Air conditioner and control method thereof

Abstract

The embodiment of the invention discloses an air conditioner and a control method thereof, relates to the technical field of air conditioners, and aims to improve the automatic adjustment accuracy of operation parameters of the air conditioner. The air conditioner comprises a controller which is configured to obtain multiple pieces of physiological state data of a user in an area where the air conditioner is located under the condition that the air conditioner runs in a sleep mode; inputting the multiple pieces of physiological state data into a data analysis model, extracting atomic-scale correlation features among the multiple pieces of physiological state data through an atomic-scale feature extraction module in the data analysis model, and extracting component-scale correlation features among the multiple pieces of physiological state data through a component-scale feature extraction module in the data analysis model, obtaining an operation parameter adjustment value of the air conditioner through a parameter prediction module in the data analysis model based on the atomic-level correlation characteristics and the component-level correlation characteristics; and adjusting operation parameters of the air conditioner based on the operation parameter adjustment value.

Description

Technical Field

[0001] This application relates to the field of air conditioner technology, and in particular to an air conditioner and its control method. Background Technology

[0002] As people pay more attention to health and quality of life, sleep quality has gradually become one of the focuses of attention. A good sleep environment is crucial for maintaining physical and mental health, and the automatic adjustment technology of air conditioner operating parameters in sleep mode has become one of the important means to improve sleep quality.

[0003] Existing automatic adjustment technologies for operating parameters mainly rely on single physiological state data, such as heart rate data. This technology cannot fully consider the relationship between different physiological state data, and the vital signs reflected by a single physiological state data may be inconsistent, making it impossible to fully understand the user's sleep state.

[0004] Therefore, how to accurately analyze physiological data so that adjusting the operating parameters of air conditioners can effectively improve users' sleep quality has become an urgent problem to be solved. Summary of the Invention

[0005] This application provides an air conditioner and its control method to improve the accuracy of automatic adjustment of the air conditioner's operating parameters.

[0006] To achieve the above objectives, this application adopts the following technical solution.

[0007] In a first aspect, embodiments of this application provide an air conditioner, including: a controller configured to: acquire multiple physiological state data of users in the area where the air conditioner is located when the air conditioner is operating in sleep mode; input the multiple physiological state data into a data analysis model; extract atomic-level correlation features between the multiple physiological state data through an atomic-level feature extraction module in the data analysis model; extract component-level correlation features between the multiple physiological state data through a component-level feature extraction module in the data analysis model; obtain an adjustment value for the operating parameters of the air conditioner based on the atomic-level correlation features and the component-level correlation features through a parameter prediction module in the data analysis model; and adjust the operating parameters of the air conditioner based on the adjustment value for the operating parameters.

[0008] The technical solution provided in this application provides at least the following beneficial effects: This technical solution acquires multiple physiological state data of the user during sleep mode, and analyzes this data to obtain the air conditioner's operating parameter adjustment values, making the air conditioner's operating parameters more aligned with the user's physiological needs. Simultaneously, it obtains atomic-level and component-level correlation features of multiple physiological state data through the atomic-level feature extraction module and component-level feature extraction module in the data analysis model, respectively. Then, the parameter prediction module in the data analysis model determines the air conditioner's operating parameter adjustment values ​​based on these atomic-level and component-level correlation features. This allows for a more comprehensive understanding of the interaction and relationship between different physiological state data, resulting in operating parameter adjustment values ​​that better meet the user's physiological needs, providing a more comfortable and healthy sleep environment, and improving the user's sleep quality and quality of life.

[0009] In some embodiments, the atomic-level feature extraction module includes a cross-attention module and a multi-head cross-attention module. The controller extracts atomic-level correlation features between multiple physiological state data through the atomic-level feature extraction module in the data analysis model. Specifically, it is configured to: after feature extraction for each physiological state data, perform feature enhancement through the cross-attention module to obtain the physiological state features corresponding to each physiological state data; perform multi-head cross-attention operations on each physiological state feature through the multi-head cross-attention module to obtain the atomic-level features corresponding to each physiological state feature; and fuse the atomic-level features corresponding to multiple physiological state data to obtain the atomic-level correlation features between multiple physiological state data.

[0010] In some embodiments, the component-level feature extraction module includes a graph convolution module; the controller extracts component-level association features between multiple physiological state data through the component-level feature extraction module in the data analysis model, specifically configured to: determine the graph structure data corresponding to each physiological state data, and perform graph convolution operation on the graph structure data through the graph convolution module to obtain the graph data features corresponding to each physiological state data; and determine the component-level association features between multiple physiological state data based on the graph data features corresponding to each physiological state data.

[0011] In some embodiments, the controller is further configured to: acquire multiple sample data and the actual operating parameter adjustment value corresponding to each sample data; each sample data includes multiple sample physiological state data; input the sample data into an initial data analysis model to obtain the predicted operating parameter adjustment value corresponding to the sample data; determine the loss function corresponding to the sample data based on the actual operating parameter adjustment value and the predicted operating parameter adjustment value corresponding to the sample data; and iteratively train the initial data analysis model based on the loss function corresponding to the sample data to obtain the trained data analysis model.

[0012] In some embodiments, the controller iteratively trains the initial data analysis model based on the loss function corresponding to the sample data. Specifically, the controller is configured to: determine the atomic-level features corresponding to the physiological state data of each sample in the sample data; determine the similarity between multiple atomic-level features in the sample data; determine the atomic-level loss function corresponding to the sample data based on the similarity; and iteratively train the initial data analysis model based on the loss function corresponding to the sample data and the atomic-level loss function.

[0013] In some embodiments, multiple physiological state data include audio data, heart rate data, and facial data; operating parameters include at least one of the following: operating air volume, operating temperature, humidity level, air swing frequency, operating time, and fan speed.

[0014] Secondly, embodiments of this application provide a method for controlling an air conditioner, comprising: acquiring multiple physiological state data of users in the area where the air conditioner is located when the air conditioner is operating in sleep mode; inputting the multiple physiological state data into a data analysis model; extracting atomic-level correlation features between the multiple physiological state data through an atomic-level feature extraction module in the data analysis model; extracting component-level correlation features between the multiple physiological state data through a component-level feature extraction module in the data analysis model; obtaining an adjustment value for the operating parameters of the air conditioner based on the atomic-level correlation features and the component-level correlation features through a parameter prediction module in the data analysis model; and adjusting the operating parameters of the air conditioner based on the adjustment value for the operating parameters.

[0015] Thirdly, embodiments of this application provide a controller, including: one or more processors; one or more memories; wherein the one or more memories are used to store computer program code, the computer program code including computer speech, and when the one or more processors execute the computer speech, the controller executes the method provided in the second aspect.

[0016] Fourthly, embodiments of this application provide a computer-readable storage medium including computer speech, which, when executed on a computer, causes the computer to perform the methods provided in the second aspect and possible implementations.

[0017] Fifthly, embodiments of the present invention provide a computer program product that can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, the computer program product can implement the methods provided in the second aspect and possible implementations.

[0018] It should be noted that the aforementioned computer speech can be stored, in whole or in part, on a computer-readable storage medium. This computer-readable storage medium can be packaged together with the controller's processor, or it can be packaged separately from the controller's processor; this application does not impose any limitations on this.

[0019] The beneficial effects described in aspects two through five of this application can be referred to the analysis of the beneficial effects of aspect one, and will not be repeated here. Attached Figure Description

[0020] The accompanying drawings are provided to further understand the technical solutions of the present invention and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of the present invention and do not constitute a limitation on the technical solutions of the present invention.

[0021] Figure 1 This is a schematic diagram of the structure of an air conditioner provided in an embodiment of this application;

[0022] Figure 2 A hardware configuration block diagram of an air conditioner provided in an embodiment of this application;

[0023] Figure 3 This is a schematic flowchart of a control method for an air conditioner provided in an embodiment of this application;

[0024] Figure 4 A bar chart illustrating a method for adjusting operating parameters provided in an embodiment of this application;

[0025] Figure 5 A schematic flowchart illustrating a model training method provided in an embodiment of this application;

[0026] Figure 6 A schematic flowchart illustrating a method for determining a loss function provided in an embodiment of this application;

[0027] Figure 7 This is a schematic flowchart of another air conditioner control method provided in an embodiment of this application;

[0028] Figure 8 A schematic diagram illustrating another loss function determination process provided in an embodiment of this application;

[0029] Figure 9 A schematic diagram of a data processing flow provided in an embodiment of this application;

[0030] Figure 10 This is a schematic diagram of another data processing flow provided in an embodiment of this application;

[0031] Figure 11 This is a schematic diagram of another data processing flow provided in an embodiment of this application;

[0032] Figure 12 This is a schematic diagram of another data processing flow provided in an embodiment of this application;

[0033] Figure 13 This is a schematic diagram of another air conditioner provided in an embodiment of this application. Detailed Implementation

[0034] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0035] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0036] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0037] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "connected" and "linked" should be interpreted broadly, for example, as a fixed connection, a detachable connection, or an integral connection. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances. Furthermore, when describing pipelines, the terms "connected" and "linked" as used in this application have the meaning of establishing electrical connection. The specific meaning needs to be understood in conjunction with the context.

[0038] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0039] To facilitate understanding, we will first provide a brief introduction and explanation of some terms or basic concepts of technology involved in the embodiments of the present invention.

[0040] As mentioned above, traditional air conditioners adjust their operating parameters in sleep mode mostly based on single physiological state data, failing to fully understand the interaction and relationship between different vital signs, thus preventing the adjustment of operating parameters from improving the user's sleep quality.

[0041] Based on this, this application provides an air conditioner that analyzes the user's physiological state data using a data analysis model to obtain the air conditioner's operating parameter adjustment values ​​output by the data analysis model, thereby adjusting the air conditioner's operating parameters. This improves the accuracy of the automatic adjustment of the air conditioner's operating parameters, providing users with a more comfortable and healthy sleep environment, and optimizing the user's sleep quality and quality of life.

[0042] The embodiments provided in this application will now be described in detail with reference to the accompanying drawings.

[0043] Figure 1 This is a schematic diagram of the structure of an air conditioner provided as an example of an embodiment of this application, such as... Figure 1 As shown, the air conditioner 1 includes an outdoor unit 2.

[0044] It should be noted that the air conditioner can be a multi-split air conditioner, a central air conditioner, etc., and this application does not limit the type of air conditioner.

[0045] As a feasible implementation method, the air conditioner 1 and the remote control 4 are connected by communication.

[0046] As a feasible implementation method, outdoor unit 2 is usually installed outdoors for heat exchange in the indoor environment.

[0047] In some embodiments, the air conditioner 1 includes an indoor unit 3.

[0048] As a feasible implementation method, indoor unit 3, taking indoor wall-mounted units as an example, is typically installed on indoor walls (such as in bedrooms, living rooms, conference rooms, etc.). Another example is indoor floor-standing units (…). Figure 1 (Not shown in the image) is also a type of indoor unit.

[0049] It should be noted that indoor unit 3 may include at least one indoor unit, and this application does not limit the number of indoor units of the air conditioner.

[0050] In some embodiments, the air conditioner 1 includes a data acquisition device 5 ( Figure 1 (Not shown in the image) is used to collect multiple physiological state data within the area where the air conditioner 1 is located.

[0051] For example, the data acquisition device 5 can be a signal detection device to detect multiple physiological characteristics of users in the area where the air conditioner 1 is located; it can also be a motion vital sign measurement device, such as a fitness tracker. This application does not limit the type or configuration of the data acquisition device 5.

[0052] Figure 2 This is a hardware configuration block diagram of an air conditioner provided according to an exemplary embodiment of this application. For example... Figure 2 As shown, the air conditioner 1 may also include a controller 1000.

[0053] In some embodiments, the controller 1000 is the control center of the air conditioner 1, which connects various parts of the air conditioner 1 through various interfaces and lines, and executes various programs of the air conditioner 1 by running or executing programs stored in the memory and calling data stored in the memory.

[0054] In some embodiments, the controller 1000 may include one or more processing units 1001.

[0055] In the embodiments shown in this application, controller 1000 refers to a device that can generate operation control signals based on voice operation codes and timing signals to instruct air conditioner 1 to execute control methods. Exemplarily, controller 1000 can be a central processing unit (CPU), a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a programmable logic device (PLD), or any combination thereof. Controller 1000 can also be other devices with processing functions, such as circuits, devices, or software modules; this application embodiment does not impose any limitations on this.

[0056] In addition, the controller 1000 can be used to control the various components inside the air conditioner 1 so that each component can operate to achieve the predetermined functions of the air conditioner 1.

[0057] In some embodiments, such as Figure 2 As shown, the controller 1000 includes a communicator 1002.

[0058] In some embodiments, the communicator 1002 is used to establish communication connections with other network entities, such as establishing communication connections with terminal devices. The communicator 1002 may include a radio frequency (RF) module, a cellular module, a wireless fidelity (WIFI) module, and a GPS module, etc. Taking an RF module as an example, the RF module can be used for signal reception and transmission; specifically, it sends received information to the controller 1000 for processing; additionally, it transmits signals generated by the controller 1000. Typically, the RF circuit may include, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier (LNA), a duplexer, etc.

[0059] In some embodiments, such as Figure 2 As shown, the controller 1000 also includes a memory 1003.

[0060] In some embodiments, the memory 1003 may be used to store software programs and data. The controller 1000 executes various functions of the air conditioner 1 and data processing by running the software programs or data stored in the memory 1003. The memory 1003 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. The memory 1003 stores an operating system that enables the air conditioner 1 to run. In this application, the memory 1003 may store the operating system and various application programs, and may also store code that executes the control method of the air conditioner 1 provided in the embodiments of this application.

[0061] Those skilled in the art will understand that Figure 2 The hardware structure shown does not constitute a limitation on the air conditioner 1. The air conditioner 1 may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0062] As one feasible implementation method, the air conditioner 1 communicates with the remote controller 4 through the communicator 1002.

[0063] In some embodiments, the remote controller 4 has the function of communicating with the controller 1000, for example, using infrared or other communication methods. Users can set various operating parameters of the air conditioner 1 using the remote controller 4.

[0064] Figure 3 This is a schematic flowchart of a control method for an air conditioner provided in an embodiment of this application, as shown below. Figure 3 As shown, the method includes the following steps:

[0065] S101. When the air conditioner is in sleep mode, the controller acquires multiple physiological state data.

[0066] As a feasible approach, multiple physiological data points are used, including audio data, heart rate data, and facial data.

[0067] As a feasible approach, a data acquisition device can be installed in the area where the air conditioner is located, and the controller can collect multiple physiological state data of users in that area through the data acquisition device.

[0068] It should be understood that the area where the air conditioner is located can be a bedroom, a conference room, an office, or other areas where the air conditioner may be located, and the user can be any user in the area where the air conditioner is located. This application does not limit the type of area where the air conditioner is located or the user.

[0069] It should be noted that when the air conditioner is running in sleep mode, its operating parameters may not meet the user's physiological needs. Therefore, it is necessary to continuously acquire multiple physiological state data of the user in order to adjust the air conditioner's operating parameters based on these data.

[0070] S102. The controller inputs multiple physiological state data into the data analysis model to obtain the operating parameter adjustment values ​​of the air conditioner.

[0071] The data analysis model includes an atomic-level feature extraction module, a component-level feature extraction module, and a parameter prediction module.

[0072] It should be noted that the operating parameter adjustment values ​​may include operating air volume adjustment values, operating temperature adjustment values, humidity level adjustment values, air swing frequency adjustment values, operating time adjustment values, and fan speed adjustment values. In specific implementations, the operating parameter adjustment values ​​may include more or fewer adjustment values. This application does not limit the number and type of adjustment values ​​included in the operating parameter adjustment values.

[0073] As a feasible implementation method, after the controller inputs multiple physiological state data into the data analysis model, the atomic-level feature extraction module in the data analysis model extracts the atomic-level correlation features between the multiple physiological state data.

[0074] The atomic-level feature extraction module includes a cross-attention module and a multi-head cross-attention module.

[0075] As a feasible implementation method, after acquiring multiple physiological state data of the user, the controller extracts features from each physiological state data and enhances the extracted features through a cross-attention module to obtain the physiological state features corresponding to each physiological state data.

[0076] As a feasible approach, a modality-specific face can be used to encode different monomodal features through a pre-trained audio-heart rate-face multimodal pre-training neural network (CLIP) model.

[0077] Specifically, a pre-trained CLIP model can be used to encode facial and heart rate data to obtain facial feature codes and heart rate feature codes; a pre-trained ESResNeXt model (i.e., a combination of (squeeze-and-excitation, SE) and (aggregated residual transformations for deep neural networks, ResNeXt)) can be used to encode audio data to obtain audio feature codes.

[0078] Furthermore, heart rate feature encoding, facial feature encoding, and audio feature encoding are enhanced through a cross-attention module. The attention module learns the long-term dependency relationship of the enhanced heart rate feature, and the cross-frame fusion module learns the long-term dependency relationship of the enhanced facial feature and audio feature to obtain the physiological state feature corresponding to each physiological state data.

[0079] Among them, heart rate characteristic T i The calculation method is shown in formula (1):

[0080]

[0081] Facial features V i The calculation method is shown in formula (2):

[0082]

[0083] Audio Feature A i The calculation method is shown in formula (3):

[0084]

[0085] Among them, W * With b * The matrix is ​​learnable; Attention() represents the attention module, and CrossFrameFusion() represents the cross-frame fusion module; both contain N self-attention layers; heart rate feature T i =T1, T2, ..., T M Facial features V i =V1, V2, ..., V M Audio feature Ti =A1, A2, ..., A M .

[0086] In some embodiments, a multi-head cross-attention module is used to perform multi-head cross-attention operations on each physiological state feature to obtain the atomic-level features corresponding to each physiological state feature.

[0087] It should be noted that, in order to model the atomic-level correlation between heart rate features, facial features, and audio features, different features can be integrated using a cross-attention module, as shown in formula (4):

[0088]

[0089] in, Let m ∈ {v, a} (where V is the symbol for facial features and a is the symbol for audio features), and h be the feature type of the input cross-attention module. i ∈(1, 2, ..., M) ctn (M) ctn (This refers to the hyperparameters in the cross-attention module).

[0090] Furthermore, atomic-level features of heart rate characteristics were obtained separately. Atomic features of facial features and atomic-level features of audio features The calculation method is shown in formulas (5), (6) and (7) below:

[0091]

[0092]

[0093]

[0094] As a feasible approach, after determining the atomic-level features corresponding to each physiological state feature, the atomic-level features corresponding to multiple physiological state features are fused to obtain the atomic-level correlation features between multiple physiological state data.

[0095] As a feasible approach, due to the semantic differences between different modalities, direct fusion methods may make the final adjusted values ​​of the running parameters more ambiguous. Therefore, contrastive learning can be used to constrain the feature distribution of atomic features corresponding to multiple physiological state features, that is, atomic-level contrastive learning, to obtain atomic-level correlation features between multiple physiological state data.

[0096] Specifically, facial features and audio features are fused with heart rate features using element-wise inner products, resulting in a fused feature P. ACM It is obtained by the method shown in the following formula (8):

[0097]

[0098] Here, concat() is the concatenation method.

[0099] Furthermore, the atomic-level correlation feature P between multiple physiological state data points m′ The method for obtaining it is shown in formula (9):

[0100]

[0101] Among them, P m′ ∈(vt, at),

[0102] In some embodiments, component-level correlation features between multiple physiological state data are extracted through the component-level feature extraction module in the data analysis model.

[0103] As a feasible approach, multiple physiological state data are input into the component-level feature extraction module in the data analysis model to determine the graph structure data corresponding to each physiological state data.

[0104] The component-level feature extraction module includes a graph convolution module.

[0105] As a feasible approach, after determining the graph structure data corresponding to each physiological state data, the graph structure data is subjected to graph convolution operation by the graph convolution module in the component-level feature extraction module to obtain the graph data features corresponding to each physiological state data.

[0106] It should be noted that the component-level feature extraction module can also be called the optimal transmission module. It learns the semantic distribution between facial data and heart rate data, and audio data and heart rate data through the optimal transmission process, in order to determine the graph data features corresponding to each physiological state data.

[0107] Furthermore, guided by the transfer cost matrix obtained during the optimal transmission process, facial image structure data, audio image structure data, and heart rate image structure data are combined to learn component-level correlation features between multiple physiological state data.

[0108] Specifically, the optimal transmission process is first defined using discrete Kantorovich formulas, and the output of the graph convolution module is then transmitted optimally through this process to obtain facial graph structure data. Audio graph structure data With heart rate graph structure data The optimal semantic flow between them.

[0109] The optimal semantic flow is the component-level association feature between multiple physiological state data.

[0110] The optimal transmission flow can be between audio graph structure data and heart rate graph structure data, and between facial graph structure data and heart rate graph structure data. Audio graph structure data With heart rate graph structure data The optimal semantic flow (component-level association features) between them is obtained as shown in Equations (10) and (11):

[0111]

[0112]

[0113] Where m'∈{v,a}, C m' The transfer cost matrix (defined by the Euclidean distance, which measures G) m The distance between any two local instances, μ m' and μ t It is distributed at the edge.

[0114] In some embodiments, after determining the atomic-level correlation features and component-level correlation features between multiple physiological state data, the atomic-level correlation features and component-level correlation features are input into the parameter prediction module in the data analysis model. The parameter prediction module then makes predictions based on the atomic-level correlation features and component-level correlation features to obtain the adjustment values ​​of the air conditioner's operating parameters.

[0115] As a feasible approach, atomic-level correlation features and component-level correlation features are input into the parameter prediction module in the data analysis model. The parameter prediction module then obtains the operating parameter adjustment values ​​corresponding to heart rate features, facial features, and audio features, respectively. The average of these three operating parameter adjustment values ​​is then taken as the final operating parameter adjustment value.

[0116] Figure 4 A bar chart illustrating a method for adjusting operating parameters provided in an embodiment of this application, as shown below. Figure 4 As shown, for example, for the operating air volume adjustment value, taking the operating air volume adjustment value corresponding to facial features as 0.4, the operating air volume adjustment value corresponding to heart rate features as 0.7, and the air volume adjustment value corresponding to audio features as 0.45, the final operating air volume adjustment value is (0.4+0.7+0.45) / 3≈0.5.

[0117] For example, if the operating temperature adjustment value is 0.1 for facial features, 0.11 for heart rate features, and 0.09 for audio features, then the final operating temperature adjustment value is (0.1 + 0.11 + 0.09).

[0118] / 3 = 0.1.

[0119] For example, if the humidity adjustment value is 0.85 for facial features, 0.6 for heart rate features, and 0.55 for audio features, then the final humidity adjustment value is (0.85+0.6+0.55) / 3≈0.7.

[0120] For example, taking the sweep frequency adjustment value as 0.2 for facial features, 0.16 for heart rate features, and 0.6 for audio features, the final sweep frequency adjustment value is (0.2 + 0.16 + 0.6).

[0121] / 3 = 0.32.

[0122] For example, if the runtime adjustment value is 0.55 for facial features, 0.75 for heart rate features, and 0.4 for audio features, then the final runtime adjustment value is (0.55 + 0.75 + 0.4).

[0123] / 3 = 0.57.

[0124] For example, if the fan speed adjustment value is 0.01 for facial features, 0.05 for heart rate features, and 0.15 for audio features, then the final fan speed adjustment value is (0.01+0.05+0.15) / 3=0.07.

[0125] It should be noted that, Figure 4 The example only shows the case where the running parameter adjustment value is positive (i.e., the running parameter is increased). In the actual implementation, the running parameter adjustment value can also be negative (i.e., the running parameter is decreased). This application does not limit this.

[0126] S103. The controller adjusts the operating parameters of the air conditioner based on the operating parameter adjustment value.

[0127] For example, if the operating air volume adjustment value is 0.5, the operating air volume of the air conditioner will be increased by 0.5; if the operating temperature adjustment value is 0.1, the operating temperature of the air conditioner will be increased by 0.1℃; if the humidity setting adjustment value is 0.7, the humidity setting of the air conditioner will be increased by 0.7; if the air swing frequency adjustment value is 0.32, the air swing frequency of the air conditioner will be increased by 0.32; if the air conditioner running time adjustment value is 0.57, the running time of the air conditioner will be extended by 0.57 hours; if the air conditioner fan speed adjustment value is 0.07, the fan speed of the air conditioner will be increased by 0.07.

[0128] The technical solution provided in this application provides at least the following beneficial effects: This technical solution acquires multiple physiological state data of the user during sleep mode, and analyzes this data to obtain the air conditioner's operating parameter adjustment values, making the air conditioner's operating parameters more aligned with the user's physiological needs. Simultaneously, it obtains atomic-level and component-level correlation features of multiple physiological state data through the atomic-level feature extraction module and component-level feature extraction module in the data analysis model, respectively. Then, the parameter prediction module in the data analysis model determines the air conditioner's operating parameter adjustment values ​​based on these atomic-level and component-level correlation features. This allows for a more comprehensive understanding of the interaction and relationship between different physiological state data, resulting in operating parameter adjustment values ​​that better meet the user's physiological needs, providing a more comfortable and healthy sleep environment, and improving the user's sleep quality and quality of life.

[0129] In some embodiments, since the initial data analysis model has not been trained with a large amount of data, there may be inaccurate output results. Therefore, before inputting multiple physiological state data into the data analysis model, the initial data analysis model needs to be trained to improve the data analysis accuracy of the data analysis model.

[0130] Figure 5 This is a flowchart illustrating a model training method provided in an embodiment of this application, as shown below. Figure 5 As shown, the method includes the following steps:

[0131] S201. Obtain multiple sample data and the actual operating parameter adjustment values ​​corresponding to each sample data.

[0132] Each sample data includes multiple sample physiological state data; these multiple sample physiological state data include sample audio data, sample heart rate data, and sample facial data.

[0133] S202. Input the sample data into the initial data analysis model to obtain the adjustment values ​​of the prediction operation parameters corresponding to the sample data.

[0134] Specifically, the atomic-level feature extraction module in the initial data analysis model extracts atomic-level correlation features between physiological state data of multiple samples, the component-level feature extraction module in the initial data analysis model extracts component-level correlation features between physiological state data of multiple samples, and the parameter prediction module in the initial data analysis model obtains the adjustment values ​​of the air conditioner's operating parameters based on the atomic-level correlation features and component-level correlation features.

[0135] Specifically, atomic-level correlation features are obtained by fusing audio features, heart rate features, and facial features, while component-level correlation features are obtained by fusing audio graph structure data, heart rate graph structure data, and facial graph structure data. For details of the fusion method, please refer to step S102 above, which will not be repeated here.

[0136] S203. Determine the loss function corresponding to the sample data based on the actual operating parameter adjustment values ​​and the predicted operating parameter adjustment values ​​corresponding to the sample data.

[0137] Figure 6 This is a flowchart illustrating a method for determining a loss function provided in an embodiment of this application, as shown below. Figure 6 As shown, after inputting the atomic-level correlation features into the parameter prediction module, the adjusted values ​​of the prediction running parameters output by the parameter prediction module are obtained. Subsequently, the parameter was adjusted based on the predicted cloud parameters and the actual operating parameter adjustments. Determine the loss function corresponding to the sample data.

[0138] As a feasible implementation method, the loss function corresponding to the sample data can be the category-aware cross-entropy loss function L, which represents the relationship between multimodal data and runtime parameter adjustments. cce As shown in formula (12):

[0139]

[0140] S204. Based on the loss function corresponding to the sample data, iteratively train the initial data analysis model to obtain the trained data analysis model.

[0141] As a feasible approach, the initial data analysis model can also be iteratively trained based on the atomic loss function corresponding to the sample data and the loss function corresponding to the sample data. Figure 7 This is a schematic flowchart of another air conditioner control method provided in an embodiment of this application, used to determine the atomic-level loss function, such as... Figure 7 As shown, the method includes the following steps:

[0142] S301. Determine the atomic-level features corresponding to the physiological state data of each sample in the sample data.

[0143] It should be noted that the method for determining the atomic-level features corresponding to the physiological state data of each sample is detailed in step S102 above, and will not be repeated here.

[0144] S302. Determine the similarity between multiple atomic-level features in the sample data.

[0145] The similarity between multiple atomic-level features includes the similarity between facial features and heart rate features, the similarity between audio features and heart rate features, and the similarity between facial features and audio features.

[0146] As a feasible implementation method, the similarity CS1 between facial features and heart rate features is calculated as shown in formula (13):

[0147]

[0148] The similarity CS2 between audio features and heart rate features is calculated as shown in formula (14):

[0149]

[0150] The similarity CS3 between facial features and audio features is calculated as shown in formula (15):

[0151]

[0152] S303. Based on similarity, determine the atomic-level loss function corresponding to the sample data.

[0153] It should be noted that the atomic loss function can be defined by the similarity between facial data and heart rate data, in order to reduce the distribution inconsistency between atomic features corresponding to the physiological state data of each sample.

[0154] Figure 8 This is a schematic diagram illustrating another loss function determination process provided in an embodiment of this application, as shown below. Figure 8 As shown, based on the atomic-level features corresponding to the physiological state data of each sample, the similarity loss between facial features and heart rate features, the similarity loss between audio features and heart rate features, and the similarity loss between facial features and audio features are determined respectively. Based on the similarity loss between facial features and heart rate features, the similarity loss between audio features and heart rate features, and the similarity loss between facial features and audio features, the atomic-level loss function corresponding to the sample data is determined.

[0155] As a feasible implementation method, given atomic-level features of heart rate characteristics Atomic features of facial features and atomic-level features of audio features Atomic loss function Lacl As shown in formula (16):

[0156]

[0157] It should be noted that the atomic-level loss function L acl To maximize L acl To match the real

[0158] as well as And try to minimize the value in other cases.

[0159] As a feasible implementation method, once the atomic loss function is determined, the loss function of the data analysis module can be determined based on the loss function corresponding to the sample data and the atomic loss function, as shown in formula (17). The loss function L of the data analysis module is calculated in the following way:

[0160] L = L ACL +L cce / ||L ACL / L cce || Formula (17)

[0161] Furthermore, the initial data analysis model is iteratively trained based on the loss function of the data analysis module until the iteration stopping condition is met, resulting in a fully trained data analysis model.

[0162] It should be noted that the iteration stopping condition can be the number of training iterations reaching a preset number, the training duration reaching a preset duration, or other possible iteration stopping conditions. This application does not limit the iteration stopping condition.

[0163] Furthermore, once the data analysis model is trained, multiple physiological state data are input into the data analysis model to obtain the adjustment values ​​of the air conditioner's operating parameters.

[0164] It should be noted that the data analysis model will also optimize the model parameters based on the user's actual feedback and sleep quality. Through continuous monitoring and learning, the data analysis model can gradually optimize the adjustment strategy to provide personalized and intelligent air conditioning sleep mode control.

[0165] In some embodiments, the atomic-level correlation features between multiple physiological state data can also be achieved through, for example... Figure 9 The flowchart shown is used to obtain the results. Figure 9 This application provides a schematic diagram of a data processing flow for determining atomic-level correlation features between multiple physiological state data, such as... Figure 9 As shown, audio features, heart rate features, and facial features are fused using an element-wise fusion method to obtain atomic-level correlation features between multiple physiological state data.

[0166] Among them, the element-wise fusion method includes multi-head attention computation, vector inner product, and multilayer perceptron operation.

[0167] In some embodiments, component-level correlation features between multiple physiological state data can also be achieved through, for example... Figure 10 The flowchart shown is used to obtain the results. Figure 10 This is another data processing flowchart provided in an embodiment of the present application, used to determine component-level correlation features between multiple physiological state data, such as... Figure 10 As shown, audio graph structure data, heart rate graph structure data, and facial graph structure data are fused using a graph-based fusion method to obtain component-level correlation features between multiple physiological state data.

[0168] Among them, graph-based fusion methods include performing convolution operations through graph convolutional neural networks and performing attention calculations through graph attention networks.

[0169] In some embodiments, Figure 9 It can also be specifically implemented as Figure 11 The process shown is as follows: Figure 11 This is a schematic diagram of another data processing flow provided in an embodiment of this application, such as... Figure 11 As shown, features are extracted from audio data, heart rate data, and facial data respectively. The features extracted from the audio data are input into the audio encoder for encoding, the features extracted from the heart rate data are input into the heart rate encoder for encoding, and the features extracted from the facial data are input into the visual encoder for encoding. The encoded data are then processed by a cross-attention module and a multi-head cross-attention module for feature enhancement to obtain audio features, heart rate features, and facial features.

[0170] Furthermore, the audio features, heart rate features, and facial features are multiplied by a vector dot product to achieve feature fusion, resulting in the fused feature P. ACM And based on P ACM Identify atomic-level correlation features between multiple physiological state data.

[0171] It should be noted that regarding the fused feature P ACM For details on how to obtain it, please refer to the above formula (8), which will not be repeated here.

[0172] In some embodiments, Figure 10 The steps described herein can also be achieved through Figure 12 The process shown is implemented as follows: Figure 12 This is a schematic diagram of another data processing flow provided in an embodiment of this application, such as... Figure 12As shown, audio graph structure data, heart rate graph structure data, and facial graph structure data are determined based on audio features, heart rate features, and facial features, respectively. These data are then input into a graph convolution module, which performs convolution operations on them to obtain the feature G of the audio graph structure data. a Features of heart rate graph structure data G t and facial image structure data G v .

[0173] Furthermore, G a G t and G v V was obtained by performing layer normalization respectively. t K v+a And Q v+a Then V t K v+a And Q v+a By sequentially inputting the cross-modal attention module and the feedforward neural network, component-level distribution constraints are obtained for audio graph structure data, heart rate graph structure data, and facial graph structure data.

[0174] Meanwhile, guided by the transfer cost matrix obtained during the optimal transmission process, G... a G t and G v By combining these data, multimodal features among multiple physiological state data can be obtained.

[0175] Furthermore, based on component-level distribution constraints and multimodal characteristics, component-level correlation features among multiple physiological state data are determined.

[0176] In this embodiment of the invention, electronic products can be divided into functional modules according to the above method examples. For example, each function can be divided into its own functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0177] When dividing each function into modules according to its corresponding function. Figure 13 This is a schematic diagram of another air conditioner provided in an embodiment of this application, as shown below. Figure 13 As shown, the air conditioner 1 may include: an acquisition module 201, an input / output module 202, and a processing module 203.

[0178] The acquisition module 201 is used to acquire multiple physiological state data of users in the area where the air conditioner is located when the air conditioner is running in sleep mode.

[0179] The input / output module 202 is used to input multiple physiological state data into the data analysis model, extract atomic-level correlation features between multiple physiological state data through the atomic-level feature extraction module in the data analysis model, extract component-level correlation features between multiple physiological state data through the component-level feature extraction module in the data analysis model, and obtain the air conditioner's operating parameter adjustment values ​​based on the atomic-level correlation features and component-level correlation features through the parameter prediction module in the data analysis model.

[0180] In some embodiments, the processing module 203 is used to adjust the operating parameters of the air conditioner based on the operating parameter adjustment value.

[0181] As a feasible implementation method, the processing module 203 is also used to perform feature enhancement through the cross-attention module after feature extraction of each physiological state data to obtain the physiological state features corresponding to each physiological state data.

[0182] As a feasible implementation method, the processing module 203 is also used to perform multi-head cross-attention operation on each physiological state feature through the multi-head cross-attention module to obtain the atomic-level feature corresponding to each physiological state feature.

[0183] As a feasible implementation method, the processing module 203 is also used to fuse the atomic-level features corresponding to multiple physiological state data to obtain atomic-level correlation features between multiple physiological state data.

[0184] As a feasible implementation method, the processing module 203 is also used to determine the graph structure data corresponding to each physiological state data, and to perform graph convolution operation on the graph structure data through the graph convolution module to obtain the graph data features corresponding to each physiological state data.

[0185] As a feasible implementation method, the processing module 203 is also used to determine the component-level association features between multiple physiological state data based on the graph data features corresponding to each physiological state data.

[0186] As a feasible implementation method, the acquisition module 201 is also used to acquire multiple sample data and the actual operating parameter adjustment value corresponding to each sample data; each sample data includes multiple sample physiological state data.

[0187] As a feasible implementation method, the input / output module 202 is also used to input sample data into the initial data analysis model to obtain the adjustment values ​​of the prediction operation parameters corresponding to the sample data.

[0188] As a feasible implementation method, the processing module 203 is also used to determine the loss function corresponding to the sample data based on the actual operating parameter adjustment value and the predicted operating parameter adjustment value corresponding to the sample data.

[0189] As a feasible implementation method, the processing module 203 is also used to iteratively train the initial data analysis model based on the loss function corresponding to the sample data to obtain the trained data analysis model.

[0190] As a feasible implementation method, the processing module 203 is also used to determine the atomic-level features corresponding to the physiological state data of each sample in the sample data.

[0191] As a feasible implementation method, the processing module 203 is also used to determine the similarity between multiple atomic-level features in the sample data.

[0192] As a feasible implementation method, the processing module 203 is also used to determine the atomic-level loss function corresponding to the sample data based on similarity.

[0193] As a feasible implementation method, the processing module 203 is also used to iteratively train the initial data analysis model based on the loss function corresponding to the sample data and the atomic loss function.

[0194] In some embodiments, the air conditioner 1 may further include a storage module 204 and a communication module 205.

[0195] In some embodiments, storage module 204 may be a memory. Communication module 205 may be a transceiver, transceiver circuit, or communication interface, etc.

[0196] In this configuration, when the processing module 203 is a processor, the storage module 204 is a memory, and the communication module 205 is a transceiver, the processor, transceiver, and memory can be connected via a bus. The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc.

[0197] This invention also provides a computer-readable storage medium, which includes computer-executed speech, such that when the computer-executed speech is run on a computer, the computer performs the method provided in the above embodiments.

[0198] This invention also provides a computer program product that can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, the computer program product can implement the methods provided in the above embodiments.

[0199] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using hardware, software, firmware, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more voice or code messages on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transfer of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers.

[0200] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0201] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely exemplary; for instance, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed between each other may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. Units described as separate components may or may not be physically separate; components shown as units may be one physical unit or multiple physical units, i.e., they may be located in one place or distributed in multiple different places. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0202] Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, in essence, 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. This software product is stored in a storage medium and includes several voice commands to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0203] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An air conditioner, characterized in that, include: The controller is configured as follows: When the air conditioner is running in sleep mode, multiple physiological state data of users in the area where the air conditioner is located are obtained; The multiple physiological state data are input into a data analysis model. The atomic-level feature extraction module in the data analysis model extracts the atomic-level correlation features between the multiple physiological state data. The component-level feature extraction module in the data analysis model extracts the component-level correlation features between the multiple physiological state data. The parameter prediction module in the data analysis model obtains the operating parameter adjustment value of the air conditioner based on the atomic-level correlation features and the component-level correlation features. The operating parameters of the air conditioner are adjusted based on the aforementioned operating parameter adjustment values.

2. The air conditioner according to claim 1, characterized in that, The atomic-level feature extraction module includes a cross-attention module and a multi-head cross-attention module; The controller extracts atomic-level correlation features between the multiple physiological state data through the atomic-level feature extraction module in the data analysis model, specifically configured as follows: After feature extraction for each physiological state data, feature enhancement is performed through the cross-attention module to obtain the physiological state features corresponding to each physiological state data. The multi-head cross-attention module performs multi-head cross-attention operation on each physiological state feature to obtain the atomic-level feature corresponding to each physiological state feature. The atomic-level features corresponding to the multiple physiological state data are fused to obtain the atomic-level correlation features between the multiple physiological state data.

3. The air conditioner according to claim 1, characterized in that, The component-level feature extraction module includes a graph convolution module; The controller extracts component-level correlation features between the multiple physiological state data through the component-level feature extraction module in the data analysis model, and is specifically configured as follows: Determine the graph structure data corresponding to each physiological state data, and perform graph convolution operation on the graph structure data through the graph convolution module to obtain the graph data features corresponding to each physiological state data. Based on the graph data features corresponding to each physiological state data, component-level association features among the multiple physiological state data are determined.

4. The air conditioner according to claim 1, characterized in that, The controller is also configured to: Acquire multiple sample data and the actual operating parameter adjustment values ​​corresponding to each sample data; each sample data includes multiple sample physiological state data. The sample data is input into the initial data analysis model to obtain the adjustment values ​​of the prediction operation parameters corresponding to the sample data; Based on the actual and predicted operating parameter adjustment values ​​corresponding to the sample data, determine the loss function corresponding to the sample data; Based on the loss function corresponding to the sample data, the initial data analysis model is iteratively trained to obtain the trained data analysis model.

5. The air conditioner according to claim 4, characterized in that, The controller iteratively trains the initial data analysis model based on the loss function corresponding to the sample data, specifically configured as follows: Determine the atomic-level features corresponding to the physiological state data of each sample in the sample data; Determine the similarity between multiple atomic-level features in the sample data; Based on the similarity, determine the atomic-level loss function corresponding to the sample data; Based on the loss function and atomic loss function corresponding to the sample data, the initial data analysis model is iteratively trained.

6. The air conditioner according to claim 1, characterized in that, The multiple physiological state data include audio data, heart rate data, and facial data; the operating parameters include at least one of the following: operating air volume, operating temperature, humidity level, air swing frequency, operating time, and fan speed.

7. A control method for an air conditioner, characterized in that, The method includes: When the air conditioner is running in sleep mode, multiple physiological state data of users in the area where the air conditioner is located are obtained; The multiple physiological state data are input into a data analysis model. The atomic-level feature extraction module in the data analysis model extracts the atomic-level correlation features between the multiple physiological state data. The component-level feature extraction module in the data analysis model extracts the component-level correlation features between the multiple physiological state data. The parameter prediction module in the data analysis model obtains the operating parameter adjustment value of the air conditioner based on the atomic-level correlation features and the component-level correlation features. The operating parameters of the air conditioner are adjusted based on the aforementioned operating parameter adjustment values.

8. The method according to claim 7, characterized in that, The atomic-level feature extraction module includes a cross-attention module and a multi-head cross-attention module; The extraction of atomic-level correlation features between the multiple physiological state data through the atomic-level feature extraction module in the data analysis model includes: After feature extraction for each physiological state data, feature enhancement is performed through the cross-attention module to obtain the physiological state features corresponding to each physiological state data. The multi-head cross-attention module performs multi-head cross-attention operation on each physiological state feature to obtain the atomic-level feature corresponding to each physiological state feature. The atomic-level features corresponding to the multiple physiological state data are fused to obtain the atomic-level correlation features between the multiple physiological state data.

9. The method according to claim 7, characterized in that, The component-level feature extraction module includes a graph convolution module; The step of extracting component-level correlation features between the multiple physiological state data through the component-level feature extraction module in the data analysis model includes: Determine the graph structure data corresponding to each physiological state data, and perform graph convolution operation on the graph structure data through the graph convolution module to obtain the graph data features corresponding to each physiological state data. Based on the graph data features corresponding to each physiological state data, component-level association features among the multiple physiological state data are determined.

10. The method according to claim 7, characterized in that, The multiple physiological state data include audio data, heart rate data, and facial data; the operating parameters include at least one of the following: operating air volume, operating temperature, humidity level, air swing frequency, operating time, and fan speed.

Images