiment, the large model can be a large language model (LLM) dedicated to performing natural language processing tasks, or a multi-modal large model.

[0075] The acquisition unit 401 can use the content input at the input end as the first solution description information, that is, use the manually input content as the first solution description information; or it can obtain corresponding environment information, working condition information, production target information, equipment information, etc. as the first solution description information.

[0076] The target solution generated in this embodiment can be a production solution (such as a weaving production solution), a fault recovery solution (such as a power supply fault recovery solution), a system adjustment solution (such as a refrigeration system adjustment solution), etc. That is, this embodiment can generate target solutions corresponding to different scenarios according to the first solution description information corresponding to different scenarios.

[0077] That is to say, the first solution description information obtained by the acquisition unit 401 is information used to describe the environment, working conditions, production targets, etc. corresponding to the target solution to be generated.

[0078] After the acquisition unit 401 obtains the first solution description information in this embodiment, the retrieval unit 402 obtains multiple second solution description information from the target solution library according to the obtained first solution description information.

[0079] In this embodiment, the target solution library includes multiple historical solution description information and the corresponding second embedding vectors of each historical solution description information; further, the target solution library in this embodiment can also include the historical parameter adjustment information corresponding to each historical solution description information.

[0080] When the retrieval unit 402 obtains multiple second solution description information from the target solution library according to the first solution description information, the implementation method that can be adopted is as follows: obtain the first embedding vector corresponding to the first solution description information; calculate the vector similarity between the first embedding vector and multiple second embedding vectors in the target solution library respectively; obtain the historical solution description information corresponding to the second embedding vector whose vector similarity meets the first preset requirement from the target solution library as the second solution description information.

[0081] That is to say, in this embodiment, the vector retrieval method is used to obtain multiple second solution description information similar to the first solution description information from the target solution library, which can improve the accuracy of the obtained second solution description information, and further improve the accuracy of the parameter adjustment information generated based on the second solution description information.

[0082] Among them, when the retrieval unit 402 obtains the historical solution description information corresponding to the second embedding vector whose vector similarity meets the first preset requirement, the second embedding vectors whose vector similarity ranks among the top N can be used as the second embedding vectors whose vector similarity meets the first preset requirement, and N is a positive integer greater than or equal to 2.

[0083] Generally, different historical solution description information included in the target solution library corresponds to the same solution generation scenario, that is, the solution generation method based on the large model provided in this embodiment can be applied only to the fault recovery scenario, or only for the production scenario, or only for the system adjustment scenario, so there is no need to select the target solution library.

[0084] In addition, in order to expand the applicable scope of the solution generation method based on the large model provided in this embodiment, this embodiment can also set multiple candidate solution libraries, and different candidate solution libraries correspond to different solution generation scenarios.

[0085] Therefore, the retrieval unit 402 can also perform the following operations: obtain the scenario information corresponding to the first solution description information, and the scenario information is one of the fault recovery scenario, the production scenario, the system adjustment scenario, etc.; use the candidate solution library corresponding to the obtained scenario information as the target solution library.

[0086] That is to say, the retrieval unit 402 can also select the target solution library from multiple candidate solution libraries according to the scenario information corresponding to the first solution description information, avoiding unnecessary calculation of the vector similarity between description information (for example, calculating the vector similarity between the first solution description information corresponding to the fault recovery scenario and the historical solution description information corresponding to the production scenario), which can improve the acquisition efficiency and accuracy of the second solution description information.

[0087] It can be understood that the historical parameter adjustment information corresponding to the historical solution description information included in the target solution library can be the initial parameter adjustment information corresponding to the historical solution description information, or the updated parameter adjustment information corresponding to the historical solution description information.

[0088] In this embodiment, after the retrieval unit 402 obtains multiple second solution description information from the target solution library, the processing unit 403 uses a large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information.

[0089] In this embodiment, the parameter adjustment information generated by the processing unit 303 is data for specifically adjusting relevant equipment or technical parameters during the fault recovery process, system adjustment process, or production process.

[0090] When the processing unit 403 uses a large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information, the implementation method that can be adopted is: splicing the first solution description information and the multiple second solution description information into a prompt text; inputting the prompt text into the large model, and obtaining parameter adjustment information according to the output result of the large model.

[0091] Specifically, when the large model in this embodiment generates parameter adjustment information according to the spliced prompt text, first, the large model judges the similarity degree between the first solution description information and the multiple second solution description information according to the prompt text, then selects a second solution description information that is closest to the first solution description information from the multiple second solution description information, and finally generates parameter adjustment information according to the selected second solution description information.

[0092] It can be understood that when the large model in this embodiment generates parameter adjustment information according to the second solution description information, it can call the target solution library to use the historical parameter adjustment information corresponding to the selected second solution description information in the target solution library as the generated parameter adjustment information; or directly generate parameter adjustment information according to the selected second solution description information.

[0093] That is to say, after the retrieval of the second solution description information is completed, since the processing unit 403 performs retrieval enhancement based on the large model, during the process of generating parameter adjustment information, the semantic information of all the words in the solution description information is fully utilized, which can improve the accuracy of the generated parameter adjustment information.

[0094] In addition, when the processing unit 403 splices the first solution description information and multiple second solution description information into a prompt text, it may further include the following: obtaining judgment priority information of the solution description information; splicing the first solution description information, multiple second solution description information, and the judgment priority information into a prompt text.

[0095] That is to say, the processing unit 403 can also adjust the prompt text input to the large model so that the prompt text includes priority information for judging the similarity between solution description information, thereby improving the flexibility in generating parameter adjustment information.

[0096] In this embodiment, after the processing unit 403 generates parameter adjustment information, the generating unit 404 obtains a target solution according to the generated parameter adjustment information.

[0097] In addition to using the parameter adjustment information, the generating unit 404 can further use the first solution description information to obtain the target solution, that is, the target solution includes both the first solution description information and the parameter adjustment information.

[0098] After the generating unit 404 obtains the target solution, it can perform fault recovery, system adjustment, or production according to the parameter adjustment information in the target solution.

[0099] The solution generation device 400 based on a large model in this embodiment may further include a simulation unit 405 for performing the following: after obtaining the target solution, simulating the target solution to obtain a solution simulation result; in response to determining that the solution simulation result meets the second preset requirement, storing the target solution in the target solution library.

[0100] That is to say, after obtaining the target solution, the simulation unit 405 can also determine whether to store the obtained target solution in the target solution library through solution simulation, thereby improving the richness of the information included in the target solution library and making the target solution library more and more perfect.

[0101] The simulation unit 405 can use a target simulator to simulate the target solution, and thus obtain a solution simulation result according to the output result of the target simulator; the solution simulation process in this embodiment is to perform simulations of fault recovery, system adjustment, production process, etc. according to the parameter adjustment information in the target solution.

[0102] When the simulation unit 405 simulates the target solution to obtain the solution simulation result, it can also first obtain the scenario information corresponding to the target solution (for example, obtain it according to the first solution description information in the target solution), then determine the simulator corresponding to the scenario information as the target simulator, and finally use the determined target simulator to simulate the target solution, and obtain the solution simulation result according to the output result of the target simulator.

[0103] In this embodiment, simulators corresponding to different scenario information can be preset, such as simulators corresponding to the fault recovery scenario, simulators corresponding to the production scenario, simulators corresponding to the system adjustment scenario, etc., so as to select the target simulator according to the scenario information corresponding to the target solution, and improve the accuracy of the obtained solution simulation result.

[0104] When the simulation unit 405 determines that the solution simulation result meets the second preset requirement, it can first obtain the evaluation index of the solution simulation result, and then determine that the solution simulation result meets the second preset requirement when it is determined that the obtained evaluation index is improved compared with the preset evaluation index.

[0105] That is to say, in this embodiment, it is determined whether to store the solution simulation result in the target solution library according to the evaluation index of the solution simulation result, which can improve the quality of the target solution stored in the target solution library.

[0106] It can be understood that if the target solution only includes parameter adjustment information, the simulation unit 405 also needs to further obtain the first solution description information, and then store the first solution description information and the parameter adjustment information into the target solution library together.

[0107] In the technical solution of the present disclosure, the acquisition, storage, and application of the user's personal information involved all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0108] According to the embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0109] As Figure 5 shown, it is a block diagram of an electronic device for the method of generating a solution based on a large model according to an embodiment of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are only examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0110] As Figure 5 shown, the device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0111] Multiple components in the device 500 are connected to the I / O interface 505, including: an input unit 506, such as a keyboard, a mouse, etc.; an output unit 507, such as various types of displays, speakers, etc.; a storage unit 508, such as a magnetic disk, an optical disc, etc.; and a communication unit 509, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0112] The computing unit 501 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 501 executes the various methods and processes described above, such as the method for generating a large model-based solution. For example, in some embodiments, the method for generating a large model-based solution can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 508.

[0113] In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the method for generating a large model-based solution described above can be executed. Alternatively, in other embodiments, the computing unit 501 can be configured to execute the method for generating a large model-based solution by any other appropriate means (e.g., by means of firmware).

[0114] The various embodiments of the systems and techniques described herein can be implemented in digital electronic circuitry, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that receives data and instructions from a storage system, at least one input device, and at least one output device, and transmits the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0115] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable large model-based solution generation devices, such that when the program code is executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, executed partially on the machine as an independent software package and partially on a remote machine, or executed entirely on a remote machine or server.

[0116] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0117] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for presenting information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0118] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0119] A computer system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The relationship between the client and the server is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services (“Virtual Private Server”, or simply “VPS”). The server can also be a server of a distributed system, or a server combined with a blockchain.

[0120] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitation is imposed herein.

[0121] The above specific embodiments do not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. A method for generating a solution based on a large model, comprising: Obtaining first solution description information; According to the first solution description information, obtaining multiple second solution description information from a target solution library; Using a large model, generating parameter adjustment information according to the first solution description information and the multiple second solution description information; Obtaining a target solution according to the parameter adjustment information.

2. The method according to claim 1, wherein, The obtaining multiple second solution description information from the target solution library according to the first solution description information includes: Obtaining a first embedding vector corresponding to the first solution description information; Calculating the vector similarity between the first embedding vector and multiple second embedding vectors in the target solution library respectively; Obtaining the historical solution description information corresponding to the second embedding vector whose vector similarity meets the first preset requirement from the target solution library as the second solution description information.

3. The method according to claim 1, further comprising Obtaining scenario information corresponding to the first solution description information; Using the candidate solution library corresponding to the scenario information as the target solution library.

4. The method according to claim 1, wherein The using a large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information includes: Concatenating the first solution description information and the multiple second solution description information into a prompt text; Inputting the prompt text into the large model, and obtaining the parameter adjustment information according to the output result of the large model.

5. The method according to claim 4, wherein The concatenating the first solution description information and the multiple second solution description information into a prompt text includes: Obtaining judgment priority information of the solution description information; Concatenating the first solution description information, the multiple second solution description information and the judgment priority information into the prompt text.

6. The method according to claim 1, further comprising After obtaining the target solution, simulating the target solution to obtain a solution simulation result; In response to determining that the solution simulation result meets the second preset requirement, storing the target solution in the target solution library.

7. The method according to claim 6, wherein The simulating the target solution to obtain a solution simulation result includes: Obtaining scenario information corresponding to the target solution; Determining a simulator corresponding to the scenario information as a target simulator; Using the target simulator to simulate the target solution, and obtaining the solution simulation result according to the output result of the target simulator.

8. A device for generating a solution based on a large model, comprising: An obtaining unit, configured to obtain first solution description information; A retrieval unit, configured to obtain multiple second solution description information from a target solution library according to the first solution description information; A processing unit, configured to use a large model to generate parameter adjustment information according to the first solution description information and the multiple second solution description information; A generating unit, configured to obtain a target solution according to the parameter adjustment information.

9. An electronic device, comprising: At least one processor; And A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method according to any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are for causing the computer to execute the method according to any one of claims 1-7.

11. A computer program product, comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-7.

Images