Intelligent door and window control system

Abstract

The invention discloses an intelligent door and window control system, and relates to the field of intelligent home control. The system comprises a multi-modal data acquisition module, a multi-modal scene perception decision module, a door and window execution module, a user interaction module and a data storage module. The core innovation is a multi-modal scene perception decision-making module, and the multi-modal scene perception decision-making module realizes accurate control through three steps: preprocessing numeric type, behavioral type and state type data and extracting features; a user behavior-environment dynamic coupling model is constructed, the matching degree is calculated through a scene membership degree formula, and five types of core scenes are recognized; and finally, door and window opening and closing and locking instructions are generated according to the scene. The system breaks through the limitation of traditional single parameter decision, can dynamically adapt to user habits and environment changes, gives consideration to safety, practicability and economy, and effectively solves the problems of misjudgment and poor adaptability of a traditional system.

Description

A smart door and window control system Technical Field

[0001] This invention relates to the field of smart home control, specifically to a smart door and window control system. Background Technology

[0002] With the rapid development of the smart home industry, smart doors and windows, as an important component of home security and environmental regulation, are constantly being enriched in their functions. Existing smart door and window control systems mostly rely on data from a single sensor (such as temperature and humidity sensors or infrared human body sensors) to achieve automatic control. For example, when the indoor temperature is detected to be higher than a threshold, the window will automatically open for ventilation, and when a human body is detected approaching, the door will automatically unlock and open.

[0003] However, such systems have obvious limitations: First, data from a single sensor is easily interfered with, leading to misjudgments. For example, on a rainy day, relying solely on a humidity sensor might mistakenly trigger a window-closing command due to the operation of an indoor humidifier. Second, they lack the ability to perceive the overall scene and cannot make accurate decisions by combining multi-dimensional information such as user behavior habits and dynamic changes in the environment. For example, when a user is working in a study, the system cannot distinguish between "briefly leaving" and "going out," and may mistakenly execute window-closing and door-locking operations.

[0004] To address the aforementioned issues, there is an urgent need for an intelligent door and window control system with overall scene perception and precise decision-making capabilities, which can overcome existing technological bottlenecks through innovative perception and decision-making modules. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent door and window control system to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent door and window control system, comprising a multimodal data acquisition module, a multimodal scene perception and decision-making module, a door and window execution module, a user interaction module, and a data storage module; the multimodal data acquisition module is used to collect environmental parameters, behavioral parameters, and state parameters; the multimodal scene perception and decision-making module receives the input data from the multimodal data acquisition module, and outputs control commands after data preprocessing, scene modeling, and dynamic decision-making; the working steps of the multimodal scene perception and decision-making module are as follows: Step 1, Multi-source data preprocessing and feature extraction: The multimodal scene perception and decision-making module uses Z-score standardization to process the numerical data collected by the multimodal data acquisition module, and... Behavioral data is used to extract features such as activity area, dwell time, and operation frequency, and state-based data is converted into binary features; Step 2: Scene modeling based on a coupled model: Construct a coupled model of user behavior features and environmental dynamic features, calculate the scene matching degree using the scene membership formula, and identify the scene type; Step 3: Dynamic decision-making and instruction generation: Generate corresponding door and window opening and closing, locking and closing control instructions according to the scene type; The door and window execution module receives control instructions and executes door and window opening and closing and locking and closing actions; The user interaction module realizes information interaction between the user and the system; The user interaction module includes a mobile APP and an indoor control panel: The mobile APP supports remote viewing of door and window status, manual sending of control instructions, and setting user behavior preferences, i.e., adjusting... The system includes a weighting coefficient function; the indoor control panel uses touch operation and supports one-click switching of scene modes (such as "sleep mode" and "away mode"), and also has a voice prompt function (such as playing "doors and windows will close in 3 seconds, please confirm" when doors and windows are about to close); the data storage module is used to store the data generated by the multimodal data acquisition module, the multimodal scene perception and decision-making module, the door and window execution module, and the user interaction module.

[0007] Preferably, the multimodal data acquisition module includes an environmental sensor group, a behavior sensor group, and a state sensor group; the environmental sensor group includes a rainfall sensor, a wind speed sensor, a PM2.5 sensor, and a temperature and humidity sensor, and is used to collect environmental parameters including but not limited to outdoor rainfall, wind speed, PM2.5 concentration, and indoor and outdoor temperature and humidity; the behavior sensor group includes a millimeter-wave radar and a Bluetooth positioning module, and is used to collect indoor human activity trajectories (achieved through millimeter-wave radar), user mobile phone location information (located through Bluetooth), and behavioral parameters of door and window operation history; the state sensor group includes a photoelectric encoder and a lock state sensor, and is used to collect state parameters such as the current opening and closing status of doors and windows, lock status, and glass integrity.

[0008] Preferably, the specific implementation of the multi-source data preprocessing and feature extraction in step 1 is as follows: First, the raw data transmitted by the multimodal data acquisition module is received, and outlier removal and standardization are performed on the data; for numerical data (such as temperature, humidity, wind speed, PM2.5 concentration), the Z-score standardization formula is used to eliminate the influence of dimensions. ,in For the standardized first Numerical parameters, This is the original value of the parameter. This is the average of the parameter's historical data over the past 7 days. This is the standard deviation of the parameter's historical data for the past 7 days. For behavioral data (such as human activity trajectories and mobile phone locations), a trajectory feature extraction algorithm is used to extract key features, including but not limited to activity area (indoor / outdoor), dwell time (t≥5 minutes is considered a stay, t<5 minutes is considered a pass-through), and operation frequency (number of door and window operations per unit time). For status data (such as the opening and closing status of doors and windows), it is converted into binary features (1 indicates open / unlocked, 0 indicates closed / locked).

[0009] Preferably, the specific implementation of the coupled model-based scene modeling in step 2 is as follows: A coupled model of user behavior features and environmental dynamic features is constructed to achieve automatic identification of scene types; specifically, scene modeling uses user state-environment state as the two-dimensional core dimension, where user state is determined by behavioral features, and environment state is determined by environmental parameters. The coupling relationship between the two is calculated using the scene membership formula. ,in This represents the scene membership value. The value ranges from 0 to 1, with values ​​closer to 1 indicating a higher degree of matching for the scenario. Weighting coefficients The value ranges from 0.4 to 0.6, and is dynamically adjusted based on user habits, such as the user's workdays. ,weekend , Weights for user behavior features Environmental dynamic feature weights; user behavior feature weights The calculation method is as follows: ,in, The number of behavioral features, where These are the activity area, stay duration, and operation frequency, respectively. For the first Weights of each behavioral feature , For the first Normalized scores for each behavioral feature (range 0-1). For example, if the user's phone is located indoors (activity area score). ), stay duration 10 minutes (stay duration score) Nearly one hour of operation without doors or windows (operation frequency score) ),but Environmental dynamic feature weights The calculation method is as follows: ,in Let be the number of environmental characteristics, where let These are rainfall, wind speed, PM2.5 concentration, and indoor-outdoor temperature difference, respectively. For the first Weight of each environmental feature , For the first Normalized scores for each environmental characteristic The value ranges from 0 to 1, with a higher score indicating that the environmental parameter is more suitable for opening windows; for example, a rainfall of 0 mm ( ), wind speed 2m / s ( PM2.5 concentration 30 μg / m³ ), indoor and outdoor temperature difference 5℃ ( ),but Based on scene membership The calculation results pre-determine 5 core scenarios, including: home activity scenarios. Briefly leaving the scene Long-term outdoor scenarios Sleep scenarios Special weather scenarios are defined as rainfall ≥ 5 mm / h or wind speed ≥ 8 m / s, which are directly classified as special weather scenarios and are not subject to other restrictions. Value restrictions.

[0010] Preferably, the specific implementation of the dynamic decision-making and instruction generation in step 3 is as follows: Based on the scenario modeling results in step 2, corresponding door and window control instructions are generated, and the specific decision-making rules are as follows: Home activity scenario: If Control the opening of doors and windows to 30% (balancing ventilation and safety); if Control the opening of doors and windows to 15%; if In the following scenarios: Control doors and windows to close and unlock indoor ventilation mode (linked to the indoor fresh air system); For short-term absence: Control doors and windows to close but not lock, while simultaneously pushing the door and window status to the user's mobile phone in real time. If the user does not respond within 5 minutes, a half-lock operation is automatically performed; For long-term absence: Control doors and windows to close completely and lock, activating door and window security mode; If the status sensor detects abnormal opening of doors and windows, an alarm message is immediately sent through the user interaction module; For sleep scenarios: Control doors and windows to close to 5% opening (ensuring indoor air circulation), reducing the operating noise of the door and window actuator module, and simultaneously linking the indoor lighting system to turn off the main light; For special weather scenarios: Regardless of other current scenario types, immediately control doors and windows to close completely and lock. If an outdoor wind speed ≥12m / s is detected, the preset wind-resistant reinforcement mode of doors and windows is activated (adjusting the locking force of door and window hardware through the door and window actuator module).

[0011] Preferably, the door and window execution module includes a drive unit, a lock control unit, and a status feedback unit: the drive unit adopts a DC brushless motor, and after receiving instructions from the multimodal scene perception and decision module, drives the door and window to perform opening and closing actions, supporting stepless adjustment of 0-100% opening degree; the lock control unit adopts a dual locking structure of electromagnetic lock and mechanical lock, and executes operations including but not limited to unlocking, half-locking, and full-locking according to the decision instructions; the status feedback unit collects the opening degree and locking status information of the door and window in real time, and feeds it back to the multimodal scene perception and decision module to form a control closed loop.

[0012] Preferably, the data storage module adopts a combination of local storage and cloud storage: local storage uses an SD card to store nearly 24 hours of real-time data and control commands to ensure normal system operation in the event of a network outage; cloud storage uses a cloud server to store historical data and decision model parameters for a long time.

[0013] Compared with existing technologies, the beneficial effects of this invention are as follows: The multimodal scene perception and decision-making module of this invention, by constructing a coupled model of "user behavior-environment dynamics," overcomes the limitations of single-parameter decision-making in existing smart doors and windows, achieving accurate identification of complex scenarios and solving the problems of easy misjudgment and poor adaptability in traditional systems. For example, in a scenario where a user is at home but the outdoor PM2.5 concentration suddenly increases, the system can quickly identify environmental changes and automatically adjust the door and window opening from 30% to 5%, taking into account both user activity needs and indoor air quality.

[0014] This invention supports stepless adjustment of door and window opening and dynamic adjustment of decision parameters to adapt to the behavioral habits of different users and the environmental characteristics of different areas; the dual locking structure and security mode enhance home security, and the emergency handling mechanism for special weather scenarios reduces the risk of damage to doors and windows caused by extreme weather.

[0015] This invention employs a multi-sensor collaborative working mode, avoiding the high cost of investing in a single high-precision sensor; the local and cloud collaborative storage method reduces data storage costs while ensuring the stability of system operation. Attached Figure Description

[0016] Figure 1 is a schematic diagram of the overall system structure of the present invention; Figure 2 is a schematic diagram of the workflow of the multimodal scene perception and decision-making module of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please refer to Figures 1-2. This invention provides a technical solution: an intelligent door and window control system, including a multimodal data acquisition module, a multimodal scene perception and decision-making module, a door and window execution module, a user interaction module, and a data storage module; the multimodal data acquisition module is used to collect environmental parameters, behavioral parameters, and state parameters; the multimodal scene perception and decision-making module receives the input data from the multimodal data acquisition module, and outputs control commands after data preprocessing, scene modeling, and dynamic decision-making; the working steps of the multimodal scene perception and decision-making module are as follows: Step 1, multi-source data preprocessing and feature extraction: the multimodal scene perception and decision-making module uses Z-score standardization to process the numerical data collected by the multimodal data acquisition module. Step 1: Extract activity area, dwell time, and operation frequency features from behavioral data; convert state data into binary features. Step 2: Scene modeling based on a coupled model: Construct a coupled model of user behavior features and environmental dynamic features, calculate scene matching degree using scene membership formula, and identify scene type. Step 3: Dynamic decision-making and instruction generation: Generate corresponding door and window opening / closing and locking control instructions according to scene type. The door and window execution module receives control instructions and executes door and window opening / closing and locking actions. The user interaction module realizes information interaction between the user and the system. The user interaction module includes a mobile APP and an indoor control panel: the mobile APP supports remote viewing of door and window status, manual sending of control instructions, and setting user behavior preferences, i.e., adjusting... The system includes a weighting coefficient function; the indoor control panel uses touch operation and supports one-click switching of scene modes (such as "Sleep Mode" and "Away Mode"), and also has a voice prompt function (such as playing "Doors and windows will close in 3 seconds, please confirm" when doors and windows are about to close); the data storage module is used to store data generated by the multimodal data acquisition module, the multimodal scene perception and decision-making module, the door and window execution module, and the user interaction module. The specific content stored in the data storage module is as follows: Historical environmental parameter data: outdoor rainfall, wind speed, PM2.5 concentration, and indoor and outdoor temperature and humidity data collected by the multimodal data acquisition module (such as data collected every 10 seconds for the past 7 days, and data triggered by sudden changes in rainfall), used for numerical data standardization calculations (mean in the formula). Standard deviation (Data source).

[0019] Historical behavioral data, including indoor human activity trajectories (millimeter-wave radar detection results), user mobile phone location (Bluetooth positioning data), and the number of door and window operations per unit time, are used to optimize behavioral feature weights. .

[0020] Status and control command history data: records of door and window opening and closing degrees, lock / unlock status (unlocked / half-locked / fully locked) and corresponding control commands. Data for the past 24 hours is stored on a local SD card, and data for one year is stored in the cloud. Core weight parameters: dynamically adjusted. (Weighting coefficients for user behavior and environmental characteristics, such as 0.5 for weekdays and 0.6 for weekends), weighting of behavioral characteristics. (Activity area 0.5, stay duration 0.3, operation frequency 0.2), environmental feature weights (Rainfall of 0.4 mm).

[0021] Feature calculation parameters: Normalized score standard for behavioral data (e.g., a score of 1 for a stay duration ≥ 5 minutes), normalized score for environmental features. The judgment threshold (e.g., PM2.5≤35μg / m³ score≥0.9).

[0022] Scene determination parameter: Scene membership degree Interval thresholds (such as those for home activities) Special weather trigger thresholds (rainfall ≥ 5 mm / h, etc.).

[0023] Furthermore, the multimodal data acquisition module includes an environmental sensor group, a behavior sensor group, and a status sensor group. The environmental sensor group includes a rainfall sensor, a wind speed sensor, a PM2.5 sensor, and a temperature and humidity sensor. The environmental sensor group is used to collect environmental parameters including, but not limited to, outdoor rainfall, wind speed, PM2.5 concentration, and indoor and outdoor temperature and humidity. The behavior sensor group includes a millimeter-wave radar and a Bluetooth positioning module. The behavior sensor group is used to collect behavioral parameters such as indoor human activity trajectory (achieved through millimeter-wave radar), user mobile phone location information (located through Bluetooth), and door and window operation history. The status sensor group includes a photoelectric encoder and a lock status sensor. The status sensor group is used to collect status parameters such as the current opening and closing status of doors and windows, lock status, and glass integrity.

[0024] Furthermore, the specific implementation of multi-source data preprocessing and feature extraction in step 1 is as follows: First, the raw data transmitted by the multimodal data acquisition module is received, and outlier removal and standardization are performed on the data; for numerical data (such as temperature, humidity, wind speed, and PM2.5 concentration), the Z-score standardization formula is used to eliminate the influence of dimensions. ,in For the standardized first Numerical parameters, This is the original value of the parameter. This is the average of the parameter's historical data over the past 7 days. This is the standard deviation of the parameter's historical data for the past 7 days. For behavioral data (such as human activity trajectories and mobile phone locations), a trajectory feature extraction algorithm is used to extract key features, including but not limited to activity area (indoor / outdoor), dwell time (t≥5 minutes is considered a stay, t<5 minutes is considered a pass-through), and operation frequency (number of door and window operations per unit time). For status data (such as the opening and closing status of doors and windows), it is converted into binary features (1 indicates open / unlocked, 0 indicates closed / locked).

[0025] Furthermore, the specific implementation of scene modeling based on the coupling model in step 2 is as follows: A coupling model of user behavior features and environmental dynamic features is constructed to achieve automatic identification of scene types; specifically, scene modeling uses user state-environment state as the two-dimensional core dimension, where user state is determined by behavioral features, and environment state is determined by environmental parameters. The coupling relationship between the two is calculated using the scene membership formula. ,in This represents the scene membership value. The value ranges from 0 to 1, with values ​​closer to 1 indicating a higher degree of matching for the scenario. Weighting coefficients The value ranges from 0.4 to 0.6, and is dynamically adjusted based on user habits, such as the user's workdays. ,weekend , Weights for user behavior features Environmental dynamic feature weights; user behavior feature weights The calculation method is as follows: ,in, The number of behavioral features, where These are the activity area, stay duration, and operation frequency, respectively. For the first Weights of each behavioral feature , For the first Normalized scores for each behavioral feature (range 0-1). For example, if the user's phone is located indoors (activity area score). ), stay duration 10 minutes (stay duration score) Nearly one hour of operation without doors or windows (operation frequency score) ),but Environmental dynamic feature weights The calculation method is as follows: ,in Let be the number of environmental characteristics, where let These are rainfall, wind speed, PM2.5 concentration, and indoor-outdoor temperature difference, respectively. For the first Weight of each environmental feature , For the first Normalized scores for each environmental characteristic The value ranges from 0 to 1, with a higher score indicating that the environmental parameter is more suitable for opening windows; for example, a rainfall of 0 mm ( ), wind speed 2m / s ( PM2.5 concentration 30 μg / m³ ), indoor and outdoor temperature difference 5℃ ( ),but Based on scene membership The calculation results pre-determine 5 core scenarios, including: home activity scenarios. Briefly leaving the scene Long-term outdoor scenarios Sleep scenarios Special weather scenarios are defined as rainfall ≥ 5 mm / h or wind speed ≥ 8 m / s, which are directly classified as special weather scenarios and are not subject to other restrictions. Value restrictions.

[0026] Furthermore, the specific implementation of dynamic decision-making and instruction generation in step 3 is as follows: Based on the scenario modeling results in step 2, corresponding door and window control instructions are generated. The specific decision-making rules are as follows: Home activity scenario: If Control the opening of doors and windows to 30% (balancing ventilation and safety); if Control the opening of doors and windows to 15%; if In the following scenarios: Control doors and windows to close and unlock indoor ventilation mode (linked to the indoor fresh air system); For short-term absence: Control doors and windows to close but not lock, and simultaneously push the door and window status to the user's mobile phone in real time. If the user does not respond within 5 minutes, automatically perform a half-lock operation; For long-term absence: Control doors and windows to close completely and lock, and activate door and window security mode; If the status sensor detects abnormal opening of doors and windows, immediately send an alarm message through the user interaction module); For sleep scenarios: Control doors and windows to close to 5% opening (ensuring indoor air circulation), reduce the operating noise of the door and window actuator module, and simultaneously link the indoor lighting system to turn off the main light; For special weather scenarios: Regardless of other current scenario types, immediately control doors and windows to close completely and lock. If an outdoor wind speed of ≥12m / s is detected, the preset wind-resistant reinforcement mode for doors and windows is activated (the locking force of the door and window hardware is adjusted through the door and window execution module). Furthermore, the door and window execution module includes a drive unit, a lock control unit, and a status feedback unit: the drive unit uses a DC brushless motor, which drives the doors and windows to open and close after receiving instructions from the multimodal scene perception and decision module, supporting stepless adjustment of the opening degree from 0-100%; the lock control unit adopts a dual locking structure of electromagnetic lock and mechanical lock, and executes operations including but not limited to unlocking, half-locking, and full-locking according to the decision instructions; the status feedback unit collects the opening degree and locking status information of the doors and windows in real time, and feeds it back to the multimodal scene perception and decision module to form a control closed loop.

[0027] Furthermore, the data storage module adopts a combination of local storage and cloud storage: local storage uses SD cards to store nearly 24 hours of real-time data and control commands to ensure normal system operation in the event of a network outage; cloud storage uses cloud servers to store historical data and decision model parameters for a long time.

[0028] The specific operation flow of this invention is as follows: Initialization phase: After the system is powered on, each module completes self-test, and the multimodal scene perception and decision module loads historical parameters (such as...) from the data storage module. Weighting coefficients, behavioral feature weights Users can complete initial preference settings (such as "weekday / weekend mode" selection) through a mobile app or indoor control panel.

[0029] Data acquisition phase: The multimodal data acquisition module collects various types of data in the manner of "timed acquisition (10-second cycle) + triggered acquisition (such as when rainfall changes suddenly)" and transmits the data to the multimodal scene perception and decision-making module via the MQTT protocol.

[0030] Scene decision-making stage: The multimodal scene perception and decision-making module processes data according to the steps of "data preprocessing - scene modeling - instruction generation", and after generating control instructions, it sends them to the door and window execution module via RS485 bus.

[0031] Execution and Feedback Phase: The door and window execution module receives instructions and executes corresponding actions. The status feedback unit feeds back the execution results to the multimodal scene perception and decision-making module, and pushes door and window status information to the user through the user interaction module.

[0032] Parameter optimization phase: As an optimization method of this invention, the cloud server can analyze the historical data of the previous day at 2:00 AM every day and optimize it using the gradient descent algorithm. Weighting coefficients and The weights and optimized parameters are synchronized to the local system's data storage module via the network.

[0033] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent door and window control system, characterized in that, It includes a multimodal data acquisition module, a multimodal scene perception and decision-making module, a door and window execution module, a user interaction module, and a data storage module. The multimodal data acquisition module is used to collect environmental parameters, behavioral parameters, and state parameters. The multimodal scene perception and decision-making module receives the input data from the multimodal data acquisition module, performs data preprocessing, scene modeling, and dynamic decision-making, and then outputs control commands. The working steps of the multimodal scene perception and decision-making module are as follows: Step 1, Multi-source data preprocessing and feature extraction: The multimodal scene perception and decision-making module uses Z-score standardization to process the numerical data collected by the multimodal data acquisition module, and extracts activity areas and dwell times from the behavioral data. Step 1: Operation frequency characteristics; convert state-type data into binary characteristics. Step 2: Scene modeling based on a coupled model: construct a coupled model of user behavior characteristics and environmental dynamic characteristics, calculate scene matching degree using the scene membership formula, and identify scene types. Step 3: Dynamic decision-making and instruction generation: generate corresponding door / window opening / closing and locking control instructions according to the scene type. The door / window execution module receives control instructions and executes door / window opening / closing and locking actions. The user interaction module enables information interaction between the user and the system. The user interaction module includes a mobile APP and an indoor control panel: the mobile APP supports remote viewing of door / window status, manual sending of control instructions, and setting user behavior preferences, i.e., adjusting... Weighting coefficient function; the indoor control panel adopts touch operation, supports one-click switching of scene modes, and also has voice prompt function; the data storage module is used to store the data generated by the multimodal data acquisition module, the multimodal scene perception and decision-making module, the door and window execution module and the user interaction module.

2. The intelligent door and window control system according to claim 1, characterized in that: The multimodal data acquisition module includes an environmental sensor group, a behavior sensor group, and a status sensor group. The environmental sensor group includes a rainfall sensor, a wind speed sensor, a PM2.5 sensor, and a temperature and humidity sensor. The environmental sensor group is used to collect environmental parameters including, but not limited to, outdoor rainfall, wind speed, PM2.5 concentration, and indoor and outdoor temperature and humidity. The behavior sensor group includes a millimeter-wave radar and a Bluetooth positioning module. The behavior sensor group is used to collect behavioral parameters such as indoor human activity trajectories, user mobile phone location information, and door and window operation history. The status sensor group includes a photoelectric encoder and a lock status sensor. The status sensor group is used to collect status parameters such as the current opening and closing status of doors and windows, lock status, and glass integrity.

3. The intelligent door and window control system according to claim 1, characterized in that: The specific implementation of multi-source data preprocessing and feature extraction in step 1 is as follows: First, the raw data transmitted by the multimodal data acquisition module is received, and outlier removal and standardization are performed on the data; for numerical data, the Z-score standardization formula is used to eliminate the influence of dimensions. ,in For the standardized first Numerical parameters, This is the original value of the parameter. This is the average of the parameter's historical data over the past 7 days. This represents the standard deviation of the parameter's historical data over the past 7 days. For behavioral data, a trajectory feature extraction algorithm is used to extract key features, including but not limited to activity area, dwell time, and operation frequency. For state-related data, it is converted into binary features.

4. The intelligent door and window control system according to claim 1, characterized in that: The specific implementation of the coupled model-based scene modeling in step 2 is as follows: A coupled model of user behavior features and environmental dynamic features is constructed to achieve automatic scene type identification; the specific scene modeling uses user state-environment state as the two-dimensional core dimension, where user state is determined by behavioral features, and environment state is determined by environmental parameters. The coupling relationship between the two is calculated using the scene membership formula. ,in This represents the scene membership value. The value ranges from 0 to 1, with values ​​closer to 1 indicating a higher degree of matching for the scenario. Weighting coefficients The value range is 0.4-0.6, and it will be dynamically adjusted according to user habits. Weights for user behavior features Weights for dynamic environmental features; User behavior feature weights The calculation method is as follows: ,in, The number of behavioral features, where These are the activity area, stay duration, and operation frequency, respectively. For the first Weights of each behavioral feature ， For the first Normalized scores of individual behavioral features; weights of environmental dynamic features The calculation method is as follows: ,in Let be the number of environmental characteristics, where let These are rainfall, wind speed, PM2.5 concentration, and indoor-outdoor temperature difference, respectively. For the first Weight of each environmental feature ， For the first Normalized scores for each environmental characteristic The value ranges from 0 to 1, with a higher score indicating that the environmental parameter is more suitable for window opening; based on scene membership degree The calculation results pre-determine 5 core scenarios, including: home activity scenarios. Briefly leaving the scene Long-term outdoor scenarios Sleep scenarios Special weather scenarios are defined as rainfall ≥ 5 mm / h or wind speed ≥ 8 m / s, which are directly classified as special weather scenarios and are not subject to other restrictions. Value restrictions.

5. The intelligent door and window control system according to claim 1, characterized in that: The specific implementation of step 3, dynamic decision-making and instruction generation, is as follows: Based on the scenario modeling results from step 2, corresponding door and window control instructions are generated. The specific decision-making rules are as follows: Home activity scenario: If Control the opening of doors and windows to 30%; like Control the opening of doors and windows to 15%; like In the following scenarios: Control doors and windows to close and unlock indoor ventilation mode; For short-term absence: Control doors and windows to close but not lock, while simultaneously pushing door and window status to the user's mobile phone in real time. If the user does not respond within 5 minutes, automatically perform a half-lock operation; For long-term absence: Control doors and windows to close completely and lock, activating door and window security mode; If the status sensor detects abnormal opening of doors and windows, immediately send alarm information through the user interaction module; For sleep scenarios: Control doors and windows to close to 5% opening, reducing the operating noise of the door and window execution module, and simultaneously linking the indoor lighting system to turn off the main light; For special weather scenarios: Regardless of other current scenario types, immediately control doors and windows to close completely and lock. If an outdoor wind speed ≥12m / s is detected, activate the preset wind-resistant reinforcement mode for doors and windows.

6. The intelligent door and window control system according to claim 1, characterized in that: The door and window execution module includes a drive unit, a lock control unit, and a status feedback unit: the drive unit uses a DC brushless motor, which, after receiving instructions from the multimodal scene perception and decision module, drives the door and window to open and close, supporting stepless adjustment of the opening degree from 0-100%; the lock control unit adopts a dual locking structure of electromagnetic lock and mechanical lock, and executes operations including but not limited to unlocking, half-locking, and full-locking according to the decision instructions; the status feedback unit collects the opening degree and locking status information of the door and window in real time, and feeds it back to the multimodal scene perception and decision module to form a control closed loop.

7. The intelligent door and window control system according to claim 1, characterized in that: The data storage module adopts a combination of local storage and cloud storage: local storage uses SD cards to store nearly 24 hours of real-time data and control commands to ensure normal system operation in the event of a network outage; cloud storage uses cloud servers to store historical data and decision model parameters for a long time.

Images