Home control method and home control scene learning method and system

Abstract

The invention discloses a home control method and a home control scene learning method and system. The method comprises the following steps: acquiring multi-modal data; and analyzing the multi-modal data by using a large language model, and generating a home control scene according to an analysis result. By means of the mode, automatic generation of the personalized scene is achieved, a user does not need to manually configure complex rules, the system can automatically learn and execute appropriate control strategies, user operation is simplified, and user experience is improved.

Description

Technical Field

[0001] The present application relates to the technical field, and in particular to a home control method and a home control scene learning method and system. Background Art

[0002] With the rapid development of smart home technology, users' demand for intelligent control of home appliances is growing. However, existing smart home systems are difficult to meet users' personalized needs. Summary of the invention

[0003] The home control method and home control scene learning method and system provided in this application can realize the automatic generation of personalized scenes. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience.

[0004] In a first aspect, the present application provides a method for learning a home control scenario, the method comprising: acquiring multimodal data; analyzing the multimodal data using a large language model, and generating a home control scenario based on the analysis results.

[0005] The generating of the home control scene according to the analysis result includes: generating the trigger conditions and control action sequence in the home control scene according to the analysis result.

[0006] Among them, after generating the home control scene according to the analysis results, it includes: obtaining user behavior data; the user behavior data is generated after the home control scene is executed; using a large language model to analyze the user behavior data and adjust the control action sequence in the home control scene.

[0007] Among them, adjusting the control action sequence in the home control scene includes: adding a control action corresponding to the user behavior data in the control action sequence.

[0008] Among them, adjusting the control action sequence in the home control scene includes: deleting the control action opposite to the user behavior data in the control action sequence.

[0009] The method further includes: generating a corresponding scene description based on the home control scene; wherein the scene description is displayed when the home control scene is triggered.

[0010] In a second aspect, the present application provides a home control method, which includes: in response to the current time satisfying a trigger condition, obtaining a corresponding home control scene; the home control scene is obtained through the method provided by the first aspect; and control instructions are sent to the corresponding home appliance device according to the home control scene, so that the intelligent body in the home appliance device works according to the control instructions.

[0011] Among them, control instructions are sent to corresponding home appliances according to the home control scenario so that the intelligent agents in the home appliances work according to the control instructions, including: in response to multiple home control scenarios simultaneously satisfying trigger conditions, determining the execution order according to the priorities of the multiple home control scenarios; according to the execution order, sending control instructions to the home appliances corresponding to the home control scenarios so that the intelligent agents in the home appliances work according to the control instructions.

[0012] In the third aspect, the present application provides a home control system, which includes: a data acquisition module, a scene learning module, a scene management module, a device control module and a user interaction module; the data acquisition module, the scene learning module, the scene management module, the device control module and the user interaction module cooperate with each other to implement the method provided in the first aspect or the second aspect.

[0013] In a fourth aspect, the present application provides a home control system, which includes: an acquisition module for acquiring multimodal data; a processing module for analyzing the multimodal data using a large language model, and generating a home control scene according to the analysis results.

[0014] The beneficial effects of the present application are as follows: Different from the prior art, the home control method and home control scene learning method and system provided by the present application use a large language model to analyze multimodal data, generate home control scenes according to the analysis results, and realize automatic generation of personalized scenes. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0016] Figure 1 It is a flowchart of an embodiment of a home control scene learning method provided by the present application;

[0017] Figure 2 is a flowchart of another embodiment of the home control scene learning method provided by the present application;

[0018] Figure 3 is a flowchart of another embodiment of the home control scene learning method provided by the present application;

[0019] Figure 4 is a flowchart of another embodiment of the home control scene learning method provided by the present application;

[0020] Figure 5 It is a flowchart of an embodiment of a home control method provided by the present application;

[0021] Figure 6 is a flow chart of another embodiment of the home control method provided by the present application;

[0022] Figure 7 It is a structural schematic diagram of an embodiment of a home control system provided by the present application;

[0023] Figure 8 This is a flow chart of an embodiment of a GPT large model scene learning module provided by the present application;

[0024] Fig. 9 It is a structural schematic diagram of an embodiment of an electronic device provided by the present application;

[0025] Fig.10 It is a structural diagram of another embodiment of the home control system provided by the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It will be understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for ease of description, only some but not all structures related to the present application are shown in the drawings. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0027] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0028] With the rapid development of smart home technology, users' demand for intelligent control of home appliances is growing. However, existing smart home systems are difficult to meet users' personalized needs.

[0029] Based on this, the present application proposes to use a large language model to analyze multimodal data, generate home control scenarios based on the analysis results, and realize automatic generation of personalized scenarios. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience. Please refer to any of the following embodiments for details.

[0030] See also Figure 1 , Figure 1 1 is a flow chart of an embodiment of a method for learning a home control scene provided by the present application, the method comprising:

[0031] Step 11: Obtain multimodal data.

[0032] Among them, the multimodal data includes at least text data, image data, voice data and sensor data; the sensor data is collected by the sensors corresponding to the home appliances.

[0033] In some embodiments, text data may be user input commands and / or historical interaction records, etc. Image data may be environmental images captured by cameras on home appliances, user gestures, etc. Voice data may be user voice commands and / or environmental sounds, etc. Sensor data may be various types of sensor data such as temperature, humidity, light, and energy consumption.

[0034] Step 12: Use the large language model to analyze the multimodal data and generate a home control scenario based on the analysis results.

[0035] In some embodiments, the large language model can use a pre-trained GPT model as the core and fine-tune it for the home appliance field. The main features include:

[0036] 1. Using the Transformer architecture, it has powerful sequence modeling capabilities.

[0037] 2. Conduct pre-training on large-scale household appliance corpus to master domain knowledge.

[0038] 3. Support multi-modal input and can process fused feature representation.

[0039] 4. Use autoregressive generation to achieve flexible text generation and conversation capabilities.

[0040] In some embodiments, the large language model is used to analyze the multimodal data to obtain user behavior data, and scene self-learning is performed based on the user behavior data to generate a home control scene and a scene description corresponding to the home control scene. That is, the large language model analyzes the multimodal data and outputs the home control scene and the scene description corresponding to the home control scene.

[0041] In some embodiments, the large language model can perform scenario pattern mining on user behavior data to obtain corresponding new scenarios, and then generate corresponding home control scenarios. And the scenario description corresponding to the scenario. The scenario description can be synchronously displayed to the user when the scenario is triggered so that the user knows the specific content of the upcoming scenario. The main content of the scenario description is to describe the specific home appliances controlled in the scenario and reminder suggestions for the user.

[0042] In some embodiments, after generating a home control scene, conditional triggering rules and / or corresponding intelligent agents are still constructed based on the home control scene.

[0043] The conditional trigger rule can be an IFTTT rule.

[0044] In some embodiments, the agent can be a scene agent, that is, each home control scene corresponds to a scene agent.

[0045] In some embodiments, the agent can be either a scene agent or a device agent. The device agent is deployed in a home appliance. The scene agent can be deployed in one of the home appliances in the scene.

[0046] In this embodiment, a large language model is used to analyze multimodal data, and a home control scenario is generated based on the analysis results to achieve automatic generation of personalized scenarios. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience.

[0047] See also Figure 2 , Figure 2 : is a flow chart of another embodiment of the home control scene learning method provided by the present application, the method comprising:

[0048] Step 21: Obtain multimodal data.

[0049] Step 22: Use the large language model to perform time series analysis on the multimodal data to obtain first user behavior data.

[0050] The first user behavior data includes at least one of a user behavior sequence and time information.

[0051] In some embodiments, the multimodal data may be sorted according to the time of generation, thereby obtaining the sequence of the user's behavior within the time period.

[0052] Step 23: Use the large language model to perform scene self-learning based on the first user behavior data to generate a home control scene.

[0053] In some embodiments, after the home control scene is generated, a scene description corresponding to the home control scene is also generated synchronously. The scene description is used to provide to the user so that the user knows the overall control action and / or method of the home control scene.

[0054] In this embodiment, a large language model is used to analyze multimodal data to obtain user behavior data. The large language model is used to perform scene self-learning based on user behavior data to generate home control scenes, thereby realizing automatic generation of personalized scenes. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience.

[0055] See also Figure 3 , Figure 3 : is a flow chart of another embodiment of the home control scene learning method provided by the present application, the method comprising:

[0056] Step 31: Obtain multimodal data.

[0057] Step 32: Analyze the multimodal data using a large language model, and generate trigger conditions and control action sequences in the home control scenario based on the analysis results.

[0058] In some embodiments, a scene description corresponding to the home control scene can also be obtained according to the home control scene, wherein the scene description is displayed when the home control scene is triggered.

[0059] In some embodiments, since the multimodal data has time information, the large language model can use time as a trigger condition when learning a scene. When the time reaches the scene trigger time, the corresponding home appliance is controlled to work according to the control action sequence of the home control scene.

[0060] In some embodiments, see Figure 4 , after generating the home control scene based on the analysis results, the following process can be performed:

[0061] Step 41: Acquire second user behavior data; the second user behavior data is generated after executing the home control scenario.

[0062] As user habits change, users may manually adjust part of the control logic after executing the home control scenario. Therefore, the control action sequence in the home control scenario can be adjusted by collecting new user behavior data.

[0063] Step 42: Analyze the second user behavior data using the large language model, and adjust the control action sequence in the home control scenario.

[0064] Considering the actual needs of users, the large language model can collect user behaviors after executing the home control scene, so as to decide whether to adjust the control action sequence in the home control scene according to the latest user behaviors. For example, add control actions in the home control scene. Or reduce control actions in the home control scene. Or, add control actions in the home control scene and reduce the original control actions in the home control scene.

[0065] In some embodiments, the second user behavior data is analyzed using a large language model to obtain a target control action. The target control action corresponding to the second user behavior data is added to the control action sequence. For example, if there is no target control action in the home control scene, and there is no control action opposite to the control action corresponding to the user behavior data, then the target control action corresponding to the second user behavior data is added to the control action sequence.

[0066] For example, the control actions corresponding to the home control scene include: turning on home appliance a, turning on home appliance b, and turning on home appliance c. The target control action is turning on home appliance d. At this time, if there is no control action corresponding to the home control scene that conflicts with the target control action, the target control action is added to the home control scene.

[0067] In some embodiments, the second user behavior data is analyzed using a large language model to obtain a target control action. The control action opposite to the second user behavior data is deleted from the control action sequence. For example, if there is a control action opposite to the target control action in the home control scene, the control action opposite to the second user behavior data is deleted from the control action sequence.

[0068] For example, the control actions corresponding to the home control scene include: turning on home appliance a, turning on home appliance b, and turning on home appliance c. The target control action is turning off home appliance a. At this time, turning on home appliance a and turning off home appliance a in the home control scene conflict (are opposite), so the control action: turning on home appliance a is deleted from the control action sequence.

[0069] In this embodiment, a large language model is used to analyze multimodal data to obtain user behavior data. The large language model is used to perform scene self-learning based on user behavior data to generate home control scenes, thereby realizing automatic generation of personalized scenes. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience.

[0070] Furthermore, it can automatically adjust the control strategy in the home control scenario to adapt to changes in user habits, thereby improving the intelligence and adaptability of the system.

[0071] See also Figure 5, Figure 5 1 is a flow chart of an embodiment of a home control method provided by the present application. The method includes:

[0072] Step 51: In response to the current time satisfying a trigger condition, a corresponding home control scene is obtained.

[0073] The home control scene in this embodiment can be learned through the embodiments of the present application involving home control scene learning.

[0074] After learning several home control scenarios through the scenario learning method of the present application, the home control system obtains the corresponding home control scenario when detecting that the current time meets the trigger condition.

[0075] In some embodiments, when a home control scene is triggered, a scene description corresponding to the home scene is obtained, and the scene description is displayed, such as by voice playback or by display.

[0076] Step 52: Send control instructions to corresponding home appliances according to the home control scenario, so that the intelligent agent in the home appliances works according to the control instructions.

[0077] Since each home control scenario involves control instructions for home appliances, the control instructions can be sent to the corresponding home appliances so that the intelligent agents in the home appliances work according to the control instructions.

[0078] In some embodiments, an intelligent agent is provided in the home appliance, which can control the home appliance according to control instructions in combination with environmental parameters, so that the actual operation of the home appliance is more in line with the environment and the scene comfort is improved.

[0079] See also Figure 6 , Figure 6 1 is a flow chart of another embodiment of the home control method provided by the present application. The method includes:

[0080] Step 61: In response to the current time satisfying a trigger condition, a corresponding home control scene is obtained.

[0081] Step 62: In response to multiple home control scenarios simultaneously satisfying the triggering conditions, determining the execution order according to the priorities of the multiple home control scenarios.

[0082] In some embodiments, after learning the home control scene, corresponding priorities are set for all learned home control scenes. Furthermore, when multiple home control scenes are triggered at the same time, the execution order can be determined according to the priorities to avoid abnormal triggering of the home control scene.

[0083] Step 63: According to the execution order, control instructions are sent to the home appliances corresponding to the home control scene, so that the intelligent agents in the home appliances work according to the control instructions.

[0084] In this embodiment, the execution order is determined according to the priority of the home control scene to avoid triggering anomalies in the home control scene.

[0085] In some embodiments, the home control system includes: a data acquisition module, a scene learning module, a scene management module, a device control module and a user interaction module; the data acquisition module, the scene learning module, the scene management module, the device control module and the user interaction module cooperate with each other to implement the method of any of the above embodiments.

[0086] During scene learning, the data acquisition module can obtain multimodal data through the user interaction module, or directly obtain multimodal data. The scene learning module uses a large language model to analyze the multimodal data and generates a home control scene according to the analysis results.

[0087] The scene management module is used to manage several home control scenes learned by the scene learning module.

[0088] When performing home control, the scene management module responds to the current time meeting the trigger condition, obtains the corresponding home control scene, and sends instructions to the device control module according to the home control scene. The device control module sends control instructions to the corresponding home appliance according to the instructions, so that the intelligent agent in the home appliance works according to the control instructions.

[0089] In some embodiments, the data acquisition module, the scene learning module, the scene management module, the device control module and the user interaction module can be set in different household appliances.

[0090] See also Figure 7 The home control system 70 mainly includes the following modules: a data acquisition module 71, a data preprocessing module 72, a GPT large model scene learning module 73, a scene management module 74, a device control module 75 and a user interaction module 76.

[0091] The data collection module 71 is responsible for collecting the user's multimodal interaction data, including voice commands, APP operations, sensor data, etc.

[0092] The data preprocessing module 72 performs preprocessing operations such as cleaning and format conversion on the collected data.

[0093] The GPT large model scenario learning module 73 uses the GPT large model to analyze the user's multimodal interaction data, and automatically learns and generates personalized scenarios.

[0094] The scene management module 74 manages scene information, including scene name, trigger conditions, control strategy, etc.

[0095] The device control module 75 sends control instructions to corresponding home appliances according to the scene information and control strategy.

[0096] The user interaction module 76 provides an interface for users to interact with the system, including voice interaction, APP control, etc.

[0097] In some embodiments, Figure 8 As shown, the GPT large model scene learning module process includes:

[0098] 1. Data representation: Convert the user’s multimodal interaction data into a format that the GPT large model can understand, such as converting voice commands into text and converting sensor data into time series.

[0099] 2. Scenario pattern mining: Use the GPT large model to analyze users’ multimodal interaction data and mine user behavior patterns. For example:

[0100] User behavior sequence: [turn on bedroom light, turn off air conditioner, open curtains, play news].

[0101] Time information: 7:00 am.

[0102] Repeat pattern: Every weekday.

[0103] 3. Scenario generation: Generate new scenarios based on the mined user behavior patterns. For example:

[0104] Scene name: "Waking up on a weekday".

[0105] Trigger conditions: Time == 7:00 AM && Date type == Weekday.

[0106] Action sequence: [Turn on bedroom light, turn off air conditioner, open curtains, play news].

[0107] 4. Scene description: Use the GPT large model to generate a natural language description of the scene, which is easy for users to understand and manage. For example:

[0108] The "Wake Up on Weekdays" scene will automatically execute at 7:00 a.m. every weekday, including turning on the bedroom lights, turning off the air conditioner, opening the curtains, and playing the news.

[0109] In some embodiments, the main functions of the scene management module include:

[0110] 1. Scene storage: The scene information generated by the GPT large model is stored in the database.

[0111] 2. Scene Editing: Allows users to view, modify and delete scenes.

[0112] 3. Scene trigger: trigger the scene at the appropriate time according to the trigger conditions of the scene.

[0113] 4. Scenario execution: Execute corresponding device control operations according to the action sequence defined by the scenario.

[0114] The core logic of the control strategy includes:

[0115] 1. Scenario priority management: When multiple scenarios meet the trigger conditions at the same time, the execution order is determined according to the priority.

[0116] 2. Device conflict resolution: When multiple scenes have conflicting control instructions for the same device, the final control action is determined based on preset rules or user preferences.

[0117] 3. Condition judgment and execution: When executing a scenario, dynamically adjust the execution strategy based on the current environment status and user status.

[0118] The adaptive optimization mechanism of the system mainly includes:

[0119] 1. User feedback collection: record the user's acceptance of the automatically executed scenarios, such as whether the system's control actions were manually modified.

[0120] 2. Scenario effect evaluation: Analyze the effects of scenario execution, including energy savings, user comfort and other indicators.

[0121] 3. Automatic parameter adjustment: Automatically adjust the trigger conditions and control strategies of the scene based on user feedback and scene effect evaluation results.

[0122] 4. New scenario recommendation: Based on the user’s usage patterns and preferences, recommend new scenarios that may be useful.

[0123] In some embodiments, for example, the self-learning process of the "getting up" scene is as follows:

[0124] 1. Data collection: The system collects the user's behavior data between 6:30-7:00 in the morning for several consecutive days: turn on the bedroom light, increase the air conditioning temperature, open the curtains, play light music.

[0125] 2. Scenario learning: The GPT model analyzes these data and generates a "getting up" scenario:

[0126] Trigger condition: time == 6:45 && weekday == True.

[0127] Action sequence: [Turn on the bedroom light, increase the air conditioning temperature, open the curtains, and play soft music].

[0128] 3. Scenario execution: The system automatically executes the scenario at 6:45 am on the next working day.

[0129] 4. Adaptive optimization: If the user manually turns off the music after the scene is executed for several consecutive days, the system will automatically adjust the scene and remove the "play light music" action.

[0130] In some embodiments, for example, the cross-device, cross-modal “movie watching” scene control is as follows:

[0131] 1. The user says via voice command: "I want to watch a movie."

[0132] 2. The system recognizes that this is a trigger command for the "movie watching" scene and performs the following actions:

[0133] Dim the lights in the living room, close the curtains, turn on the TV and stereo, and adjust the air conditioning temperature to a comfortable level.

[0134] 3. The user adjusts the volume through the mobile phone app. The system records this preference and automatically adjusts the volume to the user's preferred level the next time the "movie watching" scenario is executed.

[0135] 4. After the scene ends, the system detects through sensors that the user has left the living room and automatically turns off the equipment and restores the lighting.

[0136] See also Fig. 9 , Fig. 9 1 is a schematic diagram of the structure of an embodiment of an electronic device provided by the present application. The electronic device 90 includes a processor 91, a memory 92 and a communication interface 93 coupled to the processor 91, and the memory 92 stores at least one computer program, which is used to implement the method of any of the above embodiments when loaded and executed by the processor 91.

[0137] See also Fig.10 , Fig.10 1 is a schematic diagram of another embodiment of a home control system provided by the present application. The home control system 100 comprises: an acquisition module 101 , a processing module 102 and a learning module 103 .

[0138] The acquisition module 101 is used to acquire multimodal data.

[0139] The processing module 102 is used to analyze the multimodal data using a large language model and generate a home control scenario according to the analysis results.

[0140] In some embodiments, the acquisition module 101 and the processing module 102 cooperate to implement the method of any of the above embodiments.

[0141] To summarize, the home control method and home control scene learning method, system, and electronic device provided in the present application use a large language model to analyze multimodal data, generate home control scenes based on the analysis results, and realize automatic generation of personalized scenes. Users do not need to manually configure complex rules. The system can automatically learn and execute appropriate control strategies, simplify user operations, and improve user experience.

[0142] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only illustrative, for example, the division of the modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.

[0143] If the integrated units in the above other embodiments are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processing circuit component (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), disk or optical disk and other media that can store program code.

[0144] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for learning a home control scene, characterized in that: The method comprises: Acquire multimodal data; The multimodal data is analyzed using a large language model, and a home control scenario is generated according to the analysis result.

2. The method according to claim 1, characterized in that The generating of the home control scene according to the analysis result includes: The trigger conditions and control action sequences in the home control scenario are generated according to the analysis results.

3. The method according to claim 2, characterized in that After generating the home control scene according to the analysis result, the method includes: Acquiring user behavior data; the user behavior data is generated after executing the home control scenario; The user behavior data is analyzed using the large language model to adjust the control action sequence in the home control scenario.

4. The method according to claim 3, characterized in that: The adjusting the control action sequence in the home control scene includes: A control action corresponding to the user behavior data is newly added to the control action sequence.

5. The method according to claim 3, characterized in that: The adjusting the control action sequence in the home control scene includes: A control action that is opposite to the user behavior data is deleted from the control action sequence.

6. The method according to any one of claims 1 to 5, characterized in that: The method also includes: generating a corresponding scene description based on the home control scene; wherein the scene description is displayed when the home control scene is triggered.

7. A home control method, characterized in that: The method comprises: In response to the current time satisfying a trigger condition, obtaining a corresponding home control scene; the home control scene is obtained by the method according to any one of claims 1 to 6; According to the home control scenario, control instructions are sent to the corresponding home appliances, so that the intelligent agents in the home appliances work according to the control instructions.

8. The method according to claim 7, characterized in that The sending of control instructions to the corresponding home appliance according to the home control scenario so that the intelligent agent in the home appliance works according to the control instructions includes: In response to multiple home control scenarios simultaneously satisfying the trigger condition, determining an execution order according to the priorities of the multiple home control scenarios; According to the execution order, control instructions are sent to the home appliances corresponding to the home control scene, so that the intelligent agents in the home appliances work according to the control instructions.

9. A home control system, characterized in that: The home control system includes: a data acquisition module, a scene learning module, a scene management module, a device control module and a user interaction module; The data acquisition module, the scene learning module, the scene management module, the device control module and the user interaction module cooperate with each other to implement the method described in any one of claims 1-8.

10. A home control system, characterized in that: The home control system comprises: An acquisition module, used to acquire multimodal data; A processing module is used to analyze the multimodal data using a large language model and generate a home control scenario according to the analysis results.

Images