An intelligent air conditioner adjusting method, device and electronic equipment

Abstract

The application discloses a kind of intelligent air conditioner adjusting method, device and electronic equipment, it is related to the field of intelligent air conditioner.In the method, obtain multiple indoor images;The feature extraction is carried out to each indoor image, obtain the personnel activity state corresponding to each indoor image and the time point corresponding to personnel activity state, personnel activity state includes going out state, sleep state, long sitting state and activity state;Personnel activity time sequence is input into preset activity state prediction model, and prediction activity state is obtained;According to preset operation rule, determine the air conditioner operating data corresponding to prediction activity state;Obtain outdoor environment parameter;According to outdoor environment parameter, adjust air conditioner operating data, and obtain target air conditioner operating data;Determine to run air conditioner with target air conditioner operating data.The technical scheme provided in the application is implemented, and air conditioner operating data is automatically adjusted according to user activity state and outdoor environment parameter, to reduce energy waste.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to an intelligent air conditioner adjusting method and device and electronic equipment. BACKGROUND

[0002] With global climate change and the improvement of people's living standards, air conditioners have become an indispensable device in modern life, especially in hot or cold seasons. Air conditioners not only provide a comfortable indoor environment, but also affect people's health and work efficiency. According to statistics, the utilization rate of air conditioning systems in many countries' homes and commercial buildings is very high, especially in summer, air conditioners almost run all day long, consuming a large amount of power resources. Therefore, improving the intelligent level of air conditioning systems and realizing efficient and energy-saving operation have become an important issue in modern society.

[0003] At present, the operation of most traditional air conditioning systems depends on the manual setting of users, which has obvious shortcomings in actual use. For example, users often forget to adjust the air conditioning settings when they go out, resulting in the air conditioner continuing to run in the absence of people, which will cause waste of energy.

[0004] Therefore, an intelligent air conditioner adjusting method, device and electronic equipment are urgently needed. SUMMARY

[0005] The present application provides an intelligent air conditioner adjusting method, device and electronic equipment, which automatically adjusts air conditioner operating data according to user activity state and outdoor environment parameters, reducing energy waste.

[0006] In a first aspect of the present application, an intelligent air conditioner adjusting method is provided, which comprises: acquiring multiple indoor images; performing feature extraction on each of the indoor images to obtain a personnel activity state corresponding to each of the indoor images and a time point corresponding to the personnel activity state, the personnel activity state including an outing state, a sleep state, a sedentary state and an active state; constructing a personnel activity time sequence according to the personnel activity state and the time point corresponding to the personnel activity state; inputting the personnel activity time sequence into a preset activity state prediction model to obtain a predicted activity state; determining air conditioner operating data corresponding to the predicted activity state according to a preset operating rule, the air conditioner operating data including an air conditioner operating mode and an air conditioner target temperature; acquiring outdoor environment parameters, the outdoor environment parameters including outdoor temperature and outdoor humidity; adjusting the air conditioner operating data according to the outdoor environment parameters to obtain target air conditioner operating data; and determining to run an air conditioner with the target air conditioner operating data.

[0007] By adopting the technical scheme, the indoor image is acquired, the personnel activity state and the corresponding time point are obtained by performing feature extraction on the indoor image, the personnel activity time sequence is constructed, the future personnel activity state is predicted by using the preset activity state prediction model, and the air conditioner running mode and the air conditioner target temperature are determined based on the predicted activity state and the preset running rule. Meanwhile, the outdoor environment parameter is acquired, which is used for adjusting and optimizing the determined air conditioner running data, and finally the air conditioner is controlled to run with the optimized target air conditioner running data. The scheme can accurately predict the activity state of the user at the future time by analyzing the historical activity state of the user, and adjust the air conditioner in advance, so that the air conditioner runs more intelligently and humanly, and the running parameters are more in line with the actual needs and preferences of the user. Moreover, the running data can be adaptively adjusted according to the outdoor environment parameter, which guarantees the user experience while taking into account energy saving and environmental protection. When the user state changes, the air conditioner running data can also be automatically adjusted, avoiding the trouble of frequent manual adjustment of the air conditioner, improving the user comfort, and improving the energy utilization efficiency and reducing energy waste through intelligent optimization of the air conditioner running strategy.

[0008] Optionally, the feature extraction on each indoor image to obtain the personnel activity state corresponding to each indoor image specifically includes: performing a preprocessing operation on the indoor image to obtain a target indoor image, the preprocessing operation including image size normalization, image noise removal, and image enhancement; detecting a target person from the target indoor image by using a target detection algorithm; extracting human body key points of the target person, and calculating a set relationship between each human body key point to obtain a posture feature, the posture feature including a torso angle feature and a limb angle feature; and determining the personnel activity state according to the torso angle feature and the limb angle feature.

[0009] By adopting the technical scheme, the indoor image is acquired, the personnel activity state and the corresponding time point are obtained by performing feature extraction on the indoor image, the personnel activity time sequence is constructed, the future personnel activity state is predicted by using the preset activity state prediction model, and the air conditioner running mode and the air conditioner target temperature are determined based on the predicted activity state and the preset running rule. Meanwhile, the outdoor environment parameter is acquired, which is used for adjusting and optimizing the determined air conditioner running data, and finally the air conditioner is controlled to run with the optimized target air conditioner running data. The scheme can accurately predict the activity state of the user at the future time by analyzing the historical activity state of the user, and adjust the air conditioner in advance, so that the air conditioner runs more intelligently and humanly, and the running parameters are more in line with the actual needs and preferences of the user. Moreover, the running data can be adaptively adjusted according to the outdoor environment parameter, which guarantees the user experience while taking into account energy saving and environmental protection. When the user state changes, the air conditioner running data can also be automatically adjusted, avoiding the trouble of frequent manual adjustment of the air conditioner, improving the user comfort, and improving the energy utilization efficiency and reducing energy waste through intelligent optimization of the air conditioner running strategy.

[0010] Optionally, the determining the activity state of the person according to the trunk angle feature and the limb angle feature specifically comprises: when the trunk angle feature represents that the trunk of the target person is perpendicular to the ground and the limb angle feature represents that the limbs of the target person are in an extended state, determining that the activity state of the person is an active state; when the trunk angle feature represents that the trunk of the target person is perpendicular to the ground and the limb angle feature represents that the limbs of the target person are in a bent state, determining that the activity state of the person is a sedentary state; when the trunk angle feature represents that the trunk of the target person is at a preset angle with the ground and the limb angle feature represents that the limbs of the target person are in a horizontal placement state, determining that the activity state of the person is a sleep state; and when the target person is not detected from the target indoor image within a preset time period, determining that the activity state of the person is an outing state.

[0011] By adopting the above technical solution, when the trunk is perpendicular to the ground and the limbs are extended, it is an active state, when the trunk is perpendicular but the limbs are bent, it is a sedentary state, when the trunk is at a certain angle with the ground and the limbs are horizontally placed, it is a sleep state, and if a person cannot be detected within a period of time, it is determined to be a leaving state. This method of activity state discrimination according to the human body posture and angle feature is intuitive and efficient, and fully utilizes the activity information contained in the human body posture. Through the judgment rule of the angle relationship between the key points, the personnel activity state can be quickly classified.

[0012] Optionally, before the personnel activity time sequence is input into a preset activity state prediction model to obtain a predicted activity state, the method further comprises: obtaining time sequence sample data, the time sequence sample data comprising a historical activity state sequence and a duration corresponding to the historical activity state sequence; constructing a seq2seq model, the seq2seq model comprising an encoder and a decoder, the encoder being configured to encode the input historical activity state sequence, and the decoder being configured to predict the activity state of the person at the next time; and training the seq2seq model by using the time sequence sample data to obtain the preset activity state prediction model.

[0013] By adopting the above technical solution, the historical activity state sequence and its duration are obtained as training samples, and then a seq2seq model comprising an encoder and a decoder is constructed to train the time sequence sample data, so that the seq2seq model can predict the most possible activity state of the person at the next time given the historical activity state sequence.

[0014] Optionally, the determining the air conditioner running data corresponding to the predicted activity state according to the preset running rule comprises: when the predicted activity state is an outing state, determining the air conditioner running mode as a closed mode; when the predicted activity state is a sleep state, determining the air conditioner running mode as a sleep mode and setting the air conditioner target temperature as a preset sleep temperature; when the predicted activity state is a sedentary state, determining the air conditioner running mode as a cooling mode or a heating mode and setting the air conditioner target temperature as a preset comfortable temperature; and when the predicted activity state is an active state, determining the air conditioner running mode as a blowing mode.

[0015] By using the above technical solution, when the predicted activity state is an outing state, the air conditioner running mode is a closed mode to save energy; when the predicted activity state is a sleep state, the air conditioner enters a sleep mode and a preset sleep temperature is set; when the predicted activity state is a sedentary state, a cooling or heating mode is started and a preset comfortable temperature is adjusted; and when the predicted activity state is an active state, a blowing mode is switched to keep air circulating. The air conditioner control scheme based on the user activity state fully considers the air conditioner use demand of the user in different activity states, automatically switches the working mode and temperature parameters of the air conditioner according to the change of the predicted activity state, and has better user experience.

[0016] Optionally, the adjusting the air conditioner running data according to the outdoor environment parameter to obtain target air conditioner running data comprises: acquiring historical outdoor environment parameters and corresponding user temperature adjustment behavior data, the user temperature adjustment behavior data comprising a user temperature adjustment time, a user temperature adjustment amplitude and a target temperature after temperature adjustment; constructing a user temperature adjustment behavior prediction model, the user temperature adjustment behavior prediction model being used to predict user temperature adjustment behavior according to a current outdoor environment parameter; inputting the outdoor temperature and the outdoor humidity into the user temperature adjustment behavior prediction model to obtain a predicted temperature adjustment amplitude; and adjusting the air conditioner target temperature according to the predicted temperature adjustment amplitude to obtain the target air conditioner running data.

[0017] By adopting the technical solution, the historical outdoor environment parameters and the user temperature adjustment behavior data are acquired, a user temperature adjustment behavior prediction model is constructed, the current outdoor temperature and outdoor humidity are input into the user temperature adjustment behavior prediction model to obtain the predicted user temperature adjustment behavior, and the air conditioner target temperature is corrected by using the predicted user temperature adjustment behavior, so that the operation parameters are closer to the real needs of the user. The scheme utilizes the machine learning technology, learns the correlation between the historical temperature adjustment behavior of the user and the historical outdoor environment parameters, predicts the temperature adjustment behavior that the user may take under the current outdoor environment condition, and realizes the dynamic self-adaptation of the air conditioner parameters and the outdoor environment. Compared with the fixed set air conditioner parameters, the adaptive adjustment mode can effectively cope with the change of the outdoor temperature and humidity, and when the outdoor environment changes sharply, such as the temperature rises or drops sharply, the user adjustment preference is predicted by the user temperature adjustment behavior prediction model, the air conditioner is adjusted in advance, and the use experience caused by the change of the environment is avoided. Meanwhile, the environment and behavior data are accumulated for a long time, the user temperature adjustment behavior prediction model is iteratively updated, the air conditioner control strategy can be continuously optimized, and the habits and preferences of the user are more suitable.

[0018] Optionally, the constructing the user temperature adjustment behavior prediction model specifically includes: taking the historical outdoor environment parameters as sample features and the user temperature adjustment behavior data as sample labels to construct a training sample set; selecting a preset machine learning algorithm, training by using the training sample set to obtain a machine learning model for predicting the user temperature adjustment behavior, and taking the machine learning model as the user temperature adjustment behavior prediction model.

[0019] By adopting the technical solution, the historical outdoor environment parameters are taken as sample features, the user temperature adjustment behavior data are taken as sample labels, a training sample set is constructed, a machine learning algorithm is applied for training to obtain a user temperature adjustment behavior prediction model, the mapping relationship between a large number of outdoor environment parameters and user temperature adjustment behaviors is learned, the temperature adjustment preference of the user under different environment conditions can be described, the user habits can be learned for self-evolution, and the indoor environment can be adjusted in a more intelligent and personalized manner.

[0020] In a second aspect of the present application, an intelligent air conditioner adjusting device is provided, comprising: an acquisition module and a processing module, wherein: the acquisition module is configured to acquire a plurality of indoor images; the processing module is configured to perform feature extraction on each of the indoor images to obtain a personnel activity state corresponding to each of the indoor images and a time point corresponding to the personnel activity state, the personnel activity state including an outing state, a sleep state, a sedentary state, and an active state; the processing module is further configured to construct a personnel activity time sequence according to the personnel activity state and the time point corresponding to the personnel activity state; the processing module is further configured to input the personnel activity time sequence into a preset activity state prediction model to obtain a predicted activity state; the processing module is further configured to determine air conditioner operation data corresponding to the predicted activity state according to a preset operation rule, the air conditioner operation data including an air conditioner operation mode and an air conditioner target temperature; the acquisition module is further configured to acquire outdoor environment parameters, the outdoor environment parameters including an outdoor temperature and an outdoor humidity; the processing module is further configured to adjust the air conditioner operation data according to the outdoor environment parameters to obtain target air conditioner operation data; and the processing module is further configured to determine an air conditioner to operate in the target air conditioner operation data.

[0021] In a third aspect of the present application, an electronic device is provided, comprising a processor, a memory, a user interface, and a network interface, the memory is configured to store instructions, the user interface and the network interface are both configured to communicate with other devices, and the processor is configured to execute the instructions stored in the memory to enable the electronic device to perform the method according to any one of the above aspects.

[0022] In a fourth aspect of the present application, a computer-readable storage medium is provided, which stores instructions, when the instructions are executed, perform the method according to any one of the above aspects.

[0023] In summary, the one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0024] 1. By acquiring indoor images, the personnel activity state and the corresponding time point are obtained by feature extraction of the indoor images, the personnel activity time sequence is constructed, the future personnel activity state is predicted by using the preset activity state prediction model, and the air conditioner running mode and the air conditioner target temperature are determined based on the predicted activity state and the preset running rule. At the same time, the outdoor environment parameters are obtained, which are used to adjust and optimize the determined air conditioner running data, and finally the air conditioner is controlled to run with the optimized target air conditioner running data. This scheme can accurately predict the future activity state of the user by analyzing the user's historical activity state, and adjust the air conditioner in advance, so that the air conditioner runs more intelligently and humanely, and the running parameters are more in line with the actual needs and preferences of the user. And also can adjust the running data according to the outdoor environment parameters, while ensuring the user experience, and taking into account energy saving and environmental protection. When the user's state changes, the air conditioner running data can also be automatically adjusted, avoiding the user's frequent manual adjustment of the air conditioner. While improving user comfort, the energy utilization efficiency is improved and energy waste is reduced through intelligent optimization of the air conditioner running strategy.

[0025] 2. The acquired indoor images are preprocessed, including size normalization, image noise removal and image enhancement, to optimize the image quality for subsequent analysis. Then a target detection algorithm is used to detect the target personnel in the image, and then the human key points are extracted to calculate the posture features. The human trunk angle features and limb angle features are analyzed to determine different personnel activity states. This method of locating the human body first and then analyzing the key point posture can accurately and efficiently determine different activity states, providing reliable state determination basis for subsequent activity prediction and air conditioner control.

[0026] 3. When the predicted activity state is an outing state, the air conditioner running mode is closed to save energy; when the predicted activity state is a sleep state, the air conditioner enters a sleep mode and sets a preset sleep temperature; when the predicted activity state is a sedentary state, the cooling or heating mode is started and adjusted to a preset comfortable temperature; when the predicted activity state is an active state, the air conditioner is switched to a blowing mode to keep the air circulating. This air conditioner control scheme based on user activity state fully considers the air conditioner usage needs of users in different activity states, automatically switches the working mode and temperature parameters of the air conditioner according to the change of the predicted activity state, and has better user experience. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 is a flowchart of an intelligent air conditioner adjustment method disclosed by an embodiment of the present application;

[0028] Figure 2 is a module schematic diagram of an intelligent air conditioner adjustment device disclosed by an embodiment of the present application;

[0029] Figure 3Fig. 1 is a structural schematic diagram of an electronic device according to an embodiment of the present application.

[0030] Marker explanation: 201, acquisition module; 202, processing module; 300, electronic device; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION

[0031] In order for those skilled in the art to better understand the technical solutions in the specification, the technical solutions in the specification will be clearly and completely described below in conjunction with the drawings in the embodiments of the specification. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments.

[0032] In the description of the embodiments of the present application, the words such as "for example" or "for instance" are used to mean an example, illustration, or description. Any embodiment or design solution described as "for example" or "for instance" in the embodiments of the present application should not be interpreted as more preferred or more advantageous than other embodiments or design solutions. Rather, the words such as "for example" or "for instance" are intended to present the relevant concept in a specific manner.

[0033] In the description of the embodiments of the present application, the term "a plurality of" means two or more. For example, a plurality of systems means two or more systems, and a plurality of screen terminals means two or more screen terminals. In addition, the terms "first" and "second" are used for description purposes only, and should not be interpreted or implied to indicate or imply relative importance or implicitly indicate the indicated technical features. Therefore, the features defined with "first" and "second" can explicitly or implicitly include one or more of the features. The terms "include", "contain", "have" and their variants mean "include but are not limited to", unless otherwise specifically emphasized.

[0034] The present application provides a smart air conditioner adjustment method, referring to Figure 1 , Figure 1 Fig. 1 is a flowchart of a smart air conditioner adjustment method according to an embodiment of the present application. The method is applied to a server, and includes steps S101 to S108, which are as follows:

[0035] Step S101: Acquire a plurality of indoor images.

[0036] In step S101, the server periodically or on demand obtains real-time image data from the camera by establishing a communication connection with the indoor camera. The camera can be a common visible light camera or a special camera with infrared or depth sensing capability to adapt to different environmental lighting conditions and data acquisition requirements. During data acquisition, the server establishes a connection through the communication interface (such as USB, HDMI, network interface) with the camera, and sends image acquisition instructions to the camera according to the preset acquisition frequency (such as every 5 seconds). After receiving the instruction, the camera starts shooting images and transmits image data to the server.

[0037] In the intelligent air conditioning regulation system, protecting user privacy is a crucial task. In order to minimize the impact on user privacy during the acquisition of indoor images, the server can take the following improvement measures: using low-resolution or blurred cameras, the server can choose a camera with lower resolution (such as below 300,000 pixels), or perform blur processing on the original image during image acquisition, so that sensitive information such as personnel faces and identity features in the image cannot be clearly identified, but enough information can be retained for behavior state analysis.

[0038] Step S102: feature extraction is performed on each indoor image to obtain the personnel activity state corresponding to each indoor image and the time point corresponding to the personnel activity state, and the personnel activity state includes the state of going out, the state of sleeping, the state of sitting for a long time and the state of activity.

[0039] In step S102, the indoor image is preprocessed to obtain a target indoor image, the preprocessing operation includes image size normalization, image noise removal and image enhancement; a target detection algorithm is used to detect target personnel from the target indoor image; the body key points of the target personnel are extracted, and the set relationship between each body key point is calculated to obtain the posture feature, the posture feature includes the trunk angle feature and the limb angle feature; the personnel activity state is determined according to the trunk angle feature and the limb angle feature.

[0040] Specifically, after the server receives the indoor image, it first performs a series of preprocessing operations on the indoor image to improve the accuracy and efficiency of subsequent feature extraction. The preprocessing operations include image size normalization, image noise removal, and image enhancement; among them, image size normalization is to uniformly scale or crop indoor images of different resolutions and sizes to a specified standard size (such as 640x480 pixels) to facilitate subsequent feature extraction; image noise removal uses image filtering algorithms (such as median filtering, Gaussian filtering) to denoise the indoor image, reducing noise, burrs and other interference information in the image, and improving image quality; image enhancement uses image enhancement techniques (such as histogram equalization, contrast stretching) to optimize the image, making the target personnel and key features in the image more prominent and clear. After the preprocessing step, the server can obtain the target indoor image, and the server detects the target personnel from the image to locate and extract the area where the personnel are located. The target detection algorithm uses the YOLO algorithm based on deep learning, which learns the visual features and spatial distribution of the human body by training a large number of human image data, so as to quickly and accurately detect personnel targets in new images. The YOLO algorithm divides the target indoor image into several grids, each grid predicts several bounding boxes, and then deletes redundant bounding boxes through non-maximum suppression to finally determine the position of the target personnel.

[0041] For the detected target personnel area, the server further extracts the key point information of the human body, such as the position and posture of the head, torso, and limb parts. After obtaining the target, the server further extracts the key point features of the human body. Human key points usually refer to some joint points and end points of the human body, such as the top of the head, the neck, the shoulders, the elbows, the wrists, the hips, the knees, and the ankles. The server uses the OpenPose human key point extraction algorithm, which learns the spatial relationship of human skeleton structure and joint points by training human posture data sets, so as to accurately locate the coordinates of the key points in the target personnel image.

[0042] After extracting the human key points, the server needs to calculate the set relationship between each key point to obtain the human posture features. Here, two types of features are calculated: torso angle features and limb angle features. The torso angle feature reflects the inclination of the human torso, which can be obtained by calculating the included angle between the line connecting the neck key point and the hip key point and the vertical direction. The limb angle feature reflects the bending degree of the human limbs, which can be obtained by calculating the included angle between the lines connecting the upper arm, lower arm, thigh, and calf joint points. For example, the included angle of the two vectors formed by the shoulder key point, elbow key point, and wrist key point can be calculated as the angle feature of the upper arm. By calculating these angle features, the server can obtain a quantitative description of the human posture. The server judges the personnel activity state according to the torso angle feature and the limb angle feature.

[0043] In a possible implementation, the personnel activity state is determined according to the torso angle feature and the limb angle feature, and specifically includes: when the torso angle feature indicates that the torso of the target personnel is perpendicular to the ground and the limb angle feature indicates that the limbs of the target personnel are in an extended state, determining that the personnel activity state is an active state; when the torso angle feature indicates that the torso of the target personnel is perpendicular to the ground and the limb angle feature indicates that the limbs of the target personnel are in a bent state, determining that the personnel activity state is a sedentary state; when the torso angle feature indicates that the torso of the target personnel is at a preset angle with the ground and the limb angle feature indicates that the limbs of the target personnel are all in a horizontal placement state, determining that the personnel activity state is a sleep state; and when the target personnel is not detected from the target indoor image within a preset time period, determining that the personnel activity state is an outing state.

[0044] Specifically, the personnel activity state includes an active state, a sedentary state, a sleep state, and an outing state; for the judgment of the active state, the server analyzes the torso angle feature of the target personnel, and judges whether the torso is perpendicular to the ground. The server realizes this by calculating the included angle between the torso central axis and the ground normal vector, and when the included angle is close to 90 degrees (such as between 85 degrees and 95 degrees), it can be considered that the torso is perpendicular to the ground. Meanwhile, the server analyzes the limb angle feature of the target personnel, and judges whether the limbs are in an extended state. The server realizes this by calculating the included angle between the connecting line between the limb joints and the torso central axis, and when the included angle is close to 180 degrees (such as between 170 degrees and 190 degrees), it can be considered that the limbs are in an extended state. When the torso is perpendicular to the ground and the limbs are in an extended state, the server can determine that the personnel is currently in an active state, such as walking.

[0045] For the judgment of the sedentary state, the server analyzes the torso angle feature of the target personnel, and judges whether the torso is perpendicular to the ground. Similar to the judgment of the active state, when the included angle between the torso central axis and the ground normal vector is close to 90 degrees (such as between 85 degrees and 95 degrees), it can be considered that the torso is perpendicular to the ground. The server analyzes the limb angle feature of the target personnel, and judges whether the limbs are in a bent state. The server realizes this by calculating the included angle between the connecting line between the limb joints and the torso central axis, and when the included angle is obviously less than 180 degrees (such as between 90 degrees and 150 degrees), it can be considered that the limbs are in a bent state. When the torso is perpendicular to the ground and the limbs are in a bent state, the server can determine that the personnel is currently in a sedentary state, such as performing sedentary activities such as office work, study, and watching TV.

[0046] For the judgment of sleep state, the server analyzes the trunk angle features of the target person, judges whether it presents a preset angle with the ground, and realizes it by calculating the included angle between the trunk center axis and the ground normal vector. When the included angle is close to 0 degrees or 180 degrees (such as-10 degrees to 10 degrees, or 170 degrees to 190 degrees), it can be considered that the trunk is in a horizontal state with the ground. The server analyzes the limb angle features of the target person, judges whether they are all in a horizontal placement state, and realizes it by calculating the included angle between the connecting line between the limb joints and the ground normal vector. When the included angle is close to 0 degrees or 180 degrees (such as-10 degrees to 10 degrees, or 170 degrees to 190 degrees), it can be considered that the limbs are in a horizontal placement state. When the trunk is in a horizontal state with the ground and the limbs are all in a horizontal placement state, the server can judge that the person is currently in a sleep state.

[0047] For the judgment of going out state, the server will continuously monitor the target indoor image in a preset time period (such as 30 minutes), and judge whether there is a person appearing through the target person detection algorithm. If the target person cannot be detected from the image in the preset time period (such as continuously for 30 minutes), the server can judge that the person is currently in a going out state.

[0048] Step S103: According to the personnel activity state and the time point corresponding to the personnel activity state, a personnel activity time sequence is constructed.

[0049] In step S103, the server first reads the personnel activity state recognition results in a period of time (such as the past 1 week) from the database, including the activity state label (such as "active state", "sleep state", etc.) and timestamp information (such as "2023-05-20 08:30:00") corresponding to each time point. These data are usually arranged in time order with time as the primary key and activity state as the value. The server constructs a complete personnel activity time sequence according to the chronological order of the timestamp. The personnel activity time sequence is recorded in fixed time intervals (such as 1 minute), which records the activity state information corresponding to each time point, forming a continuous state change sequence, such as "... active state, active state, sedentary state, sedentary state, sleep state, sleep state...".

[0050] Step S104: Input the personnel activity time sequence into a preset activity state prediction model to obtain a predicted activity state.

[0051] Before step S104, the method further comprises: obtaining time series sample data, the time series sample data comprising a historical activity state sequence and a duration corresponding to the historical activity state sequence; constructing a seq2seq model, the seq2seq model comprising an encoder and a decoder, the encoder being configured to encode the input historical activity state sequence, and the decoder being configured to predict a next time personnel activity state; and training the seq2seq model by using the time series sample data to obtain a preset activity state prediction model.

[0052] Specifically, the server reads personnel activity state recognition results in a certain time range (e.g., the past month) from the database as time series sample data. Each time series sample data is composed of two parts: a historical activity state sequence, which records the activity state change of the personnel in a period of time (e.g., 1 hour); and a duration sequence, which records the time length of each activity state. For example, a sample data can be expressed as:

[0053] Historical activity state sequence: ["sedentary state", "sedentary state", "active state", "active state", "sleep state"]

[0054] Duration sequence: [20 minutes, 10 minutes, 15 minutes, 5 minutes, 10 minutes]

[0055] The server divides the obtained time series sample data into a training set, a validation set and a test set according to a certain proportion, wherein the training set is used for parameter learning of the model, the validation set is used for hyperparameter tuning of the model, and the test set is used for performance evaluation of the model. The server constructs a preset activity state prediction model based on a Seq2Seq (Sequence-to-Sequence) model, which is used to predict the trend of activity state change in a future period of time according to an activity state sequence in a past period of time. The Seq2Seq model consists of two parts: an encoder (Encoder) responsible for encoding the input historical activity state sequence into a fixed-length vector representation; and a decoder (Decoder) responsible for generating the output future activity state sequence step by step according to the encoded vector. The encoder uses a recurrent neural network to input the historical activity state sequence one by one and obtains an encoded vector containing historical information at the last time step. The decoder also uses a recurrent neural network to generate the future activity state sequence step by step, taking the encoded vector as the initial state. At each time step, the decoder generates the activity state at the next time step based on the current hidden state and the predicted activity state at the previous time step. In addition, the server can also introduce an attention mechanism between the encoder and the decoder, so that the decoder can adaptively focus on the output of the encoder at different time steps when generating the activity state at each time step, improving the prediction accuracy. The server trains the constructed Seq2Seq model using the divided training set data, optimizes the parameters of the model, so that it can as accurately as possible predict the trend of future activity state change according to the historical activity state sequence.

[0056] During the training process, the server sets a loss function (such as cross-entropy loss) to measure the difference between the predicted activity state sequence and the real activity state sequence, and uses a backpropagation algorithm and an optimizer (such as Adam) to minimize the loss function and adjust the model parameters. During the training process, the server will regularly evaluate the performance of the model on the validation set, and dynamically adjust the learning rate and regularization coefficient according to the evaluation results to prevent the model from overfitting or underfitting. After multiple rounds of iterative training, the server obtains a performance-optimized Seq2Seq activity state prediction model (i.e., the preset activity state prediction model), which can be used for actual personnel activity state prediction tasks.

[0057] In step S104, the server reads the personnel activity state recognition results in the latest period of time (e.g., the past 1 hour) from the database, and constructs a personnel activity time sequence as input data of the preset activity state prediction model. The server inputs the personnel activity time sequence into the preset activity state prediction model, starts the forward inference process of the model, and automatically generates an activity state sequence in the future period of time. In the inference process, the encoder part of the preset activity state prediction model reads the input personnel activity state sequence, updates the hidden state step by step, and outputs an encoding vector containing historical information at the last time step. Then, the decoder part of the preset activity state prediction model takes the encoding vector as the initial state, and generates the predicted future activity state sequence step by step. At each time step, the decoder generates the probability distribution of each activity state at the next time according to the current hidden state and the predicted activity state at the last time, and selects the predicted activity state with the maximum probability as the prediction result.

[0058] Step S105: According to the preset operation rule, the air conditioner operation data corresponding to the predicted activity state is determined, and the air conditioner operation data includes the air conditioner operation mode and the air conditioner target temperature.

[0059] In step S105, when the predicted activity state is the out-of-home state, the air conditioner operation mode is determined as the off mode; when the predicted activity state is the sleep state, the air conditioner operation mode is determined as the sleep mode, and the air conditioner target temperature is set as the preset sleep temperature; when the predicted activity state is the sedentary state, the air conditioner operation mode is determined as the cooling mode or the heating mode, and the air conditioner target temperature is set as the preset comfortable temperature; when the predicted activity state is the active state, the air conditioner operation mode is determined as the air supply mode.

[0060] Specifically, when the server predicts that the user activity state is the out-of-home state, it indicates that the user is not currently in the room, and there is no need to maintain the comfortable temperature in the room. The server sets the operation mode of the air conditioner to the off mode, i.e., completely stops the operation of the air conditioner, to avoid unnecessary energy waste. When the server predicts that the user activity state is the sleep state, it indicates that the user is currently resting, and the comfort requirement for temperature is high, and the user is sensitive to noise. The server sets the operation mode of the air conditioner to the sleep mode, i.e., operates at a low wind speed and noise, provides stable and uniform air supply, and creates a quiet and comfortable sleep environment. The server sets the target temperature of the air conditioner to the preset sleep temperature, which is usually 2-3 degrees Celsius lower than the normal environment temperature, e.g., 26 degrees Celsius. The preset sleep temperature is the best sleep temperature range obtained according to a large amount of user usage data and feedback, and scientific research on human comfort, which is not limited in the present application. When the server predicts that the user activity state is the sedentary state, it indicates that the user is currently engaged in sedentary static activities such as office work, study, and watching TV, and the comfort requirement for temperature is high, and the user hopes to maintain a certain air circulation.

[0061] The server sets the operation mode of the air conditioner to the cooling mode or the heating mode, and the specific mode is selected according to the difference between the current indoor and outdoor temperatures and the historical preferences of the user. When the indoor temperature is higher than the outdoor temperature and is higher than the upper limit of the user's comfortable temperature (such as 28 degrees Celsius), the server will start the cooling mode to reduce the room temperature to the comfortable temperature range. When the indoor temperature is lower than the outdoor temperature and is lower than the lower limit of the user's comfortable temperature (such as 20 degrees Celsius), the server will start the heating mode to raise the room temperature to the comfortable temperature range. The server sets the target temperature of the air conditioner to the preset comfortable temperature, which is usually between 22-26 degrees Celsius, and is fine-tuned according to the personal preferences and historical settings of the user. The preset comfortable temperature is the best sitting comfortable temperature range obtained from the physiological research of human body heat balance and a large amount of user usage data and feedback, and can be adjusted according to actual conditions, which is not limited in the present application. When the server predicts that the user's activity state is the active state, it indicates that the user is currently engaged in indoor activities of a certain intensity, such as housework, indoor exercise, etc., and generates more heat, which requires higher air quality and circulation. The server sets the air conditioner operation mode to the air supply mode to enhance indoor air circulation through higher air supply speed to timely discharge carbon dioxide, sweat, etc. generated by user activities, and maintain the freshness and dryness of indoor air.

[0062] Step S106: Obtain outdoor environment parameters, including outdoor temperature and outdoor humidity.

[0063] In step S106, during the installation and deployment of the intelligent air conditioner regulation system, temperature sensors and humidity sensors need to be installed at appropriate locations outdoors to collect real-time outdoor environment parameters. The server obtains the outdoor temperature through the temperature sensor and the outdoor humidity through the humidity sensor.

[0064] Step S107: Adjust the air conditioner operation data according to the outdoor environment parameters to obtain target air conditioner operation data.

[0065] In step S107, historical outdoor environment parameters and corresponding user temperature adjustment behavior data are obtained, including the user's temperature adjustment time, temperature adjustment amplitude, and target temperature after temperature adjustment; a user temperature adjustment behavior prediction model is constructed, which is used to predict user temperature adjustment behavior according to the current outdoor environment parameters; the outdoor temperature and outdoor humidity are input into the user temperature adjustment behavior prediction model to obtain the predicted temperature adjustment amplitude; the target temperature of the air conditioner is adjusted according to the predicted temperature adjustment amplitude to obtain the target air conditioner operation data.

[0066] Specifically, the server obtains historical outdoor environment parameters and corresponding user temperature adjustment behavior data as the data basis for training the user temperature adjustment behavior prediction model. The historical outdoor environment parameters can be obtained from meteorological departments or self-built environmental monitoring equipment, mainly including outdoor temperature, humidity indicators. The user temperature adjustment behavior data can be obtained by recording the actual operation of the user, mainly including the user's temperature adjustment time, temperature adjustment amplitude and target temperature after temperature adjustment. For example, the server can record the time point of each adjustment of the air conditioner temperature of the user, the temperature value before and after temperature adjustment, and the temperature adjustment amplitude (such as +1℃, -2℃, etc.).

[0067] In one possible implementation, the user temperature adjustment behavior prediction model is constructed, specifically including: taking the historical outdoor environment parameters as sample features and the user temperature adjustment behavior data as sample labels to construct a training sample set; selecting a preset machine learning algorithm, training using the training sample set to obtain a machine learning model for predicting user temperature adjustment behavior, and taking the machine learning model as the user temperature adjustment behavior prediction model.

[0068] Specifically, the server prepares a data set for training the machine learning model, that is, a training sample set. The training sample set consists of two parts: sample features and sample labels. Here, the sample features refer to factors affecting the user's temperature adjustment behavior, mainly historical outdoor environment parameters such as outdoor temperature and outdoor humidity; the sample labels are the actual temperature adjustment behavior of the user under specific environmental conditions, including temperature adjustment time, temperature adjustment amplitude, target temperature after temperature adjustment, etc. The server matches the outdoor environment parameters with the corresponding user temperature adjustment behavior by collecting and organizing historical data over a period of time to construct a complete training sample. For example, if the server collects that when the outdoor temperature is 30℃ and the humidity is 50%, the user adjusts the air conditioner temperature by 3℃, a training sample can be constructed, where the sample features are 30℃ and 50%, and the sample label is -3℃. In this way, the server can obtain a training sample set with a large amount of data covering different environmental conditions and user behavior. Next, the server selects a suitable machine learning algorithm to construct the user temperature adjustment behavior prediction model. In this application, a neural network algorithm is used to construct the user temperature adjustment behavior prediction model. After selecting the machine learning algorithm, the server trains the algorithm using the training sample set to obtain the final user temperature adjustment behavior prediction model. Specifically, the server inputs the training sample set into the selected machine learning algorithm, automatically extracts the rules and patterns of user temperature adjustment behavior from the training sample set through continuous learning and optimization, and encapsulates these rules and patterns in a mathematical model. The training process usually requires multiple iterations, and each iteration updates the parameters of the model to better fit the training data. Finally, the server takes the trained machine learning model as the user temperature adjustment behavior prediction model and integrates it into the intelligent air conditioner regulation system.

[0069] After obtaining the trained user temperature adjustment behavior prediction model, the server can use the model to intelligently adjust the air conditioner operation data according to the current outdoor environment parameters. Specifically, the server inputs the currently collected outdoor temperature and outdoor humidity into the user temperature adjustment behavior prediction model, and the model predicts the temperature adjustment amplitude that the user is likely to perform under the current environment. For example, the current outdoor temperature is 35°C and the humidity is 60%, the server inputs these two values into the user temperature adjustment behavior prediction model, and the user temperature adjustment behavior prediction model may predict that the user will lower the air conditioner temperature by 2°C. The adjusted air conditioner target temperature, together with other air conditioner operation parameters (such as air conditioner operation mode), constitutes the final target air conditioner operation data.

[0070] Step S108: Determine to run the air conditioner with the target air conditioner operation data.

[0071] In step S108, the server generates the optimized target air conditioner operation data according to the user temperature adjustment behavior prediction model in step S107 and the current outdoor environment parameters. The server encapsulates the target air conditioner operation data into control instructions and sends them to the smart air conditioner controller in the user's home through the network (such as Wi-Fi, mobile network). After receiving the control instructions issued by the server, the air conditioner controller parses the target operation parameters in the instructions and converts them into control signals that can be directly executed by the air conditioner device. The air conditioner controller sends the control signals converted from the target operation parameters to the air conditioner device through wired or wireless means (such as infrared, WIFI, Bluetooth, etc.). After receiving the control signals, the microprocessor of the air conditioner device adjusts the working state of the compressor, fan, electronic expansion valve, etc. of the air conditioner according to the instructions in the signals, and controls the refrigeration / heating power and air supply temperature.

[0072] For example, if the target air conditioner operation data requires the air conditioner to run in cooling mode and adjust the indoor temperature to 26°C, the air conditioner device will start the compressor, compress the refrigerant and send it to the heat exchanger of the indoor unit, and at the same time control the indoor unit fan to run at medium speed to send cold air into the room until the indoor temperature reaches 26°C and stabilizes around that temperature.

[0073] Reference Figure 2The application further provides a smart air conditioner adjusting device, which is a server, and the server comprises an acquisition module 201 and a processing module 202, wherein: the acquisition module 201 is configured to acquire multiple indoor images; the processing module 202 is configured to perform feature extraction on each indoor image to obtain a personnel activity state corresponding to each indoor image and a time point corresponding to the personnel activity state, wherein the personnel activity state comprises an outing state, a sleep state, a long sitting state and an activity state; the processing module 202 is further configured to construct a personnel activity time sequence according to the personnel activity state and the time point corresponding to the personnel activity state; the processing module 202 is further configured to input the personnel activity time sequence into a preset activity state prediction model to obtain a predicted activity state; the processing module 202 is further configured to determine air conditioner operation data corresponding to the predicted activity state according to a preset operation rule, wherein the air conditioner operation data comprises an air conditioner operation mode and an air conditioner target temperature; the acquisition module 201 is further configured to acquire outdoor environment parameters, wherein the outdoor environment parameters comprise an outdoor temperature and an outdoor humidity; the processing module 202 is further configured to adjust the air conditioner operation data according to the outdoor environment parameters to obtain target air conditioner operation data; and the processing module 202 is further configured to determine the air conditioner to operate in the target air conditioner operation data.

[0074] In a possible implementation, the processing module 202 performs feature extraction on each indoor image to obtain a personnel activity state corresponding to each indoor image, specifically comprising: the processing module 202 performs a preprocessing operation on the indoor image to obtain a target indoor image, wherein the preprocessing operation comprises image size normalization, image noise removal and image enhancement; the processing module 202 detects a target person from the target indoor image by using a target detection algorithm; the processing module 202 extracts human key points of the target person and calculates a set relationship between each human key point to obtain a posture feature, wherein the posture feature comprises a torso angle feature and a limb angle feature; and the processing module 202 determines the personnel activity state according to the torso angle feature and the limb angle feature.

[0075] In a possible implementation, the processing module 202 determines the personnel activity state according to the torso angle feature and the limb angle feature, specifically comprising: when the torso angle feature represents that the torso of the target person is perpendicular to the ground and the limb angle feature represents that the limbs of the target person are in an extended state, the processing module 202 determines that the personnel activity state is an activity state; when the torso angle feature represents that the torso of the target person is perpendicular to the ground and the limb angle feature represents that the limbs of the target person are in a bent state, the processing module 202 determines that the personnel activity state is a long sitting state; when the torso angle feature represents that the torso of the target person is at a preset angle with the ground and the limb angle feature represents that the limbs of the target person are in a horizontal placement state, the processing module 202 determines that the personnel activity state is a sleep state; and when the target person is not detected from the target indoor image within a preset time period, the processing module 202 determines that the personnel activity state is an outing state.

[0076] In a possible implementation, before the processing module 202 inputs the personnel activity time sequence into the preset activity state prediction model, the method further includes: the acquiring module 201 acquires time sequence sample data, the time sequence sample data including a historical activity state sequence and a duration corresponding to the historical activity state sequence; the processing module 202 constructs a seq2seq model, the seq2seq model including an encoder and a decoder, the encoder being configured to encode the input historical activity state sequence, and the decoder being configured to predict a personnel activity state at a next time; and the processing module 202 trains the seq2seq model by using the time sequence sample data, to obtain the preset activity state prediction model.

[0077] In a possible implementation, the processing module 202 determines the air conditioner running data corresponding to the predicted activity state according to a preset running rule, specifically including: when the predicted activity state is an outing state, the processing module 202 determines that the air conditioner running mode is a closed mode; when the predicted activity state is a sleep state, the processing module 202 determines that the air conditioner running mode is a sleep mode, and sets the air conditioner target temperature to a preset sleep temperature; when the predicted activity state is a sedentary state, the processing module 202 determines that the air conditioner running mode is a cooling mode or a heating mode, and sets the air conditioner target temperature to a preset comfortable temperature; and when the predicted activity state is an active state, the processing module 202 determines that the air conditioner running mode is a supply air mode.

[0078] In a possible implementation, the processing module 202 adjusts the air conditioner running data according to an outdoor environment parameter, to obtain target air conditioner running data, specifically including: the acquiring module 201 acquires historical outdoor environment parameters and corresponding user temperature adjustment behavior data, the user temperature adjustment behavior data including a user temperature adjustment time, a temperature adjustment amplitude and a target temperature after temperature adjustment; the processing module 202 constructs a user temperature adjustment behavior prediction model, the user temperature adjustment behavior prediction model being configured to predict user temperature adjustment behavior according to a current outdoor environment parameter; the processing module 202 inputs an outdoor temperature and an outdoor humidity into the user temperature adjustment behavior prediction model, to obtain a predicted temperature adjustment amplitude; and the processing module 202 adjusts the air conditioner target temperature according to the predicted temperature adjustment amplitude, to obtain the target air conditioner running data.

[0079] In a possible implementation, the processing module 202 constructs a user temperature adjustment behavior prediction model, specifically including: the processing module 202 constructs a training sample set by taking historical outdoor environment parameters as sample features and taking user temperature adjustment behavior data as sample labels; the processing module 202 selects a preset machine learning algorithm, trains the training sample set by using the preset machine learning algorithm, to obtain a machine learning model for predicting user temperature adjustment behavior, and takes the machine learning model as the user temperature adjustment behavior prediction model.

[0080] It should be noted that the apparatus provided in the above examples is only used as an example for the division of the above functional modules in realizing the functions thereof, and in actual applications, the above functions can be completed by different functional modules according to the needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the above-described functions. In addition, the apparatus and method embodiments provided in the above examples belong to the same concept, and the specific implementation process is detailed in the method embodiments, which will not be described here.

[0081] The present application also provides an electronic device. Referring to Figure 3 , Figure 3 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. The electronic device 300 can include at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0082] The communication bus 302 is used to realize the connection and communication between the components.

[0083] The user interface 303 can include a display screen (Display) and a camera (Camera), and the optional user interface 303 can further include a standard wired interface and a wireless interface.

[0084] The network interface 304 can optionally include a standard wired interface and a wireless interface (such as a WI-FI interface).

[0085] The processor 301 can include one or more processing cores. The processor 301 connects various parts within the server through various interfaces and lines, performs various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 305, and calling data stored in the memory 305. Alternatively, the processor 301 can be implemented in at least one of a hardware form of a digital signal processing (DSP), a field-programmable gate array (FPGA), and a programmable logic array (PLA). The processor 301 can integrate a combination of one or more of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes operating systems, user interfaces, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 301, but can be realized by a separate chip.

[0086] The memory 305 can include a random access memory (RAM) and a read-only memory (ROM). Alternatively, the memory 305 includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 305 can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 305 can also be at least one storage device located away from the aforementioned processor 301. Referring to Figure 3 The memory 305 as a computer storage medium can include an operating system, a network communication module, a user interface module, and an application program of an intelligent air conditioning regulation method.

[0087] In Figure 3In the electronic device 300 shown, the user interface 303 is mainly used to provide an interface for the user to input, and obtain data input by the user; and the processor 301 can be used to invoke an application program of the intelligent air conditioner regulating device stored in the memory 305, and when executed by one or more processors 301, the electronic device 300 performs the method described in one or more of the above embodiments. It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all expressed as a combination of a series of actions, but those skilled in the art should know that the present application is not limited to the action sequence described, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0088] The present application also provides a computer-readable storage medium, which stores instructions. When executed by one or more processors 301, the electronic device 300 performs the method described in one or more of the above embodiments.

[0089] In the above embodiments, the description of each embodiment has its own focus, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0090] In the several embodiments provided by the present application, it should be understood that the disclosed device can be implemented in other manners. For example, the division of the units is merely a logical function division, and there can be another division manner in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical or other forms.

[0091] The units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0092] In addition, each functional unit in the embodiments of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware, or in the form of software functional units.

[0093] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable memory. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a memory and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the embodiments of the present application. The aforementioned memory includes: a U disk, a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0094] The above-described are only exemplary embodiments of the present disclosure, and cannot limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. Other embodiments of the present disclosure will be readily apparent to those skilled in the art upon considering the specification and practicing the true principles of the present disclosure.

[0095] The present application is intended to cover any variations, uses or adaptive changes of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and examples are only considered as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A smart air conditioning regulation method, characterized by, The method comprises: obtaining multiple indoor images; extracting features from each of the indoor images to obtain a personnel activity state corresponding to each of the indoor images and a time point corresponding to the personnel activity state, the personnel activity state including an outing state, a sleep state, a sedentary state, and an active state; constructing a personnel activity time sequence according to the personnel activity state and the time point corresponding to the personnel activity state; inputting the personnel activity time sequence into a preset activity state prediction model to obtain a predicted activity state; determining air conditioner operation data corresponding to the predicted activity state according to a preset operation rule, the air conditioner operation data including an air conditioner operation mode and an air conditioner target temperature; obtaining outdoor environment parameters, the outdoor environment parameters including an outdoor temperature and an outdoor humidity; adjusting the air conditioner operation data according to the outdoor environment parameters to obtain target air conditioner operation data; determining to operate an air conditioner with the target air conditioner operation data, wherein the adjusting the air conditioner operation data according to the outdoor environment parameters to obtain target air conditioner operation data specifically comprises: obtaining historical outdoor environment parameters and corresponding user temperature adjustment behavior data, the user temperature adjustment behavior data including a user temperature adjustment time, a user temperature adjustment amplitude, and a target temperature after temperature adjustment; constructing a user temperature adjustment behavior prediction model, the user temperature adjustment behavior prediction model being configured to predict user temperature adjustment behavior according to a current outdoor environment parameter; inputting the outdoor temperature and the outdoor humidity into the user temperature adjustment behavior prediction model to obtain a predicted temperature adjustment amplitude; adjusting the air conditioner target temperature according to the predicted temperature adjustment amplitude to obtain the target air conditioner operation data.

2. The method of claim 1, wherein, the extracting features from each of the indoor images to obtain a personnel activity state corresponding to each of the indoor images specifically comprises: performing a preprocessing operation on the indoor images to obtain target indoor images, the preprocessing operation including image size normalization, image noise removal, and image enhancement; detecting a target person from the target indoor images using a target detection algorithm; extracting human key points of the target person and calculating a set relationship between each of the human key points to obtain a posture feature, the posture feature including a torso angle feature and a limb angle feature; determining the personnel activity state according to the torso angle feature and the limb angle feature.

3. The method of claim 2, wherein, the determining the personnel activity state according to the torso angle feature and the limb angle feature specifically comprises: when the torso angle feature indicates that the torso of the target person is perpendicular to the ground and the limb angle feature indicates that the limbs of the target person are in an extended state, determining that the personnel activity state is an active state; when the torso angle feature indicates that the torso of the target person is perpendicular to the ground and the limb angle feature indicates that the limbs of the target person are in a bent state, determining that the personnel activity state is a sedentary state; when the torso angle feature indicates that the torso of the target person is at a preset angle to the ground and the limb angle feature indicates that the limbs of the target person are in a horizontally placed state, determining that the personnel activity state is a sleep state; When the target person is not detected from the target indoor image within a preset time period, it is determined that the person activity state is an outing state.

4. The method of claim 1, wherein, Before the personnel activity time sequence is input into a preset activity state prediction model to obtain a predicted activity state, the method further includes: obtaining time sequence sample data, the time sequence sample data including a historical activity state sequence and a duration corresponding to the historical activity state sequence; constructing a seq2seq model, the seq2seq model including an encoder and a decoder, the encoder being configured to encode the input historical activity state sequence, and the decoder being configured to predict a personnel activity state at a next time point; training the seq2seq model using the time sequence sample data to obtain the preset activity state prediction model.

5. The method of claim 1, wherein, The preset operation rule includes: when the predicted activity state is an outing state, determining that an air conditioner operation mode is a closed mode; when the predicted activity state is a sleep state, determining that the air conditioner operation mode is a sleep mode, and setting an air conditioner target temperature to a preset sleep temperature; when the predicted activity state is a sedentary state, determining that the air conditioner operation mode is a cooling mode or a heating mode, and setting the air conditioner target temperature to a preset comfortable temperature; when the predicted activity state is an activity state, determining that the air conditioner operation mode is a blowing mode.

6. The method of claim 1, wherein, The method of constructing the user temperature adjustment behavior prediction model includes: using the historical outdoor environment parameters as sample features and the user temperature adjustment behavior data as sample labels to construct a training sample set; selecting a preset machine learning algorithm, training the training sample set to obtain a machine learning model for predicting user temperature adjustment behavior, and using the machine learning model as the user temperature adjustment behavior prediction model.

7. An intelligent air conditioning regulating device, characterized by, The device includes an acquisition module (201) and a processing module (202), wherein: the acquisition module (201) is configured to acquire multiple indoor images; the processing module (202) is configured to extract features from each of the indoor images to obtain a personnel activity state corresponding to each of the indoor images and a time point corresponding to the personnel activity state, the personnel activity state including an outing state, a sleep state, a sedentary state, and an activity state; the processing module (202) is further configured to construct a personnel activity time sequence according to the personnel activity state and the time point corresponding to the personnel activity state; the processing module (202) is further configured to input the personnel activity time sequence into a preset activity state prediction model to obtain a predicted activity state; the processing module (202) is further configured to determine air conditioner operation data corresponding to the predicted activity state according to a preset operation rule, the air conditioner operation data including an air conditioner operation mode and an air conditioner target temperature; the acquisition module (201) is further configured to acquire outdoor environment parameters, the outdoor environment parameters including an outdoor temperature and an outdoor humidity; the processing module (202) is further configured to adjust the air conditioner operation data according to the outdoor environment parameters to obtain target air conditioner operation data, wherein, The outdoor environment parameter is used to adjust the air conditioner operation data to obtain target air conditioner operation data, and the method specifically comprises the following steps: Obtain historical outdoor environment parameters and corresponding user temperature adjustment behavior data, wherein the user temperature adjustment behavior data comprises user temperature adjustment time, temperature adjustment amplitude and target temperature after temperature adjustment; Construct a user temperature adjustment behavior prediction model, wherein the user temperature adjustment behavior prediction model is used to predict user temperature adjustment behavior according to current outdoor environment parameters; Input the outdoor temperature and the outdoor humidity into the user temperature adjustment behavior prediction model to obtain a predicted temperature adjustment amplitude; Adjust the target temperature of the air conditioner according to the predicted temperature adjustment amplitude to obtain the target air conditioner operation data; The processing module (202) is further used to determine to run the air conditioner with the target air conditioner operation data.

8. An electronic device, comprising: The electronic device (300) comprises a processor (301), a memory (305), a user interface (303) and a network interface (304), the memory (305) is used to store instructions, the user interface (303) and the network interface (304) are used to communicate with other devices, and the processor (301) is used to execute the instructions stored in the memory (305) to enable the electronic device (300) to execute the method in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores instructions, and when the instructions are executed, the method in any one of claims 1-6 is executed.

Images