Home mode control method and device based on large model

Abstract

The invention discloses a home mode control method and device based on a large model, and relates to the technical field of smart home, and the method comprises the steps: obtaining the behavior information of a target person in a specific function region; based on the behavior information, determining a physical state and a behavior intention of the target person; and starting a home mode matched with the target person based on the physical state and the behavior intention of the target person. According to the method and the device provided by the embodiment of the invention, the appropriate home mode can be automatically matched according to the current sensing information, the intelligent degree of home mode management is improved, and the user experience is improved.

Description

Technical Field

[0001] This application relates to the field of smart home technology, and in particular to a home mode control method and device based on a large model. Background Technology

[0002] Smart devices are becoming increasingly common in daily life. Users can control smart devices through voice commands. However, voice control often only allows control of a specific smart device at a time. In a given scenario, multiple commands may be required to achieve a suitable home environment.

[0003] Currently, users can trigger automatic operations on smart devices by setting simple rules.

[0004] However, on the one hand, setting rules in advance requires users to have a certain level of knowledge, and the automatic mining of rules also requires a certain amount of data accumulation, resulting in a significant delay. Furthermore, the rules set by users and the rules automatically mined in the background cannot be perfectly matched, and the triggering of smart devices requires strict adherence to certain rules. On the other hand, some rules are very complex, requiring complex configurations from users, and the application also needs to provide support for these complex rules. For algorithmic mining, the more complex the combination of rules, the lower the probability of their implementation; therefore, data scarcity makes it impossible to mine these rules. Thirdly, user rules are configured by users based on their own understanding, and the automatic mining by algorithms also operates on a very small amount of data within the system itself; the system itself cannot utilize existing world knowledge. Summary of the Invention

[0005] This application provides a home mode control method and device based on a large model to overcome the shortcomings of the prior art. It automatically matches a suitable home mode based on the current perception information, thereby improving the intelligence level of home mode management and enhancing the user experience.

[0006] Firstly, this application provides a home mode control method based on a large model, including: Obtain behavioral information of target personnel within a specific functional area; Based on the behavioral information, the physical state and behavioral intentions of the target person are determined; Based on the target person's physical condition and behavioral intentions, activate the home mode that matches the target person.

[0007] According to the home mode control method based on a large model provided in this application, the step of obtaining behavioral information of target personnel within a specific functional area includes: Capture the actions of the target personnel within a predetermined timeframe in a specific functional area in real time; The actions of the target person within a predetermined time period are analyzed to obtain the behavioral information of the target person.

[0008] According to the home mode control method based on a large model provided in this application, the step of obtaining behavioral information of target personnel within a specific functional area includes: Acquire the interaction information between the target person and the intelligent interactive devices within a specific functional area; The interaction information is analyzed to obtain the behavioral information of the target personnel within the specific functional area.

[0009] According to the home mode control method based on a large model provided in this application, the step of activating a home mode matching the target person based on the target person's physical state and behavioral intentions includes: Based on the target person's physical state and behavioral intentions, respectively, obtain the target person's physical state parameters and behavioral intention parameters; The target person's physical state parameters and behavioral intention parameters are input into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model; wherein, the home pattern matching model is trained based on the sample physical state parameters, sample behavioral intention parameters and sample home patterns that match the sample behavioral information; based on the home pattern matching result, the home pattern matching the target person is activated.

[0010] According to this application, a home mode control method based on a large model is provided, wherein the home mode matching model includes a feature extraction layer, a weight allocation layer, and a home mode matching layer; Correspondingly, the step of inputting the target person's physical state parameters and behavioral intention parameters into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model includes: The body state parameters and behavioral intention parameters of the target person are input into the feature extraction layer to obtain the body state feature representation and behavioral intention representation of the target person output by the feature extraction layer. The target person's physical state feature representation and behavioral intention representation are input into the weight allocation layer to obtain the first weight value corresponding to the target person's physical state feature representation and the second weight value corresponding to the behavioral intention representation output by the weight allocation layer. The first weight value is used to characterize the importance of the physical state feature representation, and the second weight value is used to characterize the importance of the behavioral intention representation. The target person's physical state characteristics and behavioral intentions, along with the first weight value corresponding to the target person's physical state characteristics and the second weight value corresponding to the behavioral intentions, are input into the home pattern matching layer to obtain the home pattern matching result output by the home pattern matching layer.

[0011] According to this application, a home mode control method based on a large model is provided, wherein the home mode matching layer includes an initial mode matching layer and a deep mode matching layer; the initial mode matching layer is used to determine the historical home mode stored in the initial mode matching layer that matches the first feature representation of the input; the deep mode matching layer is used to correct the historical home mode according to the second feature representation of the target user; Correspondingly, the step of inputting the target person's physical state feature representation and behavioral intention representation, the first weight value corresponding to the target person's physical state feature representation, and the second weight value corresponding to the behavioral intention representation into the home pattern matching layer to obtain the home pattern matching result output by the home pattern matching layer specifically includes: The first feature representation and the second feature representation are determined based on the magnitude of the weight values; The first feature representation is input into the initial pattern matching layer to obtain the initial matching result of the home pattern output by the initial pattern matching layer; The initial home pattern matching result and the second feature representation are input into the deep pattern matching layer to obtain the home pattern matching result output by the deep pattern matching layer.

[0012] According to the home mode control method based on a large model provided in this application, the step of determining the first feature representation and the second feature representation based on the magnitude of the weight values ​​specifically includes: Based on the first weight value and the second weight value, the first feature representation and the second feature representation are determined; If the first weight value is greater than the second weight value, then the physical state characteristics of the target person are represented as the first feature representation, and the behavioral intention of the target person is represented as the second feature representation; If the second weight value is greater than the first weight value, then the behavioral intention of the target person is represented as the first feature representation, and the physical state feature of the target person is represented as the second feature representation.

[0013] Secondly, this application also provides a home mode control device based on a large model, comprising: The behavior information acquisition module is used to acquire behavior information of target personnel within a specific functional area; A status determination module is used to determine the status of the target person based on the behavioral information; The home mode activation module is used to activate a home mode that matches the target person's status.

[0014] Thirdly, this application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement any of the above-described home mode control methods based on a large model.

[0015] Fourthly, this application also provides a computer-readable storage medium comprising a stored program, wherein the program, when executed, implements the home mode control method based on the large model as described above.

[0016] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the home mode control method based on a large model as described above.

[0017] The home mode control method and device based on a large model provided in this application acquires the behavioral information of a target person in a specific functional area; determines the physical state and behavioral intention of the target person based on the behavioral information; and activates a home mode matching the target person based on the physical state and behavioral intention of the target person. By using a multimodal large model to realize the intelligent activation of the home mode based on the current perception information, the user experience is improved. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of the hardware environment for a home mode control method based on a large model according to an embodiment of this application.

[0021] Figure 2 This is a flowchart illustrating the home mode control method based on a large model provided in the embodiments of this application.

[0022] Figure 3 This is a schematic diagram illustrating the process of matching home patterns based on physical state and behavioral intentions, provided in an embodiment of this application.

[0023] Figure 4 This is a flowchart illustrating the home pattern matching method based on a home pattern matching model provided in an embodiment of this application.

[0024] Figure 5 This is a flowchart illustrating the method for representing the first feature and determining the second feature provided in an embodiment of this application.

[0025] Figure 6 This is a specific embodiment of the home mode control method based on a large model provided in this application.

[0026] Figure 7 This is a schematic diagram of the structure of a home mode control device based on a large model provided in an embodiment of this application.

[0027] Figure 8 This is a schematic diagram of the electronic device provided in this application. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] According to one aspect of the embodiments of this application, a home mode control method based on a large model is provided. This large model-based home mode control method is widely applicable to whole-house intelligent digital control application scenarios such as smart homes, smart home ecosystems, smart home device ecosystems, and intelligence house ecosystems. Optionally, in this embodiment, the above-mentioned large model-based home mode control method can be applied to, for example... Figure 1 The hardware environment shown consists of terminal device 102 and server 104. For example... Figure 1As shown, server 104 is connected to terminal device 102 via a network and can be used to provide services (such as application services) to the terminal or clients installed on the terminal. A database can be set up on the server or independently of the server to provide data storage services for server 104. Cloud computing and / or edge computing services can be configured on the server or independently of the server to provide data processing services for server 104.

[0031] The aforementioned network may include, but is not limited to, at least one of the following: wired network, wireless network. The aforementioned wired network may include, but is not limited to, at least one of the following: wide area network, metropolitan area network, local area network. The aforementioned wireless network may include, but is not limited to, at least one of the following: Wi-Fi (Wireless Fidelity), Bluetooth. The terminal device 102 may not be limited to PC, mobile phone, tablet computer, smart air conditioner, smart range hood, smart refrigerator, smart oven, smart stove, smart washing machine, smart water heater, smart washing equipment, smart dishwasher, smart projector, smart TV, smart clothes rack, smart curtains, smart audio-visual equipment, smart socket, smart speaker, smart speaker box, smart fresh air equipment, smart kitchen and bathroom equipment, smart bathroom equipment, smart robot vacuum cleaner, smart window cleaning robot, smart mopping robot, smart air purifier, smart steam oven, smart microwave oven, smart water heater, smart air purifier, smart water dispenser, smart door lock, etc.

[0032] In real-world applications, smart home appliances are becoming increasingly common in daily life. However, the so-called "intelligence" simply means that users can control appliances with their voice when needed, or trigger automatic operation by setting simple rules. These rules often require manual creation by the user or multiple user actions followed by automatic discovery by a background algorithm. This often fails to meet users' expectations of true intelligence. In other words, user voice control is essentially just a voice remote control. User-designed rules require pre-setting and present a learning curve for users. On one hand, automatic discovery by the background algorithm requires a certain amount of data accumulation and has a significant time lag. Furthermore, both manually created and automatically discovered rules lack flexibility, requiring strict adherence to certain conditions to trigger. For example, a rule might set the air conditioner to 20 degrees Celsius when the room temperature reaches 30 degrees Celsius. This rule essentially means the user feels hot, so the air conditioner will turn on. However, if the room temperature is currently 29.5 degrees Celsius, this rule won't trigger. Or, if the user has a cold and the temperature reaches 30 degrees Celsius, they might not want to turn on the air conditioner, but the rule will still activate it. On the other hand, some of the conditions mentioned above may be very complex, requiring users to make complex configurations. The application also needs to provide support for these complex conditions for users to obtain funding. Furthermore, for algorithmic mining, the more complex the combination of conditions, the less likely they are to occur; therefore, data scarcity makes it impossible to discover these rules. Thirdly, the system itself cannot utilize existing global knowledge. User rules are configured by users based on their own understanding, and the algorithm's automatic mining is also based on a very small amount of data within the system. Different users have different devices at home, and the rules accumulated within the system or some rules configured by users cannot be directly applied to new users. Each new user's rules need to be set from scratch, or the system needs to re-mine the rules. With global knowledge, these differences can be well accommodated. For example, if user A has an air conditioner and a rule for turning it on when it's hot, and user B is a new user with only a fan, the rule for turning on the air conditioner when it's hot cannot be automatically matched to user B. However, if the system can utilize global knowledge, it's different. Global knowledge includes common sense: both air conditioners and electric fans can be used for cooling. If we could utilize global knowledge, the rule of turning on the air conditioner when it's hot could automatically be changed to turning on the fan when it's hot in user B's home. Fourthly, existing systems' various automation rules do not utilize multimodal information such as images. For example, in the case of the user having a cold, if the system could use multimodal information, such as detecting through a camera that the user has a runny nose, and then combining this with global knowledge, it would also determine that the air conditioner should not be turned on in that situation.

[0033] In view of the above problems, this application provides a home mode control method based on a large model. By utilizing multimodal large model world knowledge and understanding of multimodal information, combined with corresponding user interaction logic, the home mode control is made more intelligent.

[0034] Figure 2 This is a flowchart illustrating the home mode control method based on a large model provided in this application embodiment, such as... Figure 2 As shown, the method includes: Step 110: Obtain behavioral information of target personnel within a specific functional area; Step 120: Based on the behavioral information, determine the target person's physical state and behavioral intentions; Step 130: Based on the target person's physical condition and behavioral intentions, activate the home mode that matches the target person.

[0035] The above steps will be explained in detail below with reference to specific embodiments.

[0036] Step 110: Obtain behavioral information of target personnel within a specific functional area; In this step, the behavioral information is obtained based on the analysis of the target person's behavior within a predetermined time period; Correspondingly, obtaining the behavioral information of target personnel within a specific functional area includes: Capture the actions of the target personnel within a predetermined timeframe in a specific functional area in real time; The actions of the target person within a predetermined time period are analyzed to obtain the behavioral information of the target person.

[0037] Optionally, the behavioral information is obtained through intelligent interactive devices within a specific functional area; Correspondingly, obtaining the behavioral information of target personnel within a specific functional area includes: Acquire the interaction information between the target person and the intelligent interactive devices within a specific functional area; The interaction information is analyzed to obtain the behavioral information of the target personnel within the specific functional area.

[0038] It is understandable that a large area is generally divided into multiple sub-areas based on the specific functions to be implemented. Here, the entire large area can be considered as a single functional area, or the sub-areas can be matched with different home modes separately. For example, within a residence, the residence can be treated as a single functional area for overall home mode matching. In this overall home mode matching, all smart devices within the residence are considered functional units of the home mode. Alternatively, a residence can be divided into leisure areas such as the living room, rest areas such as the bedroom, and dining areas such as the dining room. Functional units in the leisure area could include smart air conditioning, smart TV, smart lights, smart curtains, smart speakers, and smart monitoring equipment in the living room; smart air conditioning, smart lights, smart curtains, and smart speakers in the bedroom; and smart lights, smart air conditioning, and smart water supply equipment in the dining area. This application does not specifically limit the functional units.

[0039] Step 120: Based on the behavioral information, determine the target person's physical state and behavioral intentions; Step 110 obtains behavioral information about the target person within a specific functional area. This information can include physiological characteristics reflecting the target person's physical condition, such as a runny nose, sneezing, or a voice output saying "I have a cold today, I don't feel well," or yawning, turning off the lights, and covering up with a blanket when feeling tired. It can also include behavioral actions reflecting the target person's intentions, such as turning on the projector, setting food on the table, and sitting in a dining chair. Through this behavioral information, information such as the target person's physical condition and what they want to do can be obtained, thus determining the target person's physical condition and behavioral intentions.

[0040] Specifically, based on the behavioral information, determining the target person's physical state and behavioral intentions can be achieved using large models such as chat GPT or other semantic models. This application does not specifically limit the determination process. Step 130: Based on the target person's physical condition and behavioral intentions, activate the home mode that matches the target person.

[0041] In existing technologies, rules for the activation of each smart device are often pre-set. This means that the activation conditions for each device may be isolated, and the smart device may only detect the characteristics that require its activation. For example, if the pre-set rule is to turn on the air conditioner when the room temperature reaches 30 degrees Celsius, then the air conditioner will only automatically turn on when the room temperature reaches 30 degrees Celsius. However, this rule may deviate from the user's actual intention: to turn on the air conditioner when it's hot. Or, if the user is unwell, such as having a cold, but the room temperature reaches 32 degrees Celsius, the smart air conditioner may still turn on, without taking the user's actual physical condition into account. Alternatively, the rule may still be set to turn on the air conditioner when the temperature reaches 30 degrees Celsius, but the user's actual intention is to cool down when it's hot. However, if the user only has a smart fan, according to the above rule, even if the temperature reaches 30 degrees Celsius, the smart fan will not automatically start. This can lead to a poor user experience.

[0042] In this embodiment, the activation and adjustment of smart devices are not limited to preset rules, but focus on the user's physical state and behavioral intentions. Furthermore, the user's physical state and behavioral intentions affect not only a specific device, but also all smart devices in the entire functional area where the user resides; this can be considered as a home mode composed of the different states of these smart devices. In step 130, based on the target person's physical state and behavioral intentions, a home mode matching the target person is activated. Compared to existing technologies that influence the state of specific devices based on a certain rule, this approach utilizes multimodal information and its understanding, combined with corresponding user interaction logic, to make home mode control more intelligent and improve user experience.

[0043] Based on the above embodiments, Figure 3 This is a schematic diagram illustrating the process of matching home patterns based on physical state and behavioral intentions, as provided in an embodiment of this application. Figure 3 As shown, step 130 specifically includes: Step 131: Based on the target person's physical state and behavioral intention, obtain the target person's physical state parameters and behavioral intention parameters respectively; Step 132: Input the target person's physical state parameters and behavioral intention parameters into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model; wherein, the home pattern matching model is trained based on the sample physical state parameters, sample behavioral intention parameters of the sample person and the sample home patterns that match the sample physical state parameters and sample behavioral intention parameters. Step 133: Based on the home mode matching result, activate the home mode matched with the target person.

[0044] Specifically, after obtaining the target person's physical state and behavioral intentions, it is also necessary to obtain the target person's physical state parameters and behavioral intention parameters based on the target person's physical state and behavioral intentions respectively.

[0045] Furthermore, the body state parameters and behavioral intention parameters of the target person are input into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model; wherein, the home pattern matching model is trained based on the sample body state parameters, sample behavioral intention parameters and sample home patterns that match the sample behavioral information.

[0046] Optionally, the home pattern matching model is a pre-trained model used to determine the home pattern matching the target person's behavior information based on the body state parameters and sample behavior intention parameters corresponding to the input target person's behavior information, and output the home pattern matching result. Here, the home pattern matching result can be "complete match" or "partial match". "Complete match" indicates that the home pattern matched by the target person's behavior information is a historical home pattern matched with historical behavior information that is completely identical to the target person's current behavior information. "Partial match" indicates that the home pattern matched by the target person's behavior information is a historical home pattern matched with historical behavior information that is partially identical to the target person's current behavior information. When the matching result is a partial match, it is necessary to adjust the matching result according to the target person's real-time behavior information to obtain a home pattern that completely matches the current behavior information. "Complete match" is a special embodiment of the entire method implementation. In order to fully cover all embodiments, this application embodiment is described according to the case of "partial match", or the home pattern control method based on a large model provided in this application embodiment is applicable to both cases, so it is not necessary to make specific limitations on this.

[0047] In addition, before executing step 132, a home pattern matching model can be pre-trained. Specifically, the home pattern matching model can be trained in the following way: First, collect a large amount of sample behavior information of sample people in a specific functional area, and obtain the sample body state and sample behavior intention of the sample person corresponding to each sample behavior information. Based on the sample body state and sample behavior intention of the sample person, obtain the sample body state parameters and sample behavior intention parameters of the sample behavior information, and at the same time determine the sample home pattern corresponding to the sample behavior information.

[0048] Based on the above embodiments, in this method, the home pattern matching model includes a feature extraction layer, a weight allocation layer, and a home pattern matching layer; correspondingly, Figure 4 This is a flowchart illustrating the home pattern matching method based on a home pattern matching model provided in an embodiment of this application, as shown below. Figure 4As shown, step 132 specifically includes: Step 1321: Input the body state parameters and behavioral intention parameters of the target person into the feature extraction layer to obtain the body state feature representation and behavioral intention representation of the target person output by the feature extraction layer; Step 1322: Input the physical state feature representation and behavioral intention representation of the target person into the weight allocation layer to obtain the first weight value corresponding to the physical state feature representation and the second weight value corresponding to the behavioral intention representation of the target person output by the weight allocation layer. The first weight value is used to characterize the importance of the physical state feature representation, and the second weight value is used to characterize the importance of the behavioral intention representation. Step 1323: Input the physical state feature representation and behavioral intention representation of the target person, the first weight value corresponding to the physical state feature representation and the second weight value corresponding to the behavioral intention representation of the target person into the home pattern matching layer to obtain the home pattern matching result output by the home pattern matching layer.

[0049] Specifically, in this method, the home pattern matching layer includes an initial pattern matching layer and a deep pattern matching layer; the initial pattern matching layer is used to determine the historical home patterns stored in the initial pattern matching layer that match the first feature representation of the input; the deep pattern matching layer is used to correct the historical home patterns according to the second feature representation of the target user; correspondingly, Figure 5 A flowchart illustrating the method for representing the first feature and determining the second feature provided in an embodiment of this application is shown below. Figure 5 As shown, step 1323 specifically includes: Step 1323-1: Determine the first feature representation and the second feature representation based on the magnitude of the weight values; In this step, the first feature representation and the second feature representation can be determined based on the first weight value and the second weight value.

[0050] Optionally, if the first weight value is greater than the second weight value, then the physical state characteristics of the target person are represented as the first feature representation, and the behavioral intention of the target person is represented as the second feature representation; Optionally, if the second weight value is greater than the first weight value, then the behavioral intention of the target person is represented as the first feature representation, and the physical state feature of the target person is represented as the second feature representation.

[0051] Step 1323-2: Input the first feature representation into the initial pattern matching layer to obtain the initial matching result of the home pattern output by the initial pattern matching layer; Step 1323-3: Input the initial home pattern matching result and the second feature representation into the deep pattern matching layer to obtain the home pattern matching result output by the deep pattern matching layer.

[0052] Figure 6 This is a specific embodiment of the home mode control method based on a large model provided in this application, such as... Figure 6 As shown, Scenario 1 in this embodiment represents some actions of a user watching a movie at home, such as language interaction between the user and the smart device. For example, the user might output, "The heating is broken, it's a bit cold," and the smart device can understand the user's intent based on semantic understanding: needing heating. Another example is the user outputting, "The air is dry," which can be understood as needing humidification; or "The light through the window is uncomfortable," which can be understood as needing to close the curtains, etc. Through interaction with various related smart devices, the smart devices achieve intelligent adjustment. When a certain state is reached, and the user is satisfied with the current state or feels comfortable, the current home mode and the user's usage scenario 1 (the movie-watching scenario) will be saved via voice or other means. The current home mode and the rules matching scenario 1 will also be stored in the rule base. There are many ways to express the command to save the current state; this embodiment does not specifically limit this.

[0053] In other words, the user has many devices in their home, including various ambient lights, smart curtains, air conditioners, humidifiers, etc. The user wants to watch a movie, and since it's winter and the weather is dry, they turn on the humidifier. The heating in the house seems to be malfunctioning, so they turn on the air conditioner's heating mode, close the windows, and adjust the ambient lights. Then the user says, "Save current state," and the system saves the states of all the devices. The next time the user wants to watch a movie, the system detects this and automatically executes the previously saved state. However, the multimodal big model uses common sense to determine that the current air humidity is suitable; turning on the humidifier might make the air too humid, so it doesn't turn it on. It also finds the current temperature comfortable, so it doesn't turn on the air conditioner either. It only closes the windows and adjusts the ambient lights.

[0054] Based on this, when smart devices in the home obtain the user's current behavior information and analyze it to determine if the user might want to watch a movie, they will search for a matching home mode in the rule base. Based on this home mode, they will perform deep adaptation. For example, if the air conditioner is on in a previously matched home mode, but the air conditioner is broken and cannot respond in the current movie-watching scenario, the heating function can be implemented by other devices in the home, such as turning on a fan. By matching functions, not just devices, the home mode is fully adapted and then activated.

[0055] As can be seen from any of the above embodiments, the home mode control based on a large model provided in this application, compared with traditional methods of adapting to smart devices, does not require users to manually configure rules, nor does it require large-scale data mining in the background to create device linkage rules. Users can simply say "save the current home state" to save the current state of each appliance, such as whether the air conditioner is on, the current temperature, whether the lights are on, and their brightness. Furthermore, the multimodal large model automatically applies the previously saved device states based on the current perception information, making the triggering of saved device states more intelligent.

[0056] It is understandable that, such as Figure 6 Multimodal large-scale intelligent decision-making can form a specific rule base by assigning certain weights to various behavioral information or by learning user habits and preferences through interaction between the model and the user.

[0057] The home mode control method based on a large model provided in this application obtains the behavioral information of target personnel in a specific functional area; determines the physical state and behavioral intention of the target personnel based on the behavioral information; and activates a home mode matching the target personnel based on the physical state and behavioral intention of the target personnel. By using a multimodal large model to realize the intelligent activation of the home mode based on the current perception information, the user experience is improved.

[0058] The home mode control device based on a large model provided in this application is described below. The home mode control device based on a large model described below can be referred to in correspondence with the home mode control method based on a large model described above.

[0059] Based on any of the above embodiments Figure 7 This is a schematic diagram of the structure of a home mode control device based on a large model provided in an embodiment of this application, as shown below. Figure 7 As shown, the home mode control device based on a large model includes a behavior information acquisition module 710, a status determination module 720, and a home mode activation module 730. Among them, the behavior information acquisition module 710 is used to acquire the behavior information of target personnel within a specific functional area; The status determination module 720 is used to determine the status of the target person based on the behavioral information; Home mode activation module 730 is used to activate a home mode that matches the target person based on the target person's status.

[0060] The home mode control device based on a large model provided in this application acquires behavioral information of target personnel within a specific functional area; determines the physical state and behavioral intentions of the target personnel based on the behavioral information; and activates a home mode matching the target personnel based on the physical state and behavioral intentions of the target personnel. By using a multimodal large model to realize the intelligent activation of the home mode based on the current perception information, the user experience is improved.

[0061] Based on any of the above embodiments, the home mode activation module 730 in this device includes: The submodule for obtaining body state and behavioral intention is used to obtain the body state parameters and behavioral intention parameters of the target person based on their body state and behavioral intention, respectively. The home pattern matching submodule is used to input the target person's physical state parameters and behavioral intention parameters into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model; wherein, the home pattern matching model is trained based on the sample physical state parameters, sample behavioral intention parameters of the sample person and sample home patterns that match the sample physical state parameters and sample behavioral intention parameters. The home mode activation submodule activates the home mode matched with the target person based on the home mode matching result.

[0062] Based on any of the above embodiments, the home pattern matching model in this device includes a feature extraction layer, a weight allocation layer, and a home pattern matching layer; Correspondingly, the home mode matching submodule includes: The feature extraction unit is used to input the body state parameters and behavioral intention parameters of the target person into the feature extraction layer to obtain the body state feature representation and behavioral intention feature representation of the target person output by the feature extraction layer; The weight value acquisition unit is used to input the physical state feature representation and behavioral intention representation of the target person into the weight allocation layer to obtain the first weight value corresponding to the physical state feature representation and the second weight value corresponding to the behavioral intention representation of the target person output by the weight allocation layer. The first weight value is used to characterize the importance of the physical state feature representation, and the second weight value is used to characterize the importance of the behavioral intention representation. The matching unit is used to input the physical state feature representation and behavioral intention representation of the target person, the first weight value corresponding to the physical state feature representation and the second weight value corresponding to the behavioral intention representation of the target person into the home pattern matching layer, so as to obtain the home pattern matching result output by the home pattern matching layer.

[0063] Based on any of the above embodiments, in this device, the behavior information acquisition module 610 is specifically used for: Capture the actions of the target personnel within a predetermined timeframe in a specific functional area in real time; The actions of the target person within a predetermined time period are analyzed to obtain the behavioral information of the target person.

[0064] Based on any of the above embodiments, the behavior information acquisition module 610 in this device is further configured to: Acquire the interaction information between the target person and the intelligent interactive devices within a specific functional area; The interaction information is analyzed to obtain the behavioral information of the target personnel within the specific functional area.

[0065] Based on any of the above embodiments, in this device, the matching unit is specifically used for: The first feature representation and the second feature representation are determined based on the magnitude of the weight values; The first feature representation is input into the initial pattern matching layer to obtain the initial matching result of the home pattern output by the initial pattern matching layer; The initial home pattern matching result and the second feature representation are input into the deep pattern matching layer to obtain the home pattern matching result output by the deep pattern matching layer.

[0066] Based on any of the above embodiments, in this device, the matching unit is further configured to: Based on the first weight value and the second weight value, the first feature representation and the second feature representation are determined; If the first weight value is greater than the second weight value, then the physical state characteristics of the target person are represented as the first feature representation, and the behavioral intention of the target person is represented as the second feature representation; If the second weight value is greater than the first weight value, then the behavioral intention of the target person is represented as the first feature representation, and the physical state feature of the target person is represented as the second feature representation.

[0067] Figure 8 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 8As shown, the electronic device may include a processor 810, a communications interface 820, a memory 830, and a communication bus 840. The processor 810, communications interface 820, and memory 830 communicate with each other via the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute a home mode control method based on a large model. This method includes: acquiring behavioral information of a target person within a specific functional area; determining the target person's physical state and behavioral intentions based on the behavioral information; and activating a home mode matched to the target person based on their physical state and behavioral intentions.

[0068] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0069] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the home mode control method based on a large model provided by the above methods. The method includes: acquiring behavioral information of a target person in a specific functional area; determining the physical state and behavioral intention of the target person based on the behavioral information; and activating a home mode matching the target person based on the physical state and behavioral intention of the target person.

[0070] In another aspect, this application also provides a computer-readable storage medium, the computer-readable storage medium including a stored program, wherein the program executes the home mode control method based on the large model provided by the above methods when it runs, the method including: acquiring behavioral information of a target person in a specific functional area; determining the physical state and behavioral intention of the target person based on the behavioral information; and activating a home mode matched to the target person based on the physical state and behavioral intention of the target person.

[0071] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0072] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A home mode control method based on a large model, characterized in that, include: Obtain behavioral information of target personnel within a specific functional area; Based on the behavioral information, the physical state and behavioral intentions of the target person are determined; Based on the target person's physical condition and behavioral intentions, activate the home mode that matches the target person.

2. The home mode control method based on a large model according to claim 1, characterized in that, The acquisition of behavioral information of target personnel within a specific functional area includes: Capture the actions of the target personnel within a predetermined timeframe in a specific functional area in real time; The actions of the target person within a predetermined time period are analyzed to obtain the behavioral information of the target person.

3. The home mode control method based on a large model according to claim 1, characterized in that, The acquisition of behavioral information of target personnel within a specific functional area includes: Acquire the interaction information between the target person and the intelligent interactive devices within a specific functional area; The interaction information is analyzed to obtain the behavioral information of the target personnel within the specific functional area.

4. The home mode control method based on a large model according to claim 1, characterized in that, The step of activating a home mode matching the target person based on their physical condition and behavioral intentions includes: Based on the target person's physical state and behavioral intentions, respectively, obtain the target person's physical state parameters and behavioral intention parameters; The target person's physical state parameters and behavioral intention parameters are input into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model; wherein, the home pattern matching model is trained based on the sample physical state parameters, sample behavioral intention parameters and sample home patterns that match the sample behavioral information; based on the home pattern matching result, the home pattern matching the target person is activated.

5. The home mode control method based on a large model according to claim 4, characterized in that, The home pattern matching model includes a feature extraction layer, a weight allocation layer, and a home pattern matching layer; Correspondingly, the step of inputting the target person's physical state parameters and behavioral intention parameters into the home pattern matching model to obtain the home pattern matching result output by the home pattern matching model includes: The body state parameters and behavioral intention parameters of the target person are input into the feature extraction layer to obtain the body state feature representation and behavioral intention representation of the target person output by the feature extraction layer. The target person's physical state feature representation and behavioral intention representation are input into the weight allocation layer to obtain the first weight value corresponding to the target person's physical state feature representation and the second weight value corresponding to the behavioral intention representation output by the weight allocation layer. The first weight value is used to characterize the importance of the physical state feature representation, and the second weight value is used to characterize the importance of the behavioral intention representation. The target person's physical state characteristics and behavioral intentions, along with the first weight value corresponding to the target person's physical state characteristics and the second weight value corresponding to the behavioral intentions, are input into the home pattern matching layer to obtain the home pattern matching result output by the home pattern matching layer.

6. The home mode control method based on a large model according to claim 5, characterized in that, The home pattern matching layer includes an initial pattern matching layer and a deep pattern matching layer; the initial pattern matching layer is used to determine the historical home patterns stored in the initial pattern matching layer that match the first feature representation of the input; the deep pattern matching layer is used to correct the historical home patterns according to the second feature representation of the target user; Correspondingly, the step of inputting the target person's physical state feature representation and behavioral intention representation, the first weight value corresponding to the target person's physical state feature representation, and the second weight value corresponding to the behavioral intention representation into the home pattern matching layer to obtain the home pattern matching result output by the home pattern matching layer specifically includes: The first feature representation and the second feature representation are determined based on the magnitude of the weight values; The first feature representation is input into the initial pattern matching layer to obtain the initial matching result of the home pattern output by the initial pattern matching layer; The initial home pattern matching result and the second feature representation are input into the deep pattern matching layer to obtain the home pattern matching result output by the deep pattern matching layer.

7. The home mode control method based on a large model according to claim 6, characterized in that, The step of determining the first feature representation and the second feature representation based on the weight values ​​specifically includes: Based on the first weight value and the second weight value, the first feature representation and the second feature representation are determined; If the first weight value is greater than the second weight value, then the physical state characteristics of the target person are represented as the first feature representation, and the behavioral intention of the target person is represented as the second feature representation; If the second weight value is greater than the first weight value, then the behavioral intention of the target person is represented as the first feature representation, and the physical state feature of the target person is represented as the second feature representation.

8. A home mode control device based on a large model, characterized in that, include: The behavior information acquisition module is used to acquire behavior information of target personnel within a specific functional area; A status determination module is used to determine the status of the target person based on the behavioral information; The home mode activation module is used to activate a home mode that matches the target person's status.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program, when executed, performs the home mode control method based on a large model as described in any one of claims 1 to 7.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to execute the home mode control method based on a large model according to any one of claims 1 to 7 through the computer program.

Images