Air conditioner and control method thereof

Abstract

The application provides an air conditioner and a control method thereof, comprising obtaining a deep model based on a soft attention mechanism, the deep model being used to obtain air conditioner control data in line with user habits according to environment data; obtaining real-time environment data, and obtaining real-time air conditioner control data according to the deep model and the real-time environment data. Then, the air conditioner adjusts the operation parameters of the air conditioner according to the real-time air conditioner control data, and adjusts the operation state of the air conditioner to an air conditioner state in line with the current user use habits. The application solves the problem that the air conditioner cannot automatically adjust the operation state according to user demand, improves the intelligence of the air conditioner, reduces the time spent by the user in manually adjusting the air conditioner, and improves the user experience.

Description

Technical Field

[0001] This invention relates to the field of automatic air conditioning control technology, and in particular to a control method for an air conditioner and an air conditioner. Background Technology

[0002] Currently, users typically rely on manually adjusting various parameters of traditional air conditioners to suit their different needs. This manual adjustment method requires users to spend time and effort setting the air conditioner's status, and users also need to repeatedly adjust the air conditioner's status based on changes in the surrounding environment and their comfort level. Finding the most suitable and comfortable air conditioner setting takes a significant amount of time; moreover, most air conditioners lack sufficient intelligence and cannot automatically adjust parameters to meet user needs; furthermore, air conditioners cannot be customized with a personalized automatic adjustment mechanism based on each user's usage patterns. Summary of the Invention

[0003] In view of the above problems, the present invention is proposed to provide an air conditioner and its control method that overcome or at least partially solve the above problems, thereby solving the problem that the air conditioner cannot automatically adjust its operating status according to user needs and achieving the goal of improving user experience.

[0004] Furthermore, this invention reduces the operation time for users to adjust the operating parameters of the air conditioner.

[0005] Furthermore, the present invention can customize a unique automatic operation mechanism based on the user's usage.

[0006] Specifically, the present invention provides a control method for an air conditioner, comprising:

[0007] A deep model based on a soft attention mechanism is obtained, which is used to obtain air conditioner control data that conforms to user habits based on environmental data;

[0008] Real-time environmental data is acquired, and real-time air conditioner control data is obtained based on the depth model and the real-time environmental data.

[0009] Optionally, the acquisition of a deep model based on a soft attention mechanism includes:

[0010] Acquire historical data, including historical environmental data and historical air conditioner control data of the user under each of the historical environmental data.

[0011] Using the historical environmental data as input and the corresponding historical air conditioner control data as output, a preset neural network model is trained to obtain the deep model.

[0012] Optionally, the acquisition of historical data includes:

[0013] Acquire initial historical data and filter the initial historical data according to the air conditioner usage time to obtain the historical data.

[0014] Optionally, the deep model includes an attention mechanism layer and a multi-layer perception mechanism. Obtaining real-time air conditioner control data based on the deep model and the real-time environmental data includes:

[0015] The attention mechanism layer is used to weight the real-time environmental data; and

[0016] The real-time air conditioner control data is obtained by using the multi-layer sensing mechanism and the weighted real-time environmental data.

[0017] Optionally, the historical environmental data includes one or more of the following: user perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity.

[0018] The historical air conditioner control data includes one or more of the following: air conditioner on / off status, user-set air conditioner mode, user-set temperature, user-set fan speed, continuous air conditioner operation time, and user's air conditioner usage time under different air conditioner states.

[0019] Optionally, the weighting of the real-time environmental data using the attention mechanism layer includes:

[0020] The real-time environmental data is linearly mapped to transform it into a multi-dimensional domain.

[0021] Obtain the attention distribution, and calculate the weighted average of the weighted real-time environmental data based on the attention distribution.

[0022] Optionally, the multi-layer sensing mechanism includes an input layer, a hidden layer, and an output layer. The step of using the multi-layer sensing mechanism to obtain the real-time air conditioner control data based on the weighted real-time environmental data includes:

[0023] The weighted real-time environmental data is obtained through the input layer;

[0024] Intermediate output data is obtained through the hidden layer based on the weighted real-time environmental data.

[0025] The intermediate output data is processed by the output layer to obtain the real-time air conditioner control data.

[0026] Optionally, the activation function of the hidden layer is either the sigmoid function or the tanh function.

[0027] Optionally, the step of acquiring real-time environmental data and obtaining real-time air conditioner control data based on the depth model and the real-time environmental data includes:

[0028] The real-time environmental data is acquired once at a set interval, and after each acquisition of the real-time environmental data, the real-time air conditioner control data is obtained based on the depth model and the real-time environmental data.

[0029] The present invention also provides an air conditioner, the air conditioner including a controller, the controller including a memory, a processor and a machine-executable program stored in the memory and running on the processor, and the processor implementing the control method described above when executing the machine-executable program.

[0030] This invention proposes a control method for an air conditioner that automatically adjusts operating parameters based on user habits. By inputting real-time environmental data into a deep model based on a soft attention mechanism, real-time air conditioner control data conforming to user habits is obtained. The soft attention mechanism assigns weights to the real-time environmental data, allowing the deep model to obtain control data that aligns with user habits based on the weights calculated by the soft attention mechanism. This invention solves the problem of air conditioners failing to automatically adjust their operating status according to user needs, improving the intelligence of the air conditioner. Furthermore, it reduces the time users spend manually adjusting the air conditioner, thereby enhancing the user experience.

[0031] Furthermore, this invention can train a unique automatic adjustment mechanism based on each user's usage to meet the user's personalized needs.

[0032] The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments of the invention in conjunction with the accompanying drawings. Attached Figure Description

[0033] The following sections will describe some specific embodiments of the invention in detail by way of example and not limitation, with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar parts or portions. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings:

[0034] Figure 1 This is a schematic flowchart of a control method for an air conditioner according to an embodiment of the present invention;

[0035] Figure 2 This is a schematic flowchart of generating a depth model according to an embodiment of the present invention;

[0036] Figure 3 This is a schematic flowchart of a control method for an air conditioner according to another embodiment of the present invention;

[0037] Figure 4 This is a schematic structural diagram of a controller in an air conditioner according to an embodiment of the present invention. Detailed Implementation

[0038] The following reference Figures 1 to 4 This invention describes an air conditioner and its control method according to embodiments of the present invention. In this description, it should be understood that 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 indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature, that is, include one or more of that feature. In the description of the present invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified. When a feature "includes or contains" one or more of the features it encompasses, unless otherwise specifically described, this indicates that other features are not excluded and may be further included.

[0039] In the description of this embodiment, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0040] Figure 1 This is a schematic flowchart of an air conditioner control method according to an embodiment of the present invention. The control method includes the following steps:

[0041] Step S101: Obtain a deep model based on a soft attention mechanism. This deep model is used to obtain air conditioner control data that conforms to user habits based on environmental data.

[0042] Step S102: Obtain real-time environmental data, and based on the depth model and real-time environmental data, obtain real-time air conditioner control data.

[0043] Specifically, when the air conditioner enters the adaptive function mode, it acquires a deep model based on a soft attention mechanism and real-time environmental data. Then, it inputs the real-time environmental data into the deep model, which obtains real-time air conditioner control data based on the real-time environmental data. Subsequently, the air conditioner adjusts its operating parameters based on the real-time air conditioner control data, adjusting the air conditioner's state to match the user's current usage habits.

[0044] This invention inputs real-time environmental data into a deep model based on a soft attention mechanism to obtain real-time air conditioner control data that conforms to user habits. The deep model uses the soft attention mechanism to calculate the weights of the real-time environmental data, and then obtains the air conditioner control data frequently used by the user based on the weighted real-time environmental data. This invention solves the problem that air conditioners cannot automatically adjust their operating status according to user needs, improving the intelligence of the air conditioner; on the other hand, it reduces the time and number of adjustments required by the user to manually adjust the air conditioner, thus improving the user experience.

[0045] Furthermore, the present invention can adjust the depth model according to the user's usage data, so that the air conditioner control data output by the depth model is consistent with or less than a preset ratio value of the air conditioner control data manually adjusted by the user when the air conditioner is in non-adaptive mode, and the adjustment time of the air conditioner is consistent with or less than the preset ratio value, thereby realizing "personal customization". Users no longer need to manually adjust the control data of the air conditioner, saving users the number of operations and improving the user's experience and comfort.

[0046] In some embodiments of the present invention, obtaining a deep model based on a soft attention mechanism includes the following steps:

[0047] Step S201: Obtain historical data, which includes historical environmental data and historical air conditioner control data of the user under each historical environmental data.

[0048] Step S202: Using historical environmental data as input and corresponding historical air conditioner control data as output, train the preset neural network model to obtain a deep model.

[0049] In this embodiment, historical environmental data includes user-perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity. Historical air conditioner control data includes at least the air conditioner on / off status, user-set air conditioner mode, user-set temperature, user-set fan speed, continuous air conditioner operation time, and user air conditioner usage time under different air conditioner states.

[0050] In this embodiment, the training of the preset neural network model includes two processes. The first process is to input historical environmental data into the preset neural network model to obtain preset air conditioner control data, and calculate the difference between the historical air conditioner control data and the preset air conditioner control data. The second process is to determine whether the difference is less than the set value of the corresponding data. When the difference is less than the set value of the corresponding data, the preset neural network model is trained into a deep model. When the difference is not less than the set value of the corresponding data, the first process is returned to be executed until the difference between the historical air conditioner control data and the preset air conditioner control data is less than the set value, at which point the preset neural network model is trained into a deep model that conforms to the user's usage habits.

[0051] This invention trains a preset neural network model into a deep model that conforms to user habits using historical data. The deep model can obtain air conditioner control data that conforms to user habits based on the input real-time environmental data, making the obtained air conditioner control data more accurate and more in line with user habits. It eliminates the need for manual operation by the user, improves the intelligence of the air conditioner, and enhances the user's comfort and experience.

[0052] In some embodiments of the present invention, historical data is obtained by acquiring initial historical data and filtering the initial historical data according to the air conditioner usage time to obtain historical data.

[0053] In this embodiment, the initially collected historical data includes all data on changes in the air conditioner's operating parameters, including data from multiple adjustments made by the user within a short period. Some of this data does not conform to the user's usage habits and needs to be filtered out, either by deleting it from the air conditioner's memory or from the cloud database. For example, if a data point shows that the proportion of time a user spent using the air conditioner in that state to the total continuous operating time of the air conditioner is less than a set value, then that data point will be deleted from the database or from the air conditioner's memory.

[0054] This invention obtains historical data by filtering initial historical data, making the deep model trained based on the historical data more accurate. This, in turn, makes the air conditioner control data obtained by inputting real-time environmental data into the deep model more accurate and more in line with user habits. It eliminates the need for manual operation by the user and reduces the number of operations required.

[0055] In some embodiments of the present invention, historical environmental data includes one or more of the following: user perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity.

[0056] Historical air conditioner control data includes one or more of the following: air conditioner on / off status, user-set air conditioner mode, user-set temperature, user-set fan speed, continuous air conditioner operation time, and user usage time of the air conditioner under different air conditioner states.

[0057] In this embodiment, some parameters of historical environmental data and historical air conditioner control data are shown. In the air conditioner control method of the present invention, historical environmental data and historical air conditioner control data include, but are not limited to, the above parameters. Depending on actual needs or geographical location differences, the parameters of historical environmental data and historical air conditioner control data can be changed, added, or deleted.

[0058] Accordingly, the real-time environmental data in step S102 above includes one or more of the user's perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity; the real-time air conditioner control data includes one or more of the air conditioner's on / off status, user-set air conditioner mode, user-set temperature, and user-set fan speed.

[0059] This invention trains a preset neural network model into a deep model based on historical data. The more historical data there is, the closer the deep model trained from the preset neural network model is to the user's usage habits. As a result, the real-time air conditioner control data obtained after inputting real-time environmental data into the deep model is more accurate, eliminating the need for manual adjustment by the user, improving user comfort, and reducing the number of operations required by the user.

[0060] In some embodiments of the present invention, the deep model includes an attention mechanism layer and a multi-layer perception mechanism; based on the deep model and real-time environmental data, real-time air conditioner control data is obtained, including:

[0061] An attention mechanism layer is used to weight real-time environmental data; and a multi-layer perception mechanism is used to obtain real-time air conditioner control data based on the weighted real-time environmental data.

[0062] In this embodiment, the deep model comprises two parts: a first attention mechanism layer and a second multi-layer perception mechanism. The first attention mechanism layer is used to perform weighted averaging on the input real-time environmental data, calculate the weights of each parameter of the real-time environmental data, and obtain weighted real-time environmental data. The weighted real-time environmental data is then used as the input to the multi-layer perception mechanism. The multi-layer perception mechanism processes and analyzes the weighted real-time environmental data to obtain real-time air conditioner control data that conforms to the user's usage habits. Based on the real-time air conditioner control data, the operating parameters of the air conditioner are changed, and the operating state of the air conditioner is adjusted to conform to the user's current usage habits.

[0063] The attention mechanism layer employed in this invention performs a weighted average of the input real-time environmental data, calculating the weights of each parameter in the real-time environmental data. This allows for more accurate air conditioner control data obtained from the deep model, making the air conditioner's operating state more closely aligned with user habits. The multi-layer perception mechanism used in this invention can process the weighted real-time environmental data into air conditioner control data that conforms to user habits, enabling the air conditioner to adjust its operating state based on this control data, thereby improving user comfort.

[0064] In some embodiments of the present invention, an attention mechanism layer is used to weight real-time environmental data, including:

[0065] Linear mapping is performed on real-time environmental data to transform it into a multi-dimensional domain.

[0066] Obtain the attention distribution and calculate a weighted average of the weighted real-time environmental data based on the attention distribution.

[0067] In this embodiment, real-time environmental data includes one or more of the following: user perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity. The attention mechanism layer includes a soft attention mechanism. The attention mechanism layer uses the principle of a soft attention mechanism to perform a weighted average of the input real-time environmental data, calculating the weights of different input real-time environmental data. The method used is as follows: when selecting real-time environmental data, instead of selecting one from n real-time environmental data points, a weighted average of the n input real-time environmental data points is calculated, and then the weighted average of the n input real-time environmental data points is input into the multilayer perception mechanism for calculation.

[0068] For example, when a scene comes into human view, attention is often drawn to key elements, such as dynamic points or striking colors, while the static elements may be temporarily ignored. Similarly, when searching for information about people in an image, people tend to focus on areas that match the person's features, ignoring those that don't. This demonstrates a reasonable and effective allocation of attention.

[0069] The air conditioner control method of the present invention employs a soft attention mechanism comprising three parts: calculating attention score, calculating attention distribution, and calculating weighted average. Common models for calculating attention score include additive model, dot product model, scaled dot product model, and bilinear model. The model used in the air conditioner control method of the present invention can be replaced according to actual needs to achieve the same output effect. Calculating attention distribution involves using the softmax function to convert the attention score to 0-1 to obtain the attention distribution of each real-time environmental data. Calculating weighted average involves using a weighted summation method to perform a weighted average of the output result of the attention distribution and the input real-time environmental data to obtain the weighted real-time environmental data.

[0070] The softmax function is:

[0071]

[0072] Among them, s i and s j To calculate the attention score, s i It is from s1 to s n The i-th output result in the sequence, s j It is from s1 to s n The j-th output result.

[0073] For example, a data set X = [x1, x2, x3, ..., x...] n ].

[0074] Where x1 represents the first type of environmental data, x2 represents the second type of environmental data, and x3 represents the third type of environmental data. n This is the Nth type of environmental data, which includes, but is not limited to, user perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity.

[0075] Step 1: Perform a linear mapping on X to transform it into a multidimensional domain, thus obtaining Y.

[0076] Step 2: Calculate the attention distribution of X, obtaining the attention distribution sequence a:

[0077] a = [a1, a2, a3, ..., a n ].

[0078] Where a1, a2, a3, ..., a n Let x1, x2, x3, ..., x n The corresponding attention distribution.

[0079] Step 3: Calculate the weighted average of the input information:

[0080]

[0081] x i It is the i-th data in data X, a i This is the i-th data point in the attention distribution sequence a. The real-time environment data is weighted and averaged based on the input real-time environment data and the obtained attention distribution to obtain the weighted real-time environment data.

[0082] This invention employs an attention mechanism layer to weight real-time environmental data, resulting in weighted real-time environmental data. Weighting the real-time environmental data allocates data according to weights, highlighting key data points. This ensures that the weighted real-time environmental data, when input into the multi-layer sensing mechanism, yields more accurate and user-friendly real-time air conditioner control data, ultimately improving user comfort.

[0083] In some embodiments of the present invention, the multi-layer sensing mechanism includes an input layer, a hidden layer, and an output layer; employing the multi-layer sensing mechanism, real-time air conditioner control data is obtained based on weighted real-time environmental data, including:

[0084] The weighted real-time environmental data is obtained through the input layer.

[0085] Intermediate output data is obtained through a hidden layer based on weighted real-time environmental data.

[0086] The intermediate output data is processed by the output layer to obtain real-time air conditioner control data.

[0087] In this embodiment, the output data of the first part of the attention mechanism layer is weighted real-time environmental data, including weighted user perceived temperature, weighted indoor temperature, weighted indoor humidity, weighted outdoor temperature, and weighted indoor humidity. The output data of the hidden layer is intermediate data output by the multilayer perception mechanism.

[0088] The second part of the deep model of this invention is a multi-layer perception mechanism, which consists of three parts: an input layer, a hidden layer, and an output layer. The input layer receives the weighted real-time environmental data from the first part's attention mechanism layer. The hidden layer processes and analyzes the weighted real-time environmental data to obtain intermediate output data that conforms to user habits. This intermediate output data cannot be directly transmitted to the air conditioner so that it can adjust its operating state accordingly. Instead, it needs to be converted into air conditioner control commands by the output layer, which then processes and integrates these commands into real-time air conditioner control data that conforms to user habits.

[0089] This invention processes weighted real-time environmental data through an input layer, a hidden layer, and an output layer in a multi-layer sensing mechanism to obtain real-time air conditioner control data that conforms to user habits. The air conditioner automatically adjusts its operating status based on this control data, eliminating the need for manual adjustment by the user, reducing the number of user operations, and improving the intelligence of the air conditioner.

[0090] In some embodiments of the present invention, the activation function of the hidden layer is the sigmoid function or the tanh function.

[0091] The hidden layer and the input layer are fully connected. Assuming the data of the input layer is represented by a vector Z, the output of the hidden layer is f(W1Z+b1), where W1 is the weight coefficient, b1 is the bias, and the activation function f can be the commonly used sigmoid function or tanh function.

[0092] The sigmoid function is:

[0093] The tanh function is:

[0094] In the above S(y), y and tanh(i) are both W1Z+b1, where W1 is the weight coefficient, Z is the vector, and b1 is the bias.

[0095] In this invention, intermediate output data of the hidden layer is calculated by using the sigmoid or tanh activation function. The intermediate output data cannot be directly transmitted to the air conditioner so that the air conditioner can directly adjust its operating status based on the intermediate output data. Instead, the intermediate output data needs to be converted into air conditioner control commands through the output layer. Then, the output layer processes and integrates the air conditioner control commands into real-time air conditioner control data that conforms to the user's usage habits.

[0096] The present invention employs an activation function in the hidden layer to make the air conditioner control data obtained from the hidden layer more accurate, more in line with user habits, and reduce the number of operations required by the user.

[0097] In some embodiments of the present invention, real-time environmental data is acquired, and real-time air conditioner control data is obtained based on the depth model and the real-time environmental data. This includes acquiring real-time environmental data once at a set interval, and obtaining real-time air conditioner control data based on the depth model and the real-time environmental data after each acquisition of real-time environmental data.

[0098] Specifically, the air conditioner acquires real-time environmental data at set intervals. After each acquisition, the real-time environmental data is input into a deep model. The deep model then sequentially uses an attention mechanism layer and a multi-layer perception mechanism to obtain real-time air conditioner control data and sends it to the air conditioner. The air conditioner then adjusts its operating parameters based on the real-time air conditioner control data to ensure that the air conditioner's operating status matches the user's usage habits.

[0099] Each time real-time environmental data is input into the deep model, the deep model is updated according to the real-time environmental data to obtain a deep model that conforms to the user's current usage habits. This, in turn, yields air conditioner control data that conforms to the user's current usage habits. The air conditioner then adjusts its operating parameters based on this control data to adjust its operating state to conform to the user's current usage habits.

[0100] Preferably, the duration is set to 5 minutes, but this can be changed according to the user's actual needs.

[0101] This invention obtains air conditioner control data by inputting real-time environmental data into a deep model at set intervals. Based on this control data, the air conditioner's operating state is adjusted to match the user's current usage habits, improving user comfort and reducing the number of operations required, thus enhancing the user experience. Furthermore, with increasing updates, the air conditioner can become increasingly intelligent, further improving its convenience, comfort, and intelligence.

[0102] In some optional embodiments of the present invention, the deep model is stored in the cloud, and historical environmental data, historical air conditioner control data, as well as the collected real-time environmental data and the air conditioner control data output by the deep model are stored in a database in the cloud.

[0103] In some optional embodiments of the present invention, the deep model, historical environmental data, historical air conditioner control data, and real-time collected environmental data and air conditioner control data output by the deep model are stored in the air conditioner's memory 520. When the storage quantity in the air conditioner's memory 520 reaches a set value, the air conditioner will filter and delete some data to maintain the weight ratio of the stored data and increase the available storage space.

[0104] In some optional embodiments of the present invention, the air conditioner may enter the adaptive function mode in ways including but not limited to the adaptive function mode button on the air conditioner's remote control, the operation button on the air conditioner's display screen, the mobile device connected to the air conditioner via Bluetooth or WiFi, and the APP operation button on the mobile device.

[0105] In some embodiments of the present invention, a public deep model trained based on mass data is stored in the cloud or the air conditioner's memory 520, so that users can use the air conditioner's adaptive function mode even when there is less data in the database.

[0106] In some embodiments of the present invention, reference is made to Figure 3 After the user turns on the air conditioner, the air conditioner will prompt the user whether to select the adaptive function mode.

[0107] If you choose no, you can manually adjust the air conditioner's operating parameters and adjust its operating status.

[0108] If you select "Yes," the air conditioner will enter adaptive function mode. And once in adaptive function mode:

[0109] First, the air conditioner will acquire real-time environmental data from the surrounding environment, including one or more of the following: the user's perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity.

[0110] The air conditioner then uploads the acquired environmental data to the cloud. The deep learning model stored in the cloud receives the environmental data, processes and analyzes it to obtain air conditioner control data that matches the user's usage habits. The cloud then sends the obtained air conditioner control data to the air conditioner.

[0111] Finally, the air conditioner adjusts its operating parameters and operating status based on the obtained air conditioner control data to conform to the user's usage habits, meet the user's needs, and thus improve the user experience.

[0112] The adaptive function mode refers to the mode in which the control method of the air conditioner of the present invention is executed. The cloud or memory 520 stores a database of deep models and user usage data. The air conditioner control data includes one or more of the following: air conditioner on / off status, user-set air conditioner mode, user-set temperature, and user-set fan speed.

[0113] In addition, this invention can automatically learn and match the appropriate air conditioning status according to the user's usage habits, reducing the complexity of user operation; on the other hand, as the number of times the user uses it increases, the air conditioner can become increasingly intelligent, improving the convenience, comfort and intelligence of the air conditioner.

[0114] The present invention also provides an air conditioner, the air conditioner including a controller 500, the controller 500 including a memory 520, a processor 510 and a machine-executable program 41 stored in the memory 520 and running on the processor 510, and the processor 510 implementing the control method described above when executing the machine-executable program 41.

[0115] In this embodiment, the air conditioner includes a controller 500, which controls various parameters of the air conditioner, including one or more of the following: air conditioner on / off status, user-set air conditioner mode, user-set temperature, and user-set fan speed.

[0116] In addition, the controller 500 can control the indoor temperature sensor to detect the indoor temperature, the outdoor temperature sensor to detect the outdoor temperature, the outdoor humidity sensor to detect the outdoor humidity, the indoor humidity sensor to detect the indoor ambient humidity, and the indoor infrared sensor to detect the user's body surface temperature.

[0117] The air conditioner uses the above detection methods to obtain one or more of the following in real time when it is in adaptive mode: indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity. It also adjusts one or more of the following: air conditioner on / off status, user-set air conditioning mode, user-set temperature, and user-set fan speed. This ensures that the air conditioner's operating parameters match the user's usage habits and improves the user experience.

[0118] The controller 500 includes a memory 520 and a processor 510. The memory 520 stores at least the air conditioner switch status, user-set air conditioner mode, user-set temperature, user-set fan speed, indoor temperature, indoor humidity, outdoor temperature, outdoor humidity, and a machine-executable program 41 corresponding to the above control method. When the machine-executable program 41 running on the processor 510 implements the control method described in any of the above embodiments, it ensures that the air conditioner is in an operating state that conforms to the user's usage habits.

[0119] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0120] For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection (electronic device) having one or more wires, a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, a computer-readable medium can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0121] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0122] Therefore, those skilled in the art should recognize that although numerous exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications conforming to the principles of the present invention can be directly determined or derived from the disclosure of the present invention without departing from the spirit and scope of the invention. Thus, the scope of the present invention should be understood and construed as covering all such other variations or modifications.

Claims

1. A control method for an air conditioner, characterized in that, include: A deep model based on a soft attention mechanism is obtained, which is used to obtain air conditioner control data that conforms to user habits based on environmental data; Acquire real-time environmental data, and obtain real-time air conditioner control data based on the depth model and the real-time environmental data; The deep model includes an attention mechanism layer and a multi-layer perception mechanism. The process of obtaining real-time air conditioner control data based on the deep model and the real-time environmental data includes: The attention mechanism layer is used to weight the real-time environmental data; and Using the aforementioned multi-layer sensing mechanism, the real-time air conditioner control data is obtained based on the weighted real-time environmental data; The weighting of the real-time environmental data using the attention mechanism layer includes: The real-time environmental data is linearly mapped to transform it into a multi-dimensional domain. Obtain the attention distribution, and calculate the weighted average of the weighted real-time environmental data based on the attention distribution.

2. The control method according to claim 1, characterized in that, The acquisition of deep models based on soft attention mechanisms includes: Acquire historical data, including historical environmental data and historical air conditioner control data of the user under each of the historical environmental data. Using the historical environmental data as input and the corresponding historical air conditioner control data as output, a preset neural network model is trained to obtain the deep model.

3. The control method according to claim 2, characterized in that, The acquisition of historical data includes: Acquire initial historical data and filter the initial historical data according to the air conditioner usage time to obtain the historical data.

4. The control method according to claim 2, characterized in that, The historical environmental data includes one or more of the following: user perceived temperature, indoor temperature, indoor humidity, outdoor temperature, and outdoor humidity. The historical air conditioner control data includes one or more of the following: air conditioner on / off status, user-set air conditioner mode, user-set temperature, user-set fan speed, continuous air conditioner operation time, and user's air conditioner usage time under different air conditioner states.

5. The control method according to claim 1, characterized in that, The multi-layer sensing mechanism includes an input layer, a hidden layer, and an output layer. The method of obtaining real-time air conditioner control data based on the weighted real-time environmental data using this multi-layer sensing mechanism includes: The weighted real-time environmental data is obtained through the input layer; Intermediate output data is obtained through the hidden layer based on the weighted real-time environmental data. The intermediate output data is processed by the output layer to obtain the real-time air conditioner control data.

6. The control method according to claim 5, characterized in that, The activation function of the hidden layer is either the sigmoid function or the tanh function.

7. The control method according to claim 1, characterized in that, The step of acquiring real-time environmental data and obtaining real-time air conditioner control data based on the depth model and the real-time environmental data includes: The real-time environmental data is acquired once at a set interval, and after each acquisition of the real-time environmental data, the real-time air conditioner control data is obtained based on the depth model and the real-time environmental data.

8. An air conditioner, characterized in that, include: A controller, comprising a memory, a processor, and a machine-executable program stored in the memory and running on the processor, wherein the processor, when executing the machine-executable program, implements the control method according to any one of claims 1 to 7.

Images