A method and device for executing intention based on digital twin

Abstract

The present invention provides a method and device for executing intention based on digital twins, including: responding to input instructions; identifying language information in the input instructions based on a pre-trained target language model; classifying the input instructions according to the language information to determine the classification result; determining the target intention instruction according to the classification result and the language information; according to the target intention instruction, causing the data processing subsystem to execute and determine the execution result according to the target intention instruction; causing the digital twin model to display the execution result. The present invention enables the target language model to accurately identify the intention of the input instruction, and accurately interact with the digital twin model according to the intention, helping the digital twin system to be more intelligent, improving the interaction effect between the digital twin and artificial intelligence, and improving the operating performance and efficiency of the digital twin system.

Description

Technical Field

[0001] The present application relates to the field of digital twin technology, and more specifically, to a method and device for executing intentions based on digital twins. Background Art

[0002] Digital twins are a rapidly developing field that helps companies improve product quality, efficiency and reduce costs by building virtual entities in a digital environment. Artificial intelligence, a state-of-the-art natural language processing technology, has broad and far-reaching application prospects in this field. With the development of artificial intelligence technology, digital twins will have a wider range of application scenarios and more efficient solutions.

[0003] At present, digital twins and artificial intelligence can be simply integrated and simple dialogue and question-answering can be carried out in the scene. However, in the face of various input instructions, the instructions may include logical actions for the digital twin model, or questions and answers for the digital twin model, including multiple scenes. The current technology cannot effectively identify the intention of the instruction and then form an accurate execution action, resulting in poor interaction between digital twins and artificial intelligence. It also requires the development of different interaction logics in the face of different digital twin scene requirements, resulting in increased costs. Summary of the invention

[0004] The present invention provides an intention execution method and device based on digital twins, which accurately identify the intention of input instructions through a pre-trained language model, and accurately interact with the digital twin model based on the intention, thereby helping the digital twin system to become more intelligent, improving the interaction effect between digital twins and artificial intelligence, and improving the operating performance and efficiency of the digital twin system.

[0005] In a first aspect, the present invention provides an intention execution method based on digital twins, the method being applied to a digital twin system, the digital twin system including a digital twin model, a target language model and a data processing subsystem, the method comprising:

[0006] Responding to an input instruction; recognizing language information in the input instruction based on the pre-trained target language model;

[0007] classifying the input instruction according to the language information to determine a classification result;

[0008] Determine a target intention instruction according to the classification result and the language information;

[0009] According to the target intention instruction, the data processing subsystem is caused to execute according to the target intention instruction and determine the execution result;

[0010] Enable the digital twin model to display the execution result.

[0011] Preferably, the classifying the input instruction according to the language information to determine the classification result comprises:

[0012] Performing semantic decomposition on the language information to extract a set of candidate words in the language information;

[0013] Using the target language model to perform a clustering operation on the candidate word set to determine a candidate keyword set;

[0014] Extracting the keyword from the candidate keyword set using a maximum margin correlation algorithm;

[0015] Inputting the keyword into a pre-trained keyword question classifier to output a classification label to which the keyword belongs;

[0016] Determine the classification result to which the input instruction belongs according to the classification label;

[0017] The classification result includes: question and answer category and / or command category.

[0018] Preferably, determining the target intention instruction according to the classification result and the language information includes:

[0019] If the classification result is the question-and-answer category, determining that the target intention instruction is a question-and-answer instruction according to the keyword and the target language model;

[0020] If the classification result is the command category, the target intended instruction is determined to be a command instruction based on the keyword and the target language model.

[0021] Preferably, the step of causing the data processing subsystem to execute and determine the execution result according to the target intention instruction comprises:

[0022] If the target intention instruction is the question-and-answer instruction, causing the data processing subsystem to match the target answer in the digital twin model according to the question-and-answer instruction;

[0023] The execution result is determined according to the target answer.

[0024] Preferably, the step of causing the data processing subsystem to execute and determine the execution result according to the target intention instruction comprises:

[0025] If the target intention instruction is the command instruction, causing the data processing subsystem to act according to the command instruction to determine the action behavior in the digital twin model;

[0026] The execution result is determined according to the action behavior.

[0027] Preferably, before identifying the language information in the input instruction based on the pre-trained target language model, the method further includes:

[0028] Determining scene information related to the digital twin model;

[0029] The original language model is iteratively trained using the scene information as a training sample to determine the target language model.

[0030] Preferably, before identifying the language information in the input instruction based on the pre-trained target language model, the method further includes:

[0031] Extracting data information of the actual scene corresponding to the digital twin model;

[0032] Update the data information into the digital twin model.

[0033] In a second aspect, the present invention provides an intention execution device based on digital twins, comprising:

[0034] A language information recognition module, used to respond to input instructions; recognize language information in the input instructions based on the pre-trained target language model;

[0035] A classification result determination module, used for classifying the input instruction according to the language information to determine a classification result;

[0036] A target intention instruction determination module, used to determine the target intention instruction according to the classification result and the language information;

[0037] An execution result determination module, used to make the data processing subsystem execute and determine the execution result according to the target intention instruction according to the target intention instruction;

[0038] The execution result display module is used to enable the digital twin model to display the execution result.

[0039] In a third aspect, the present invention provides a readable medium, comprising execution instructions. When a processor of an electronic device executes the execution instructions, the electronic device executes any method described in the first aspect.

[0040] In a fourth aspect, the present invention provides an electronic device, comprising a processor and a memory storing execution instructions, wherein when the processor executes the execution instructions stored in the memory, the processor executes any method described in the first aspect.

[0041] The present invention provides a method and device for executing intention based on digital twins, which responds to input instructions; identifies language information in the input instructions based on a pre-trained target language model; classifies the input instructions according to the language information to determine the classification result; determines the target intention instruction according to the classification result and the language information; according to the target intention instruction, enables the data processing subsystem to execute and determine the execution result according to the target intention instruction; and enables the digital twin model to display the execution result. The present invention enables the target language model to accurately identify the intention of the input instruction, and accurately interacts with the digital twin model according to the intention, helping the digital twin system to be more intelligent, improving the interaction effect between the digital twin and artificial intelligence, and improving the operating performance and efficiency of the digital twin system.

[0042] The further effects of the above-mentioned non-conventional preferred manner will be described below in conjunction with specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the existing technical solutions, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A schematic diagram of an intention execution method based on digital twins provided in one embodiment of the present invention;

[0045] Figure 2 A schematic diagram of another intention execution method based on digital twins provided in one embodiment of the present invention;

[0046] Figure 3 A schematic diagram of an intention execution device based on digital twins provided in one embodiment of the present invention;

[0047] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] Today, as human society is highly developing digital technology, digital twins have become a popular technology application in more and more industries. Digital twins can help humans digitally replicate objects, systems, processes, etc. in the real physical world to achieve the goal of integrating virtual and real things and improve business efficiency and operational value in different industries. It is a concept that combines the physical world with the digital world.

[0050] It simulates and reflects physical entities or processes by creating a real-time, interactive digital model. Digital twins can be used in various fields, including manufacturing, energy, transportation, and medical care. By combining with technologies such as artificial intelligence, big data analysis, and the Internet of Things, they can realize the intelligence and optimization of the spatial environment.

[0051] Digital twins are a rapidly developing field that helps companies improve product quality, efficiency and reduce costs by building virtual entities in a digital environment. Artificial intelligence, a state-of-the-art natural language processing technology, has broad and far-reaching application prospects in this field. With the development of artificial intelligence technology, digital twins will have a wider range of application scenarios and more efficient solutions.

[0052] At present, digital twins and artificial intelligence can be simply integrated and simple dialogue and question-answering can be carried out in the scene. However, in the face of various input instructions, the instructions may include logical actions for the digital twin model, or questions and answers for the digital twin model, including multiple scenes. The current technology cannot effectively identify the intention of the instruction and then form an accurate execution action, resulting in poor interaction between digital twins and artificial intelligence. It also requires the development of different interaction logics in the face of different digital twin scene requirements, resulting in increased costs.

[0053] In view of this, the present invention provides an intention execution method based on digital twins. Figure 1 As shown, a specific embodiment of a method for executing an intention based on a digital twin provided by the present invention is shown. In this embodiment, the method is applied to a digital twin system, which includes a digital twin model, a target language model, and a data processing subsystem. The method includes:

[0054] Step 101, responding to an input instruction; recognizing language information in the input instruction based on a pre-trained target language model;

[0055] Specifically, the digital twin model in this embodiment is a digital model that is completely consistent with the real-world entity in a virtual space to achieve simulation, monitoring, analysis and control of its entire life cycle process. By using advanced technologies such as big data, cloud computing, and the Internet of Things, a virtual model reflecting the complete life cycle of physical equipment is created by collecting various data during the operation of the equipment, from design to manufacturing, operation, maintenance, and scrapping. The digital twin model can map various scenes in the real world. For example: community service scenes, property service scenes, traffic operation scenes, etc. Before recognizing the language information in the input instruction based on the pre-trained target language model, it also includes: extracting data information of the actual scene corresponding to the digital twin model, and updating the data information to the digital twin model. Among them, the data information can be the state data of the physical system corresponding to the digital twin model, the simulation data related to the model, etc. Assuming that the mapping scene of the digital twin model is a community, the data information can be the current number of people in the community, the number of community facilities, the status of community facilities, etc. Because the state of the actual scene changes at any time, a real-time digital twin model can be obtained by updating the data information, so that the subsequent execution results of the intention are more accurate.

[0056] The target language model refers to a deep learning model trained with a large amount of text data, which can generate natural language text or understand the meaning of language text. The target language model in this embodiment can handle a variety of natural language tasks, such as text classification, question and answer, dialogue, etc., and can be trained in advance using the elements in the corresponding scene of the above-mentioned digital twin model to make it better understand the meaning of the language text related to the digital twin model. The target language model in this embodiment can be FastGpt or Dify. Before recognizing the language information in the input instruction based on the pre-trained target language model, it also includes: determining the scene information related to the digital twin model, and iteratively training the original language model with the scene information as a training sample to determine the target language model. The scene information can be text information related to the physical system corresponding to the digital twin model. For example, assuming that the scene corresponding to the digital twin model is a community service, the scene information can be a statistical community population, then the corresponding text information can be "Statistics of the number of people in the xx community", "How many people are there in the xx community", "How many people are there in the xx community", etc. Of course, the scene information is not limited to statistics, and all the information related to the scene of the digital twin model is included. Because the scene information is closely related to the actual physical system, iterative training of the original model using the scene information as training samples can improve the positive feedback effect of the training samples, so that the final target language model can accurately understand the intention of the input instructions. After continuous iteration and updating of model parameters until the original language model converges and stops training, the target language model obtained can accurately identify instructions related to the digital twin model.

[0057] The data processing subsystem may include a console, several data processing units, and several execution units. The data processing unit may be used to transmit data between the digital twin model and the target language model. The console is used to manage and monitor the digital twin system. The execution unit may be used to execute console commands, retrieve or allocate data between the digital twin model and the target language model, so that the digital twin model or the target language model may perform related actions according to the commands.

[0058] The input instructions in this embodiment may be natural language input instructions or interface input instructions. The input instructions may be input through the above-mentioned data processing subsystem. The input instructions may be instructions related to the digital twin system. The language information in the input instructions is identified based on a pre-trained target language model. The language information may be a command or a question and answer. Because the target language model has been pre-trained, it can interpret and process the language information in the input instructions.

[0059] Step 102: classify the input instructions according to the language information to determine the classification result;

[0060] Specifically, the language information is semantically decomposed to extract a set of candidate words in the language information, the target language model is used to perform a clustering operation on the candidate word set to determine a set of candidate keywords, the maximum margin correlation algorithm is used to extract keywords from the candidate keyword set, the keywords are input into a pre-trained keyword question classifier to output the classification label to which the keywords belong, and the classification result to which the input instruction belongs is determined according to the classification label, and the classification results include: question and answer category and / or command category.

[0061] For example, if the input command is "focus the camera on the electric vehicle charging equipment in this community", the target language model can be used to decompose the language information corresponding to the input command, and the command can be decomposed into a set of several nouns and verbs, that is, a set of candidate words, such as "lens, focus, community, electric vehicle, charging equipment, charging", etc. The maximum boundary correlation algorithm is used to extract keywords from the candidate keyword set, which can be "focus, electric vehicle charging equipment". The target language model can have a built-in question classifier, and by comparing the similarity between keywords and classification labels, the classification result to which the input command belongs can be determined. Through classification, the command can be determined as a command category.

[0062] Similarly, assuming that the input command is "How many buildings are there in this community", by extracting the keywords "how many, building", and comparing the similarity between the keywords and the classification labels, the classification result to which the input command belongs can be determined. Through classification, it can be determined that the command belongs to the question and answer category.

[0063] Step 103: Determine the target intention instruction according to the classification result and the language information;

[0064] If the classification result is a question-and-answer category, the target intent instruction is determined to be a question-and-answer instruction based on the keywords and the target language model. Among them, the keywords can be extracted through the above steps, and the target language model is pre-trained. Through the recognition of keywords by the target language model, the target intent instruction is obtained as a question-and-answer instruction, that is, the intention of the input instruction in the system is: to obtain relevant answers to the questions asked.

[0065] Step 104: If the target intention instruction is a question-and-answer instruction, the data processing subsystem is made to match the target answer in the digital twin model according to the question-and-answer instruction, and the execution result is determined according to the target answer;

[0066] If the target intent instruction is a question-and-answer instruction, the data processing subsystem matches the target answer in the digital twin model according to the question-and-answer instruction, and determines the execution result according to the target answer. Specifically, through the accurate recognition of the input instruction by the above-mentioned target language model, the final target intent instruction is obtained as a question-and-answer instruction, that is, the user wants to get relevant answers through the digital twin system, for example, want to know the number of people in this community, want to know the current traffic conditions somewhere, etc. This requires the data processing subsystem to access the knowledge related to the question-and-answer instruction in the digital twin model according to the question-and-answer instruction, so as to extract the target answer. The execution result corresponding to the question-and-answer instruction is the answer to the question.

[0067] Step 105: Enable the digital twin model to display the execution results.

[0068] It can be seen from the above steps that the execution result is the answer to the question and answer instruction. The execution result is the matching result obtained for the user's intention. The execution result needs to be fed back to the user, and the relevant data of the execution result can be displayed in the digital twin model through the data processing subsystem. Specifically, the digital twin model can provide a user interface so that the user can interact with the digital twin model through a command line interface, a graphical interface, or voice, so that the execution result is displayed on the digital twin model. For example, if the input command is "How many buildings are there in this community", then there can be a scene dialog box in the digital twin model. The execution result is received in the scene dialog box and can be displayed in text or voice, "There are 20 buildings in this community. Is there anything else I can help you with?".

[0069] It can be seen from the above technical solutions that the beneficial effects of this embodiment are: accurately identifying the intention of the input command through the target language model, and accurately interacting with the digital twin model based on the intention, helping the digital twin system to respond to user questions and answers more accurately and quickly, improving the interaction effect between the digital twin and artificial intelligence, and improving the operating performance and efficiency of the digital twin system.

[0070] Figure 1 What is shown is only a basic embodiment of the method of the present invention. By performing certain optimization and expansion on this basis, other preferred embodiments of the method can be obtained.

[0071] like Figure 2 As shown, another specific embodiment of the method for executing intention based on digital twins of the present invention is shown. This embodiment is further described on the basis of the above-mentioned embodiment. In this embodiment, the method includes the following steps:

[0072] Step 201, responding to an input instruction; recognizing language information in the input instruction based on a pre-trained target language model;

[0073] Specifically, the digital twin model in this embodiment is a technology that simulates, monitors, analyzes and controls the entire life cycle of a real-world entity by building a digital model that is completely consistent with the real-world entity in a virtual space. By using advanced technologies such as big data, cloud computing, and the Internet of Things, a virtual model that reflects the complete life cycle of a physical device is created from design to manufacturing, operation, maintenance, and scrapping by collecting various data during the operation of the device. The digital twin model can map various scenes in the real world. For example: community service scenes, property service scenes, traffic operation scenes, etc. Before recognizing the language information in the input instruction based on the pre-trained target language model, it also includes: extracting data information of the actual scene corresponding to the digital twin model, and updating the data information to the digital twin model. Among them, the data information can be the state data of the physical system corresponding to the digital twin model, the simulation data related to the model, etc. Assuming that the mapping scene of the digital twin model is a community, the data information can be the current number of people in the community, the number of community facilities, the status of community facilities, etc. Because the state of the actual scene changes at any time, a real-time digital twin model can be obtained by updating the data information to make the subsequent execution results of the intention more accurate.

[0074] The target language model refers to a deep learning model trained with a large amount of text data, which can generate natural language text or understand the meaning of language text. The target language model in this embodiment can handle a variety of natural language tasks, such as text classification, question and answer, dialogue, etc., and can be trained in advance using the elements in the corresponding scene of the above-mentioned digital twin model to make it better understand the meaning of the language text related to the digital twin model. The target language model in this embodiment can be FastGpt or Dify. Before recognizing the language information in the input instruction based on the pre-trained target language model, it also includes: determining the scene information related to the digital twin model, and iteratively training the original language model with the scene information as a training sample to determine the target language model. The scene information can be text information related to the physical system corresponding to the digital twin model. For example, assuming that the scene corresponding to the digital twin model is a community service, the scene information can be a statistical community population, then the corresponding text information can be "Statistics of the number of people in the xx community", "How many people are there in the xx community", "How many people are there in the xx community", etc. Of course, the scene information is not limited to statistics, and all the information related to the scene of the digital twin model is included. Because the scene information is closely related to the actual physical system, iterative training of the original model using the scene information as training samples can improve the positive feedback effect of the training samples, so that the final target language model can accurately understand the intention of the input instructions. After continuous iteration and updating of model parameters until the original language model converges and stops training, the target language model obtained can accurately identify instructions related to the digital twin model.

[0075] The data processing subsystem may include a console, several data processing units, and several execution units. The data processing unit may be used to transmit data between the digital twin model and the target language model. The console is used to manage and monitor the digital twin system. The execution unit may be used to execute console commands, retrieve or allocate data between the digital twin model and the target language model, so that the digital twin model or the target language model may perform related actions according to the commands.

[0076] The input instructions in this embodiment may be natural language input instructions or interface input instructions. The input instructions may be input through the above-mentioned data processing subsystem. The input instructions may be instructions related to the digital twin system. The language information in the input instructions is identified based on a pre-trained target language model. The language information may be a command or a question and answer. Because the target language model has been pre-trained, it can interpret and process the language information in the input instructions.

[0077] Step 202: classify the input instructions according to the language information to determine the classification result;

[0078] Specifically, the language information is semantically decomposed to extract a set of candidate words in the language information, the target language model is used to perform a clustering operation on the candidate word set to determine a set of candidate keywords, the maximum margin correlation algorithm is used to extract keywords from the candidate keyword set, the keywords are input into a pre-trained keyword question classifier to output the classification label to which the keywords belong, and the classification result to which the input instruction belongs is determined according to the classification label, and the classification results include: question and answer category and / or command category.

[0079] For example, if the input command is "focus the camera on the electric vehicle charging equipment in this community", the target language model can be used to decompose the language information corresponding to the input command, and the command can be decomposed into a set of several nouns and verbs, that is, a set of candidate words, such as "lens, focus, community, electric vehicle, charging equipment, charging", etc. The maximum boundary correlation algorithm is used to extract keywords from the candidate keyword set, which can be "focus, electric vehicle charging equipment". The target language model can have a built-in question classifier, and by comparing the similarity between keywords and classification labels, the classification result to which the input command belongs can be determined. Through classification, the command can be determined as a command category.

[0080] Similarly, assuming that the input command is "How many buildings are there in this community", by extracting the keywords "how many, building", and comparing the similarity between the keywords and the classification labels, the classification result to which the input command belongs can be determined. Through classification, it can be determined that the command belongs to the question and answer category.

[0081] Step 203: Determine the target intention instruction according to the classification result and the language information;

[0082] If the classification result is a command category, the target intention instruction is determined to be a command instruction based on the keywords and the target language model. Among them, the keywords can be extracted through the above steps, and the target language model is pre-trained. Through the recognition of keywords by the target language model, the target intention instruction is obtained as a command instruction, that is, the intention of the input instruction in the system is: for the input instruction, a related operation needs to be obtained.

[0083] Step 204: If the target intention instruction is a command instruction, the data processing subsystem is made to act according to the command instruction to determine the action behavior in the digital twin model, and the execution result is determined according to the action behavior.

[0084] If the target intention instruction is a command instruction, the data processing subsystem is made to act according to the command instruction to determine the action behavior in the digital twin model, and the execution result is determined according to the action behavior. Specifically, through the accurate recognition of the input instruction by the above-mentioned target language model, the final target intention instruction is obtained as a command instruction, that is, the user wants to make the digital twin system obtain relevant actions corresponding to the instruction through the input instruction, for example, want to focus on a specific area in the digital twin model, want to rotate the digital twin model by a certain angle, etc. This requires the data processing subsystem to specifically change the relevant parameters in the digital twin model according to the command instruction, so that the digital twin model produces action behavior, such as zooming in, partial zooming in, zooming out, rotating, etc. Therefore, the execution result corresponding to the command instruction is the operation generated for the instruction.

[0085] Step 205: Enable the digital twin model to display the execution results.

[0086] From the above steps, it can be seen that the execution result is the operation generated for the instruction. The execution result is the matching result obtained for the user's intention. The execution result needs to be fed back to the user, and the execution result can be displayed in the digital twin model through the data processing subsystem. Specifically, the physical data in the digital twin model can be regulated by the data processing subsystem so that the digital twin model acts according to the execution result. For example, if the input instruction is "help me focus on the device with model xx", then the parameters of the device with model xx in the digital twin model can be changed through the data processing subsystem, and the device can be partially enlarged in the digital twin model, so as to complete the focus and form a display of the execution result.

[0087] It can be seen from the above technical solutions that the beneficial effects of this embodiment are: accurately identifying the intention of the input command through the target language model, and accurately interacting with the digital twin model based on the intention, helping the digital twin system to respond to user commands more accurately and quickly, improving the interaction effect between the digital twin and artificial intelligence, and improving the operating performance and efficiency of the digital twin system.

[0088] like Figure 3 As shown in FIG. 1 , a specific embodiment of an intention execution device based on digital twins of the present invention is shown. The device of this embodiment is used to execute Figures 1-2 The physical device of the method. Its technical solution is essentially consistent with the above embodiment, and the corresponding description in the above embodiment is also applicable to this embodiment. The device in this embodiment includes:

[0089] The language information recognition module 301 is configured to respond to an input instruction and recognize language information in the input instruction based on a pre-trained target language model;

[0090] A classification result determination module 302 is configured to classify the input instruction according to the language information to determine a classification result;

[0091] A target intention instruction determination module 303 is configured to determine a target intention instruction according to the classification result and the language information;

[0092] The execution result determination module 304 is configured to enable the data processing subsystem to execute and determine the execution result according to the target intention instruction according to the target intention instruction;

[0093] The execution result display module 305 is configured to enable the digital twin model to display the execution result.

[0094] Figure 4 : is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. Among them, the memory may include a memory, such as a high-speed random access memory (Random-Access Memory, RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage, etc. Of course, the electronic device may also include hardware required for other services.

[0095] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0096] The memory is used to store execution instructions. Specifically, the execution instructions are computer programs that can be executed. The memory may include internal memory and non-volatile memory, and provides execution instructions and data to the processor.

[0097] In one possible implementation, the processor reads the corresponding execution instructions from the non-volatile memory into the memory and then runs them, and can also obtain the corresponding execution instructions from other devices to form an intention execution device based on digital twins at the logical level. The processor executes the execution instructions stored in the memory to implement an intention execution method based on digital twins provided in any embodiment of the present invention through the executed execution instructions.

[0098] As described above, the present invention Figure 3 The method for executing an intention execution device based on digital twins provided in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or instructions in the form of software. The above processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0099] The steps of the method disclosed in the embodiment of the present invention can be directly embodied as being executed by a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0100] The embodiment of the present invention further proposes a readable storage medium, which stores an execution instruction. When the stored execution instruction is executed by a processor of an electronic device, the electronic device can execute an intention execution method based on digital twins provided in any embodiment of the present invention, and is specifically used to execute the following Figure 1 , Figure 2 The method shown.

[0101] The electronic device described in the above embodiments may be a computer.

[0102] Those skilled in the art will appreciate that the embodiments of the present invention may be provided as methods or computer program products. Therefore, the present invention may be implemented in the form of a complete hardware embodiment, a complete software embodiment, or a combination of software and hardware.

[0103] The various embodiments of the present invention are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0104] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0105] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A digital twin-based intention execution method, characterized in that: The method is applied to a digital twin system, wherein the digital twin system includes a digital twin model, a target language model, and a data processing subsystem, and the method includes: Responding to an input instruction; recognizing language information in the input instruction based on the pre-trained target language model; classifying the input instruction according to the language information to determine a classification result; Determine a target intention instruction according to the classification result and the language information; According to the target intention instruction, the data processing subsystem is caused to execute according to the target intention instruction and determine the execution result; Enable the digital twin model to display the execution result.

2. The method according to claim 1, characterized in that The classifying the input instruction according to the language information to determine the classification result comprises: Performing semantic decomposition on the language information to extract a set of candidate words in the language information; Using the target language model to perform a clustering operation on the candidate word set to determine a candidate keyword set; Extracting the keyword from the candidate keyword set using a maximum margin correlation algorithm; Inputting the keyword into a pre-trained keyword question classifier to output a classification label to which the keyword belongs; Determine the classification result to which the input instruction belongs according to the classification label; The classification result includes: question and answer category and / or command category.

3. The method according to claim 2, characterized in that Determining the target intention instruction according to the classification result and the language information includes: If the classification result is the question-and-answer category, determining that the target intention instruction is a question-and-answer instruction according to the keyword and the target language model; If the classification result is the command category, the target intended instruction is determined to be a command instruction based on the keyword and the target language model.

4. The method according to claim 3, characterized in that The step of causing the data processing subsystem to execute and determine the execution result according to the target intention instruction comprises: If the target intention instruction is the question-and-answer instruction, causing the data processing subsystem to match the target answer in the digital twin model according to the question-and-answer instruction; The execution result is determined according to the target answer.

5. The method according to claim 3, characterized in that: The step of causing the data processing subsystem to execute and determine the execution result according to the target intention instruction comprises: If the target intention instruction is the command instruction, causing the data processing subsystem to act according to the command instruction to determine the action behavior in the digital twin model; The execution result is determined according to the action behavior.

6. The method according to any one of claims 1 to 5, characterized in that: Before identifying the language information in the input instruction based on the pre-trained target language model, the method further includes: Determining scene information related to the digital twin model; The original language model is iteratively trained using the scene information as a training sample to determine the target language model.

7. The method according to any one of claims 1 to 5, characterized in that: Before identifying the language information in the input instruction based on the pre-trained target language model, the method further includes: Extracting data information of the actual scene corresponding to the digital twin model; Update the data information into the digital twin model.

8. An intention execution device based on digital twins, characterized in that: include: A language information recognition module, used to respond to input instructions; Recognizing language information in the input instruction based on the pre-trained target language model; A classification result determination module, used for classifying the input instruction according to the language information to determine a classification result; A target intention instruction determination module, used to determine the target intention instruction according to the classification result and the language information; An execution result determination module, used to make the data processing subsystem execute and determine the execution result according to the target intention instruction according to the target intention instruction; The execution result display module is used to enable the digital twin model to display the execution result.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method according to any one of claims 1 to 7 when executed.

10. An electronic device, comprising: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1 to 7.

Images