Air conditioner air supply control method, device and equipment and medium

Abstract

The invention relates to the technical field of artificial intelligence, and provides an air conditioner air supply control method, device and equipment and a medium, which can perform feature engineering based on data acquired by a multi-source sensor and user preference data, improve the accuracy of subsequent air supply control mode reasoning through the data of the multi-source sensor, and improve the reasoning efficiency. The user preference data enables the pattern reasoning result to be more personalized; the air supply mode prediction model trained based on the federated learning algorithm is used for mode reasoning, and the user privacy safety can be improved while rapid reasoning is achieved at edge nodes; and the decision engine is called to decide the candidate air supply modes inferred by the model to obtain the target air supply mode, air supply control is conducted on the air conditioner according to the target air supply mode, further decision making can be conducted in combination with the decision engine, the accuracy of mode control is further improved, and personalized comfort adjustment of the air supply modes of the air conditioner is achieved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to an air conditioning air supply control method, device, equipment and medium. Background Technology

[0002] With the development of smart homes and sophisticated air conditioning control, users' demand for a "comfortable but not disturbing" air supply experience is constantly increasing.

[0003] Traditional air conditioners typically determine whether to supply air based on a single temperature setting or simple infrared human body occupancy detection, which cannot simultaneously take into account individual comfort preferences, real-time changes in body surface temperature, and minimal disturbance strategies (such as windproof mode).

[0004] While some high-end air conditioning systems support facial recognition or proximity-to-turn-on / proximity-to-close, they still lack personalized decision-making based on long-term user behavior modeling.

[0005] Therefore, there is an urgent need for an air conditioning air supply control solution that features low latency, high privacy, high accuracy, and personalized control. Summary of the Invention

[0006] In view of the above, it is necessary to provide an air conditioning air supply control method, device, equipment and medium, which aims to solve the problems of lack of personalization, low privacy, low accuracy and high delay in air conditioning air supply control.

[0007] An air conditioning air supply control method, the air conditioning air supply control method comprising: In response to the air supply control command of the target air conditioner in the target area, the system uses multi-source sensors to collect sensor data and obtain user preference data. Feature engineering is performed based on the sensor data and the user preference data to obtain target features; The target features are input into the air supply pattern prediction model trained based on the federated learning algorithm to obtain candidate air supply patterns; The decision engine is invoked to make a decision on the candidate air supply modes, and the target air supply mode is obtained; The target air conditioner is controlled to supply air according to the target air supply mode.

[0008] An air conditioning air supply control device, the air conditioning air supply control device comprising: The data acquisition unit is used to respond to the air supply control command of the target air conditioner in the target area, and to acquire sensor data and user preference data using multi-source sensors. The feature engineering unit is used to perform feature engineering based on the sensor data and the user preference data to obtain target features; The input unit is used to input the target features into the air supply mode prediction model trained based on the federated learning algorithm to obtain candidate air supply modes. The decision unit is used to call the decision engine to make a decision on the candidate air supply modes and obtain the target air supply mode; The control unit is used to control the air supply of the target air conditioner according to the target air supply mode.

[0009] A computer device, the computer device comprising: A memory that stores at least one instruction; and a processor that executes the instructions stored in the memory to implement the air conditioning air supply control method.

[0010] A computer-readable storage medium storing at least one instruction, which is executed by a processor in a computer device to implement the air conditioning air supply control method.

[0011] As can be seen from the above technical solutions, this invention can perform feature engineering based on data collected from multi-source sensors and user preference data. Multi-source sensor data improves the accuracy of subsequent air supply control mode inference, and user preference data makes the mode inference results more personalized. By using an air supply mode prediction model trained based on federated learning algorithm for mode inference, fast inference can be achieved at edge nodes while improving user privacy and security. The decision engine is called to make a decision on the candidate air supply modes inferred by the model to obtain the target air supply mode, and the air supply of the air conditioner is controlled according to the target air supply mode. The decision engine can be combined to make further decisions, which further improves the accuracy of mode control and realizes personalized comfort adjustment of the air conditioner's air supply mode. Attached Figure Description

[0012] Figure 1 This is a flowchart of a preferred embodiment of the air conditioning air supply control method of the present invention.

[0013] Figure 2 This is a flowchart of a preferred embodiment of the feature engineering of the present invention.

[0014] Figure 3 This is a flowchart of a preferred embodiment of the present invention, which calls a decision engine to make decisions on candidate air supply modes.

[0015] Figure 4 This is a functional block diagram of a preferred embodiment of the air conditioning air supply control device of the present invention.

[0016] Figure 5 This is a schematic diagram of the structure of a computer device that implements the air conditioning air supply control method of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] like Figure 1 The diagram shown is a flowchart of a preferred embodiment of the air conditioning air supply control method of the present invention. Depending on different requirements, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0019] The air conditioning air supply control method is applied to one or more computer devices. The computer device is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. Its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0020] The computer device can be any electronic product that can interact with the user, such as a personal computer, tablet computer, smartphone, personal digital assistant (PDA), game console, interactive network television (IPTV), smart wearable device, etc.

[0021] The computer equipment may also include network equipment and / or user equipment. The network equipment includes, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of hosts or network servers.

[0022] The server can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0023] Artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0024] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0025] The network in which the computer device is located includes, but is not limited to, the Internet, wide area network, metropolitan area network, local area network, and virtual private network (VPN).

[0026] S10, in response to the air supply control command of the target air conditioner in the target area, uses multi-source sensors to collect sensor data and obtain user preference data.

[0027] In this embodiment, the target area can be an area equipped with air conditioning, such as a bedroom, living room, or office.

[0028] In this embodiment, the target air conditioner can be a smart air conditioner with mode adjustment function.

[0029] In this embodiment, the air supply control command can be automatically triggered when the target air conditioner is started, or it can be triggered according to user needs (such as when the user clicks a designated button on the air conditioner remote control, the air supply control command is triggered).

[0030] In this embodiment, the process of collecting sensor data using multi-source sensors and obtaining user preference data includes: Point cloud data within the target area is acquired using a multi-antenna multiple-input multiple-output (MIMO) short-range millimeter-wave radar. The infrared body surface temperature sensor is used to collect human body surface temperature distribution data within the target area; Environmental data within the target area is collected using environmental sensors. The sensor data is obtained by combining the point cloud data, the human body surface temperature distribution data, and the environmental data; The user preference data is obtained by collecting explicit preference data and implicit feedback data from the user interface.

[0031] The multi-antenna, multi-input, multi-output short-range millimeter-wave radar can generate point cloud data by transmitting nanosecond-level radio pulse signals to collect data such as indoor occupancy status, human body position coordinates, movement speed, and micro-Doppler time series in real time.

[0032] The infrared body surface temperature sensor can collect the body surface temperature distribution data using a thermopile or a thermal array.

[0033] The human body surface temperature distribution data may include the maximum skin temperature and average skin temperature within the target area.

[0034] The environmental sensors may include temperature sensors, humidity sensors, carbon dioxide sensors, etc.

[0035] The environmental data may include temperature, humidity, and air quality data within the target area.

[0036] The explicit preference data may include explicit user behaviors collected through user interfaces such as applications, voice, or remote controls, such as users manually turning "wind-following-person mode" on or off, user-uploaded airflow mode control scores through applications, and user-uploaded preferences (such as preference for medium wind speed and priority for avoiding wind at night).

[0037] The implicit feedback data may include unsatisfactory feedback determined by frequently turning off the air conditioner within a short period of time.

[0038] Through the above embodiments, multi-dimensional data can be obtained, providing a comprehensive and rich data foundation for subsequent air conditioning air supply mode inference, thereby helping to improve the accuracy of subsequent air conditioning air supply mode inference.

[0039] S11, Perform feature engineering based on the sensor data and the user preference data to obtain target features.

[0040] In this embodiment, in order to improve data quality and reduce the difficulty and accuracy of subsequent feature engineering, it is necessary to preprocess the sensor data and the user preference data first.

[0041] Specifically, before performing feature engineering based on the sensor data and the user preference data, the method further includes: The sensor data and user preference data are time-stamp aligned and sampling window interpolated to obtain the data to be processed; A Kalman filter is used to smooth the point cloud data in the data to be processed; Non-uniformity correction (NUC) is performed on the human body surface temperature distribution data in the data to be processed, and ambient light interference is removed by using a moving average filter; A bandpass filter is used to filter out instantaneous fluctuations in environmental data in the data to be processed.

[0042] Specifically, when aligning the timestamps of the sensor data and the user preference data, the timestamps of the point cloud data, the human body surface temperature distribution data, the environmental data, and the user preference data can be unified to NTP (Network Time Protocol) or a local clock, thereby eliminating time discrepancies between different devices.

[0043] When performing sampling window interpolation alignment on the sensor data and the user preference data, a sampling window of a certain size can be used to perform linear interpolation processing on various data. For example, a uniform sampling window of 200ms can be used to perform linear interpolation processing on each data, thereby ensuring that complete multi-dimensional data in the form of "location-temperature-environment" can be obtained within the same time window.

[0044] By aligning timestamps and sampling windows through interpolation, the problem of data misalignment caused by differences in sampling frequencies of multiple sensors can be solved, providing a time-consistent data foundation for subsequent feature fusion and model inference, and avoiding misjudgments caused by time deviations.

[0045] Among them, the point cloud data in the data to be processed is smoothed by using a Kalman filter, which can effectively remove high-frequency noise (such as equipment vibration, signal reflection interference, etc.).

[0046] Among them, non-uniformity correction is performed on the human body surface temperature distribution data in the data to be processed, which can eliminate temperature measurement errors caused by differences in the characteristics of the sensor itself, and the ambient light interference can be removed by using a moving average filter.

[0047] Among them, using a bandpass filter to filter out instantaneous fluctuations in environmental data (such as sudden temperature changes caused by opening a window) in the data to be processed can preserve effective trend data.

[0048] The above embodiments can effectively improve the signal-to-noise ratio of the original data, providing reliable input for subsequent feature engineering.

[0049] like Figure 2 The diagram shown is a flowchart of a preferred embodiment of the feature engineering process of the present invention. Specifically, the step of performing feature engineering based on the sensor data and the user preference data to obtain the target features includes: S110, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is used to identify human targets in the point cloud data and track the activity trajectory of the human targets; S111, A human trajectory is generated based on the activity trajectory using a Kalman filter; S112, construct spatial features including the distance and relative orientation between the human target and the air outlet of the target air conditioner based on the human trajectory; S113, construct physiological features including body surface temperature sequence, skin temperature difference, and temperature change slope based on the human body surface temperature distribution data; S114, Construct environmental features including indoor temperature and humidity, and time period unique thermal encoding based on the environmental data; S115, construct user features based on the user preference data, including previous air supply control data, user feedback tags, and user preference vectors; S116, Based on the feature fusion strategy, the spatial features, the physiological features, the environmental features, and the user features are fused to obtain the target features.

[0050] Among them, a density-based noisy spatial clustering algorithm is used to identify human targets and eliminate interfering objects from millimeter-wave radar point cloud data. Then, a stable human trajectory is generated through a Kalman filter, which can output dynamic attributes such as real-time position, movement speed, and relative air outlet orientation.

[0051] The relative orientation may include whether the human body is facing the air conditioner's air outlet, etc.

[0052] The skin temperature difference can be the difference between the skin temperature and the ambient temperature.

[0053] The slope of the temperature change can be the rate of change of body surface temperature over a certain period of time.

[0054] The unique hot encoding of the time period can be represented as a data structure similar to "daytime 8:00-20:00" and "nighttime 20:00-8:00".

[0055] The previous air supply control data may include: the previous air supply control command (such as air following people, air avoiding people), the execution result of the previous air supply control, etc.

[0056] The user feedback tags may include: satisfied, dissatisfied, etc.

[0057] It should be noted that the various features constructed above may also include other sub-features; the sub-features mentioned above are merely examples. For instance, the physiological features may also include other sub-features such as activity energy and whether one feels hot.

[0058] The feature fusion strategy may include a weighting strategy, such as configuring a weight coefficient for each data source and performing weighted calculations on each feature according to the weight coefficients of different data sources, thereby obtaining the target feature.

[0059] Through the above embodiments, raw sensor data can be transformed into structured feature vectors that can be directly used for model inference, providing comprehensive data support for personalized decision-making.

[0060] S12, the target features are input into the air supply mode prediction model trained based on the federated learning algorithm to obtain candidate air supply modes.

[0061] In this embodiment, before inputting the target features into the air supply pattern prediction model trained based on the federated learning algorithm, the method further includes: A hybrid model is constructed, comprising a short-time series prediction branch and a structured feature branch; wherein the short-time series prediction branch and the structured feature branch are fused using a model fusion processor; the short-time series prediction branch is used to capture the time dependency between body surface temperature and behavior; the structured feature branch is used to model user static preferences. Construct a training set based on the dimensions of the target features; Construct a loss function based on a misjudgment penalty mechanism; A federated learning algorithm is used to train the hybrid model on the edge node to which the target air conditioner belongs, using the training set and the loss function, to obtain a local model; The parameters of the local model are encrypted and sent to the central node, and the model parameters fed back by the central node are received. The model parameters are loaded onto the framework of the local model to obtain the air supply pattern prediction model.

[0062] The short-term series prediction branch can employ models with temporal attributes, such as the lightweight Transformer model or the LSTM (Long Short-Term Memory) model.

[0063] The structured feature branch can be modeled using the LightGBM (Light Gradient Boosting Machine) tree model to model structured historical preferences and environmental features.

[0064] Specifically, after fusing the outputs of the short-time series prediction branch and the structured feature branch using the model fusioner, the probabilities of wind supply modes such as "wind follows people", "wind avoids people", and "maintain current mode" can be output.

[0065] In constructing the loss function based on the aforementioned misjudgment penalty mechanism, the cross-entropy loss function can be used as a basis, and higher weights can be assigned to high-impact misjudgments (e.g., the weight of "falsely triggering wind avoidance" is set to 2, the weight of "falsely blowing wind" is set to 1.5, and the weight of correct classification is set to 1), thereby strengthening the model's ability to avoid key misjudgments.

[0066] Specifically, the parameters of the local model can be encrypted and sent to the central node using a Secure Aggregation (SA) algorithm or a differential privacy algorithm, thereby achieving model aggregation in the cloud without leaking user privacy.

[0067] The above embodiments enable model training while ensuring user privacy, effectively improving data security.

[0068] In this embodiment, the air supply pattern prediction model obtained after training can be lightweighted (e.g., quantized and distilled first) and then deployed on a local edge processor (e.g., an air conditioner embedded controller or single board) to reduce inference latency through local processing.

[0069] In this embodiment, since the model is trained using a federated learning algorithm, the inference and prediction of the air supply mode can be performed locally on each edge node, which effectively reduces latency and improves the efficiency of air supply mode inference.

[0070] S13, invoke the decision engine to make a decision on the candidate air supply mode and obtain the target air supply mode.

[0071] Please refer to Figure 3 This is a flowchart of a preferred embodiment of the present invention, invoking a decision engine to make a decision on candidate air supply modes. Specifically, the step of invoking the decision engine to make a decision on the candidate air supply modes and obtaining the target air supply mode includes: S130, obtain the real-time probability of the candidate air supply mode and the corresponding trigger threshold; S131, when the real-time probability is greater than the trigger threshold and the safety constraints are met, an air supply command based on the candidate air supply mode is sent to the target air conditioner; wherein, the safety constraints include a disabled time period constraint and a user preference constraint; S132, within a preset time period, when the real-time probability of the candidate air supply mode drops to a preset probability, a short-term fluctuation event is determined to have occurred, and a rollback command to the target air conditioner to revert to the original mode is sent; or S133, within the preset time period, when the real-time probability of the candidate air supply mode does not decrease to the preset probability, the candidate air supply mode is determined as the target air supply mode.

[0072] The trigger threshold can be an optimal value selected based on a large number of experiments. For example, the trigger threshold can be configured to 0.7.

[0073] The safety constraints can be configured to meet specific user needs. For example, when "nighttime wind avoidance" is configured, even if the probability of "wind avoidance mode" at night is greater than 0.6 (lower than the daytime trigger threshold of 0.7), the wind avoidance mode will still be triggered first, thus meeting the user's personalized needs.

[0074] The trigger threshold, preset duration, and preset probability can be configured based on experiments. For example, the trigger threshold can be configured to 0.7, the preset duration to 120 seconds, and the preset probability to 0.4. In this case, if the probability of a certain mode is greater than the corresponding trigger threshold (e.g., 0.7), that mode is initially triggered. Within 120 seconds after the trigger command, if the probability of the corresponding mode drops below 0.4, it is determined to be a short-term fluctuation, and the system reverts to the original mode, thus avoiding frequent switching.

[0075] Through the above embodiments, the decision engine can be invoked to make further decisions based on model reasoning, thereby effectively improving the accuracy of air conditioning air supply mode control. At the same time, due to the rollback operation, the discomfort caused by frequent switching can be avoided, and the preference adaptation logic further improves the user's subjective comfort.

[0076] S14, control the air supply of the target air conditioner according to the target air supply mode.

[0077] In this embodiment, the target air supply mode can be converted into a signal that the air conditioner actuator can recognize, and the air outlet angle (such as adjusting the blades to avoid the human body area), wind speed (such as medium wind speed), zoned air supply (such as multi-outlet air conditioner closing the air outlet in the human body area), etc. can be controlled to control the target air conditioner to supply air according to the target air supply mode.

[0078] In this embodiment, after controlling the air supply of the target air conditioner according to the target air supply mode, the method further includes: Obtain user feedback data; The air supply mode prediction model and the decision engine are optimized based on the user feedback data.

[0079] For example, the actuator of the target air conditioner can provide real-time feedback on the command execution results and store them in the local log. User feedback on the current control effect can also be collected through application pop-ups (such as a pop-up displaying "Is the current air supply mode comfortable?"), voice inquiries, and manual operation recognition (such as the user manually turning off the "wind follows person" mode being considered "unsatisfied"). This data is then periodically used as online learning samples to optimize the air supply mode prediction model and the decision engine, thereby continuously improving the accuracy of air supply mode control and ensuring that there is no direct airflow in "wind avoids person mode" and precise airflow in "wind follows person mode".

[0080] As can be seen from the above technical solutions, this invention can perform feature engineering based on data collected from multi-source sensors and user preference data. Multi-source sensor data improves the accuracy of subsequent air supply control mode inference, and user preference data makes the mode inference results more personalized. By using an air supply mode prediction model trained based on federated learning algorithm for mode inference, fast inference can be achieved at edge nodes while improving user privacy and security. The decision engine is called to make a decision on the candidate air supply modes inferred by the model to obtain the target air supply mode, and the air supply of the air conditioner is controlled according to the target air supply mode. The decision engine can be combined to make further decisions, which further improves the accuracy of mode control and realizes personalized comfort adjustment of the air conditioner's air supply mode.

[0081] like Figure 4 The diagram shown is a functional block diagram of a preferred embodiment of the air conditioning air supply control device of the present invention. The air conditioning air supply control device 11 includes a data acquisition unit 110, a feature engineering unit 111, an input unit 112, a decision unit 113, and a control unit 114. The module / unit referred to in this invention is a series of computer program segments that can be executed by a processor and perform a fixed function, and which are stored in a memory. In this embodiment, the functions of each module / unit will be described in detail in subsequent embodiments.

[0082] The acquisition unit 110 is used to acquire sensor data and obtain user preference data in response to the air supply control command of the target air conditioner in the target area.

[0083] In this embodiment, the target area can be an area equipped with air conditioning, such as a bedroom, living room, or office.

[0084] In this embodiment, the target air conditioner can be a smart air conditioner with mode adjustment function.

[0085] In this embodiment, the air supply control command can be automatically triggered when the target air conditioner is started, or it can be triggered according to user needs (such as when the user clicks a designated button on the air conditioner remote control, the air supply control command is triggered).

[0086] In this embodiment, the acquisition unit 110 uses multi-source sensors to acquire sensor data and obtains user preference data, including: Point cloud data within the target area is acquired using a multi-antenna multiple-input multiple-output (MIMO) short-range millimeter-wave radar. The infrared body surface temperature sensor is used to collect human body surface temperature distribution data within the target area; Environmental data within the target area is collected using environmental sensors. The sensor data is obtained by combining the point cloud data, the human body surface temperature distribution data, and the environmental data; The user preference data is obtained by collecting explicit preference data and implicit feedback data from the user interface.

[0087] The multi-antenna, multi-input, multi-output short-range millimeter-wave radar can generate point cloud data by transmitting nanosecond-level radio pulse signals to collect data such as indoor occupancy status, human body position coordinates, movement speed, and micro-Doppler time series in real time.

[0088] The infrared body surface temperature sensor can collect the body surface temperature distribution data using a thermopile or a thermal array.

[0089] The human body surface temperature distribution data may include the maximum skin temperature and average skin temperature within the target area.

[0090] The environmental sensors may include temperature sensors, humidity sensors, carbon dioxide sensors, etc.

[0091] The environmental data may include temperature, humidity, and air quality data within the target area.

[0092] The explicit preference data may include explicit user behaviors collected through user interfaces such as applications, voice, or remote controls, such as users manually turning "wind-following-person mode" on or off, user-uploaded airflow mode control scores through applications, and user-uploaded preferences (such as preference for medium wind speed and priority for avoiding wind at night).

[0093] The implicit feedback data may include unsatisfactory feedback determined by frequently turning off the air conditioner within a short period of time.

[0094] Through the above embodiments, multi-dimensional data can be obtained, providing a comprehensive and rich data foundation for subsequent air conditioning air supply mode inference, thereby helping to improve the accuracy of subsequent air conditioning air supply mode inference.

[0095] The feature engineering unit 111 is used to perform feature engineering based on the sensor data and the user preference data to obtain target features.

[0096] In this embodiment, in order to improve data quality and reduce the difficulty and accuracy of subsequent feature engineering, it is necessary to preprocess the sensor data and the user preference data first.

[0097] Specifically, before performing feature engineering based on the sensor data and the user preference data, the feature engineering unit 111 performs timestamp alignment and sampling window interpolation alignment on the sensor data and the user preference data to obtain the data to be processed. A Kalman filter is used to smooth the point cloud data in the data to be processed; Non-uniformity correction (NUC) is performed on the human body surface temperature distribution data in the data to be processed, and ambient light interference is removed by using a moving average filter; A bandpass filter is used to filter out instantaneous fluctuations in environmental data in the data to be processed.

[0098] Specifically, when aligning the timestamps of the sensor data and the user preference data, the timestamps of the point cloud data, the human body surface temperature distribution data, the environmental data, and the user preference data can be unified to NTP (Network Time Protocol) or a local clock, thereby eliminating time discrepancies between different devices.

[0099] When performing sampling window interpolation alignment on the sensor data and the user preference data, a sampling window of a certain size can be used to perform linear interpolation processing on various data. For example, a uniform sampling window of 200ms can be used to perform linear interpolation processing on each data, thereby ensuring that complete multi-dimensional data in the form of "location-temperature-environment" can be obtained within the same time window.

[0100] By aligning timestamps and sampling windows through interpolation, the problem of data misalignment caused by differences in sampling frequencies of multiple sensors can be solved, providing a time-consistent data foundation for subsequent feature fusion and model inference, and avoiding misjudgments caused by time deviations.

[0101] Among them, the point cloud data in the data to be processed is smoothed by using a Kalman filter, which can effectively remove high-frequency noise (such as equipment vibration, signal reflection interference, etc.).

[0102] Among them, non-uniformity correction is performed on the human body surface temperature distribution data in the data to be processed, which can eliminate temperature measurement errors caused by differences in the characteristics of the sensor itself, and the ambient light interference can be removed by using a moving average filter.

[0103] Among them, using a bandpass filter to filter out instantaneous fluctuations in environmental data (such as sudden temperature changes caused by opening a window) in the data to be processed can preserve effective trend data.

[0104] The above embodiments can effectively improve the signal-to-noise ratio of the original data, providing reliable input for subsequent feature engineering.

[0105] In this embodiment, the feature engineering unit 111 performs feature engineering based on the sensor data and the user preference data to obtain target features including: The Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is used to identify human targets in the point cloud data and track the movement trajectory of the human targets. A human trajectory is generated based on the activity trajectory using a Kalman filter; Based on the human body trajectory, a spatial feature is constructed including the distance and relative orientation between the human body target and the air outlet of the target air conditioner; Physiological features, including body surface temperature sequence, skin temperature difference, and temperature change slope, are constructed based on the human body surface temperature distribution data. Based on the environmental data, an environmental feature is constructed, including indoor temperature and humidity, and time-time unique thermal encoding. Based on the user preference data, a user feature is constructed that includes the previous air supply control data, user feedback tags, and user preference vectors. The target feature is obtained by fusing the spatial features, physiological features, environmental features, and user features based on a feature fusion strategy.

[0106] Among them, a density-based noisy spatial clustering algorithm is used to identify human targets and eliminate interfering objects from millimeter-wave radar point cloud data. Then, a stable human trajectory is generated through a Kalman filter, which can output dynamic attributes such as real-time position, movement speed, and relative air outlet orientation.

[0107] The relative orientation may include whether the human body is facing the air conditioner's air outlet, etc.

[0108] The skin temperature difference can be the difference between the skin temperature and the ambient temperature.

[0109] The slope of the temperature change can be the rate of change of body surface temperature over a certain period of time.

[0110] The unique hot encoding of the time period can be represented as a data structure similar to "daytime 8:00-20:00" and "nighttime 20:00-8:00".

[0111] The previous air supply control data may include: the previous air supply control command (such as air following people, air avoiding people), the execution result of the previous air supply control, etc.

[0112] The user feedback tags may include: satisfied, dissatisfied, etc.

[0113] It should be noted that the various features constructed above may also include other sub-features; the sub-features mentioned above are merely examples. For instance, the physiological features may also include other sub-features such as activity energy and whether one feels hot.

[0114] The feature fusion strategy may include a weighting strategy, such as configuring a weight coefficient for each data source and performing weighted calculations on each feature according to the weight coefficients of different data sources, thereby obtaining the target feature.

[0115] Through the above embodiments, raw sensor data can be transformed into structured feature vectors that can be directly used for model inference, providing comprehensive data support for personalized decision-making.

[0116] The input unit 112 is used to input the target features into the air supply mode prediction model trained based on the federated learning algorithm to obtain candidate air supply modes.

[0117] In this embodiment, before inputting the target features into the air supply pattern prediction model trained based on the federated learning algorithm, a hybrid model including a short-time series prediction branch and a structured feature branch is constructed; wherein, the short-time series prediction branch and the structured feature branch are fused using a model fusion processor; the short-time series prediction branch is used to capture the time dependency between body surface temperature and behavior; the structured feature branch is used to model user static preferences; Construct a training set based on the dimensions of the target features; Construct a loss function based on a misjudgment penalty mechanism; A federated learning algorithm is used to train the hybrid model on the edge node to which the target air conditioner belongs, using the training set and the loss function, to obtain a local model; The parameters of the local model are encrypted and sent to the central node, and the model parameters fed back by the central node are received. The model parameters are loaded onto the framework of the local model to obtain the air supply pattern prediction model.

[0118] The short-term series prediction branch can employ models with temporal attributes, such as the lightweight Transformer model or the LSTM (Long Short-Term Memory) model.

[0119] The structured feature branch can be modeled using the LightGBM (Light Gradient Boosting Machine) tree model to model structured historical preferences and environmental features.

[0120] Specifically, after fusing the outputs of the short-time series prediction branch and the structured feature branch using the model fusioner, the probabilities of wind supply modes such as "wind follows people", "wind avoids people", and "maintain current mode" can be output.

[0121] In constructing the loss function based on the aforementioned misjudgment penalty mechanism, the cross-entropy loss function can be used as a basis, and higher weights can be assigned to high-impact misjudgments (e.g., the weight of "falsely triggering wind avoidance" is set to 2, the weight of "falsely blowing wind" is set to 1.5, and the weight of correct classification is set to 1), thereby strengthening the model's ability to avoid key misjudgments.

[0122] Specifically, the parameters of the local model can be encrypted and sent to the central node using a Secure Aggregation (SA) algorithm or a differential privacy algorithm, thereby achieving model aggregation in the cloud without leaking user privacy.

[0123] The above embodiments enable model training while ensuring user privacy, effectively improving data security.

[0124] In this embodiment, the air supply pattern prediction model obtained after training can be lightweighted (e.g., quantized and distilled first) and then deployed on a local edge processor (e.g., an air conditioner embedded controller or single board) to reduce inference latency through local processing.

[0125] In this embodiment, since the model is trained using a federated learning algorithm, the inference and prediction of the air supply mode can be performed locally on each edge node, which effectively reduces latency and improves the efficiency of air supply mode inference.

[0126] The decision unit 113 is used to call the decision engine to make a decision on the candidate air supply mode and obtain the target air supply mode.

[0127] In this embodiment, the decision unit 113 invokes the decision engine to make a decision on the candidate air supply modes, and the target air supply mode is obtained as follows: Obtain the real-time probability of the candidate air supply mode and the corresponding trigger threshold; When the real-time probability is greater than the trigger threshold and the safety constraints are met, an air supply command based on the candidate air supply mode is sent to the target air conditioner; wherein, the safety constraints include a disabled time period constraint and a user preference constraint; Within a preset time period, when the real-time probability of the candidate air supply mode drops to a preset probability, a short-term fluctuation event is determined to have occurred, and a rollback command to the target air conditioner to revert to the original mode is sent; or If, within the preset time period, the real-time probability of the candidate air supply mode does not decrease to the preset probability, the candidate air supply mode is determined as the target air supply mode.

[0128] The trigger threshold can be an optimal value selected based on a large number of experiments. For example, the trigger threshold can be configured to 0.7.

[0129] The safety constraints can be configured to meet specific user needs. For example, when "nighttime wind avoidance" is configured, even if the probability of "wind avoidance mode" at night is greater than 0.6 (lower than the daytime trigger threshold of 0.7), the wind avoidance mode will still be triggered first, thus meeting the user's personalized needs.

[0130] The trigger threshold, preset duration, and preset probability can be configured based on experiments. For example, the trigger threshold can be configured to 0.7, the preset duration to 120 seconds, and the preset probability to 0.4. In this case, if the probability of a certain mode is greater than the corresponding trigger threshold (e.g., 0.7), that mode is initially triggered. Within 120 seconds after the trigger command, if the probability of the corresponding mode drops below 0.4, it is determined to be a short-term fluctuation, and the system reverts to the original mode, thus avoiding frequent switching.

[0131] Through the above embodiments, the decision engine can be invoked to make further decisions based on model reasoning, thereby effectively improving the accuracy of air conditioning air supply mode control. At the same time, due to the rollback operation, the discomfort caused by frequent switching can be avoided, and the preference adaptation logic further improves the user's subjective comfort.

[0132] The control unit 114 is used to control the air supply of the target air conditioner according to the target air supply mode.

[0133] In this embodiment, the target air supply mode can be converted into a signal that the air conditioner actuator can recognize, and the air outlet angle (such as adjusting the blades to avoid the human body area), wind speed (such as medium wind speed), zoned air supply (such as multi-outlet air conditioner closing the air outlet in the human body area), etc. can be controlled to control the target air conditioner to supply air according to the target air supply mode.

[0134] In this embodiment, after the control unit 114 controls the air supply of the target air conditioner according to the target air supply mode, it obtains user feedback data. The air supply mode prediction model and the decision engine are optimized based on the user feedback data.

[0135] For example, the actuator of the target air conditioner can provide real-time feedback on the command execution results and store them in the local log. User feedback on the current control effect can also be collected through application pop-ups (such as a pop-up displaying "Is the current air supply mode comfortable?"), voice inquiries, and manual operation recognition (such as the user manually turning off the "wind follows person" mode being considered "unsatisfied"). This data is then periodically used as online learning samples to optimize the air supply mode prediction model and the decision engine, thereby continuously improving the accuracy of air supply mode control and ensuring that there is no direct airflow in "wind avoids person mode" and precise airflow in "wind follows person mode".

[0136] As can be seen from the above technical solutions, this invention can perform feature engineering based on data collected from multi-source sensors and user preference data. Multi-source sensor data improves the accuracy of subsequent air supply control mode inference, and user preference data makes the mode inference results more personalized. By using an air supply mode prediction model trained based on federated learning algorithm for mode inference, fast inference can be achieved at edge nodes while improving user privacy and security. The decision engine is called to make a decision on the candidate air supply modes inferred by the model to obtain the target air supply mode, and the air supply of the air conditioner is controlled according to the target air supply mode. The decision engine can be combined to make further decisions, which further improves the accuracy of mode control and realizes personalized comfort adjustment of the air conditioner's air supply mode.

[0137] like Figure 5 The diagram shown is a schematic diagram of the structure of a computer device that implements the air conditioning air supply control method of the present invention.

[0138] The computer device 1 may include a memory 12, a processor 13, and a bus (the arrow in the figure represents the bus), and may also include a computer program stored in the memory 12 and executable on the processor 13, such as an air conditioning air supply control program.

[0139] Those skilled in the art will understand that the schematic diagram is merely an example of computer device 1 and does not constitute a limitation on computer device 1. Computer device 1 can be either a bus topology or a star topology. Computer device 1 may also include more or fewer other hardware or software than shown in the diagram, or different component arrangements. For example, computer device 1 may also include input / output devices, network access devices, etc.

[0140] It should be noted that the computer device 1 described is merely an example. Other existing or future electronic products that are adaptable to this invention should also be included within the scope of protection of this invention and are incorporated herein by reference.

[0141] The memory 12 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the computer device 1, such as a portable hard drive of the computer device 1. In other embodiments, the memory 12 can be an external storage device of the computer device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the computer device 1. Furthermore, the memory 12 can include both internal and external storage units of the computer device 1. The memory 12 can be used not only to store application software and various types of data installed on the computer device 1, such as the code of an air conditioning ventilation control program, but also to temporarily store data that has been output or will be output.

[0142] In some embodiments, the processor 13 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits packaged with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 13 is the control unit of the computer device 1, connecting various components of the computer device 1 via various interfaces and lines. It executes programs or modules stored in the memory 12 (e.g., executing an air conditioning ventilation control program) and calls data stored in the memory 12 to perform various functions of the computer device 1 and process data.

[0143] The processor 13 executes the operating system of the computer device 1 and various installed application programs. The processor 13 executes these application programs to implement the steps in the various air conditioning air supply control method embodiments described above, for example... Figures 1-3 The steps are shown.

[0144] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 12 and executed by the processor 13 to complete the present invention. The one or more modules / units may be a series of computer-readable instruction segments capable of performing a specific function, which describe the execution process of the computer program in the computer device 1. For example, the computer program may be divided into an acquisition unit 110, a feature engineering unit 111, an input unit 112, a decision unit 113, and a control unit 114.

[0145] The integrated unit implemented as a software functional module described above can be stored in a computer-readable storage medium. This software functional module, stored in a storage medium, includes several instructions to cause a computer device (which may be a personal computer, computer equipment, or network device, etc.) or processor to execute portions of the air conditioning air supply control method described in the various embodiments of the present invention.

[0146] If the modules / units integrated in the computer device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware devices. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above.

[0147] The computer program includes computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory, etc.

[0148] Furthermore, the computer-readable storage medium may primarily include a stored program area and a stored data area, wherein the stored program area may store the operating system, an application program required for at least one function, etc.; and the stored data area may store data created based on the use of blockchain nodes, etc.

[0149] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0150] The bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, in... Figure 5 The bus is represented by only one straight line, but this does not mean that there is only one bus or one type of bus. The bus is configured to enable communication between the memory 12 and at least one processor 13, etc.

[0151] Although not shown, the computer device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 13 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The computer device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0152] Furthermore, the computer device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish a communication connection between the computer device 1 and other computer devices.

[0153] Optionally, the computer device 1 may further include a user interface, which may be a display, an input unit (such as a keyboard), and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the computer device 1 and to display a visual user interface.

[0154] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0155] It will be understood by those skilled in the art that Figure 5 The structure shown does not constitute a limitation on the computer device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0156] Combination Figure 1 The memory 12 in the computer device 1 stores multiple instructions to implement an air conditioning air supply control method, and the processor 13 can execute the multiple instructions to achieve the following: In response to the air supply control command of the target air conditioner in the target area, the system uses multi-source sensors to collect sensor data and obtain user preference data. Feature engineering is performed based on the sensor data and the user preference data to obtain target features; The target features are input into the air supply pattern prediction model trained based on the federated learning algorithm to obtain candidate air supply patterns; The decision engine is invoked to make a decision on the candidate air supply modes, and the target air supply mode is obtained; The target air conditioner is controlled to supply air according to the target air supply mode.

[0157] Specifically, the processor 13's implementation method for the above instructions can be found in [reference needed]. Figures 1-3 The descriptions of the relevant steps in the corresponding embodiments are not repeated here.

[0158] It should be noted that all data involved in this case was legally obtained. Software tools or components not belonging to this company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.

[0159] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0160] This invention can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This invention can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This invention can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0161] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0163] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0164] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0165] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices described in this invention can also be implemented by a single unit or device through software or hardware. Terms such as "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0166] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An air conditioning supply air control method characterized by, The air conditioning air supply control method comprises: In response to the air supply control instruction of the target air conditioner in the target area, sensor data is collected by a multi-source sensor, and user preference data is obtained; Feature engineering is performed according to the sensor data and the user preference data to obtain target features; The target features are input into an air supply mode prediction model trained based on a federated learning algorithm to obtain candidate air supply modes; A decision engine is called to make decisions on the candidate air supply modes to obtain a target air supply mode; The target air conditioner is controlled according to the target air supply mode.

2. The air conditioning supply air control method according to claim 1, wherein The method further comprises the following steps before the feature engineering is performed according to the sensor data and the user preference data: Timestamp alignment and sampling window interpolation alignment are performed on the sensor data and the user preference data to obtain processed data; Kalman filter is used to smooth the point cloud data in the processed data; Non-uniformity correction is performed on the human body surface temperature distribution data in the processed data, and ambient light interference is removed by a moving average filter; A band-pass filter is used to filter the instantaneous fluctuations of the environmental data in the processed data. The method further comprises the following steps before the target features are obtained by performing feature engineering on the sensor data and the user preference data:

3. The air conditioning supply air control method according to claim 2, wherein A density-based spatial clustering algorithm with noise is used to identify human targets in the point cloud data and track the activity trajectories of the human targets; Kalman filter is used to generate human trajectory according to the activity trajectory; Spatial features are constructed according to the human trajectory; Physiological features are constructed according to the human body surface temperature distribution data; Environmental features are constructed according to the environmental data; 4. The air conditioning supply air control method according to claim 3, wherein User features are constructed according to the user preference data; The spatial features, the physiological features, the environmental features and the user features are fused based on a feature fusion strategy to obtain the target features. The method further comprises the following steps before the target features are input into the air supply mode prediction model trained based on the federated learning algorithm: A hybrid model including a short-time sequence prediction branch and a structured feature branch is constructed; wherein the short-time sequence prediction branch and the structured feature branch are output fused by a model fusioner; the short-time sequence prediction branch is used to capture the time-dependent relationship between body surface temperature and behavior; and the structured feature branch is used to model user static preferences; A training set is constructed according to the dimensions of the target features; A loss function is constructed based on a misjudgment penalty mechanism; ​ ​ 5. The air conditioning supply air control method according to claim 1, wherein ​ ​ ​ ​ adopting a federated learning algorithm, training the mixed model on an edge node to which the target air conditioner belongs by using the training set and the loss function, to obtain a local model; encrypting and sending parameters of the local model to a central node, and receiving model parameters fed back by the central node; loading the model parameters on the framework of the local model to obtain the supply mode prediction model.

6. The air conditioning supply air control method according to claim 1, wherein The calling of the decision engine on the candidate supply mode to obtain the target supply mode includes: obtaining a real-time probability of the candidate supply mode and a corresponding trigger threshold; when the real-time probability is greater than the trigger threshold and a safety constraint is met, sending a supply instruction based on the candidate supply mode to the target air conditioner; wherein the safety constraint includes a disabled time period constraint and a user preference constraint; when the real-time probability of the candidate supply mode decreases to a preset probability within a preset time length, determining that a short-term fluctuation event occurs, and sending a rollback instruction to the target air conditioner to roll back to the original mode; or when the real-time probability of the candidate supply mode does not decrease to the preset probability within the preset time length, determining the candidate supply mode as the target supply mode.

7. The air conditioning supply air control method according to claim 1, wherein After the target air conditioner is controlled according to the target supply mode, the method further includes: obtaining user feedback data; optimizing the supply mode prediction model and the decision engine according to the user feedback data.

8. An air conditioning air supply control device, characterized by comprising: The air conditioner supply control device includes: a collection unit configured to collect sensor data by using a multi-source sensor in response to a supply control instruction of a target air conditioner in a target area, and obtain user preference data; a feature engineering unit configured to perform feature engineering according to the sensor data and the user preference data to obtain target features; an input unit configured to input the target features into a supply mode prediction model trained based on a federated learning algorithm to obtain a candidate supply mode; a decision unit configured to call a decision engine to make a decision on the candidate supply mode to obtain a target supply mode; a control unit configured to control the target air conditioner according to the target supply mode.

9. A computer device, comprising: The computer device includes: a memory configured to store at least one instruction; and a processor configured to execute the instruction stored in the memory to implement the air conditioner supply control method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor in a computer device to implement the air conditioner supply control method according to any one of claims 1 to 7.

Images