Intelligent door body operation algorithm with AI adaptive regulation

Abstract

This invention discloses an AI-adaptive control algorithm for intelligent gate operation, belonging to the field of intelligent gate control technology. The algorithm includes: constructing a multimodal perception fusion module to simultaneously collect gate's own operating status data, gate area environmental perception data, and user behavior characteristic data; establishing a gate operating health assessment model to dynamically calculate the gate's current health index; generating a user behavior intent prediction model to predict the user's passage intention and mode; constructing a hierarchical adaptive decision engine to generate an optimal gate operation control strategy by combining the gate health index, user behavior intent prediction results, and environmental perception data; and executing closed-loop operation control, driving gate operation according to the control strategy and providing real-time feedback adjustments. This invention can monitor the gate's operating status in real time, accurately predict user behavior intent, and dynamically adjust gate operating parameters, improving the safety and reliability of gate operation. It is applicable to various types of intelligent gate control systems.

Description

Technical Field

[0001] This invention relates to an AI-adaptive control intelligent door operation algorithm, which falls under the field of intelligent door control technology. Specifically, it relates to an intelligent door operation algorithm based on artificial intelligence technology, which can realize real-time monitoring of door operation status, accurate prediction of user behavior intentions, and adaptive control of door operation parameters. It is applicable to door control systems of various types of electric doors, automatic doors, intelligent entrance doors, and industrial doors. Background Technology

[0002] With the rapid development of artificial intelligence and Internet of Things (IoT) technologies, intelligent door control systems have been widely applied in residential buildings, commercial buildings, industrial plants, and public facilities. Intelligent doors integrate various sensors, controllers, and actuators to achieve functions such as automatic opening and closing, anti-pinch protection, and remote control, greatly improving the convenience and security of door use. Currently, most intelligent doors on the market adopt a control method based on preset programs. Parameters such as the door's operating speed, acceleration, and opening angle are preset during installation and debugging and remain largely unchanged during subsequent use.

[0003] The existing technology has the following problems: 1. Existing algorithms lack the ability to monitor and dynamically evaluate the operating status of the door itself in real time, and cannot detect problems such as wear and aging of door components in a timely manner. As the usage time increases, the operating performance of the door will gradually decline, and may even malfunction, affecting the normal use and safety of the door. 2. Most existing algorithms adopt reactive control, which only activates the gate when the user enters the sensing area or triggers the sensor. It cannot predict the user's intention to pass in advance, resulting in a delayed gate response, which affects the user experience. At the same time, it is prone to false triggering or missed triggering in complex scenarios. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an AI adaptive control intelligent door operation algorithm. This algorithm constructs a multimodal perception fusion module to simultaneously collect door operation status data, door area environmental perception data, and user behavior feature data. It establishes a door operation health assessment model and a user behavior intention prediction model, and combines a hierarchical adaptive decision engine to achieve adaptive control of door operation parameters. This solves the problems of the inability to assess door operation status in real time, inaccurate prediction of user intention, and inability to dynamically adjust door operation parameters in existing technologies.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The AI-adaptive intelligent gate operation algorithm includes a multimodal perception fusion module, a gate operation health assessment module, a user behavior intent prediction module, a hierarchical adaptive decision engine, and a closed-loop operation control module.

[0006] The multimodal perception fusion module is used to synchronously collect data on the gate's own operating status, the gate area's environmental perception data, and user behavior feature data. It performs spatiotemporal alignment and feature-level fusion on these heterogeneous data to extract fused feature vectors. The multimodal perception fusion module includes a gate status perception unit, an environmental perception unit, and a user behavior perception unit. The gate status perception unit collects data such as motor current, voltage, speed, temperature, and gate position, speed, and acceleration through various sensors installed on the gate motor, transmission mechanism, and gate itself, comprehensively reflecting the gate's operating status. The environmental perception unit collects environmental data such as obstacles, light intensity, temperature, humidity, and wind force in the gate area through devices such as millimeter-wave radar, infrared sensors, temperature and humidity sensors, and wind force, providing a basis for environmental adaptation of the gate's operation. The user behavior perception unit collects data such as user position, movement speed, posture, orientation, and biometrics through devices such as cameras, depth cameras, and human body sensors, for predicting user behavior intentions.

[0007] The gate health assessment module is used to dynamically calculate the current health index of the gate based on historical operating data and real-time collected data. The gate health assessment model is constructed through a deep learning network. It uses historical operating data of the gate under different health conditions as training samples to learn the operating characteristic patterns of each component of the gate under different conditions such as normal operation, slight wear, moderate wear, and severe wear. In actual operation, the real-time collected gate operating data is input into the trained deep learning network. The network outputs the wear degree, aging degree, and failure risk probability of each component of the gate. These indicators are combined to obtain the overall health index of the gate. The health index adopts a scoring method of 0-100 points. The higher the score, the better the operating condition of the gate. The lower the score, the more serious the problem of the gate.

[0008] The user behavior intent prediction module is used to predict a user's travel intent and mode of travel through temporal analysis and semantic understanding of user behavior feature data. The user behavior intent prediction model adopts an architecture that combines temporal convolutional networks and attention mechanisms, which can effectively capture the temporal change patterns of user behavior feature data. First, the raw data collected by the user behavior perception unit is preprocessed to extract key features such as the user's location, movement speed, posture, and orientation. Then, these features are input into the temporal convolutional network to extract the temporal features of user behavior. Next, the features at different times are weighted through the attention mechanism to highlight the features that are more important for judging user intent. Finally, the weighted features are input into the classifier and combined with the historical user behavior database to predict whether the user needs to travel, the direction of travel, the speed of travel, and whether the user is carrying items or accompanying others.

[0009] The hierarchical adaptive decision engine combines the door health index, user behavior intention prediction results, and environmental perception data to generate the optimal door operation control strategy. The hierarchical adaptive decision engine includes a basic decision layer, a health compensation layer, and an environment adaptation layer. The basic decision layer generates a basic operation control strategy based on the user behavior intention prediction results, including basic parameters such as door opening time, opening speed, opening angle, and closing time. The health compensation layer compensates for the parameters of the basic control strategy based on the door health index. When the door health index is high, the basic control parameters remain unchanged; when the door health index decreases, the door operating speed and acceleration are appropriately reduced, and the anti-pinch sensitivity is increased to ensure the safety and stability of door operation. The environment adaptation layer further optimizes and adjusts the control strategy based on environmental perception data, reducing the door opening speed and increasing wind damping in strong wind environments, extending the door's open time in densely populated environments, and automatically turning on the door area lighting in low-light environments.

[0010] The closed-loop operation control module is used to drive the gate actuator according to the generated control strategy and provide real-time feedback on the operating status to dynamically adjust the control parameters. The closed-loop operation control module adopts the PID control algorithm, which collects the operating status data of the gate actuator in real time, compares it with the preset operating target, calculates the control deviation, and dynamically adjusts the control parameters to achieve precise control of the gate operation. At the same time, the actual operating data is fed back to the gate operation health assessment model and the user behavior intention prediction model to realize online updating and optimization of the model and continuously improve the accuracy and adaptability of the model.

[0011] In addition, the present invention also includes an anomaly handling and security protection mechanism. When an abnormality in the door operation, obstruction by an obstacle, or user safety risk is detected, a safety protection action is immediately triggered to stop the door operation or reverse the operation, and the abnormal event information is recorded. At the same time, an alarm information is sent to the management platform. The anomaly handling and security protection mechanism adopts a multi-redundancy design, including hardware redundancy and software redundancy, to ensure that the system can still work normally and protect user safety in the event of failure of a single sensor or controller.

[0012] Due to the adoption of the above technical solution, the technical progress achieved by this invention compared to the prior art is as follows: 1. This invention constructs a multimodal perception fusion module to simultaneously collect data on the door's own operating status, the door area's environmental perception data, and user behavior characteristics, thereby achieving comprehensive perception of the door's operating environment and user behavior. This provides a rich data foundation for the door's adaptive control and improves the system's perception capability and accuracy.

[0013] 2. This invention establishes a door operation health assessment model, which can monitor the operating status of each component of the door in real time, dynamically calculate the door's health index, promptly detect problems such as wear and aging of the door, and automatically adjust the door's operating parameters according to the health index, thereby extending the door's service life and improving the reliability and safety of the door's operation.

[0014] 3. This invention generates a user behavior intent prediction model. Through temporal analysis and semantic understanding of user behavior feature data, it can predict the user's passage intent and passage method in advance, realize the predictive control of the gate, shorten the gate's response time, improve the user experience, and reduce false triggering and missed triggering.

[0015] 4. This invention constructs a hierarchical adaptive decision engine, which adopts a three-layer architecture of basic decision layer, health compensation layer and environment adaptation layer. It can combine the door health index, user behavior intention prediction results and environmental perception data to generate the optimal door operation control strategy, realize the adaptive adjustment of door operation parameters, and improve the adaptability of the door to different operating environments and usage scenarios.

[0016] 5. This invention adopts a closed-loop operation control mechanism, which realizes precise control of the gate operation by providing real-time feedback on the gate's operating status and dynamically adjusting control parameters, thereby improving the stability and accuracy of the gate's operation. At the same time, it feeds back the actual operating data to the evaluation and prediction model, enabling online updates and optimization of the model and continuously improving the system's performance.

[0017] 6. This invention sets up an anomaly handling and security protection mechanism, adopts a multi-redundancy design, and can detect and handle various abnormal situations during the operation of the door in a timely manner, ensuring the personal safety of users and the safety of the door equipment, and improving the reliability and stability of the system.

[0018] 7. The algorithm of this invention has good versatility and scalability, and can be applied to various types of door control systems such as electric doors, automatic doors, smart entrance doors and industrial doors. It can be adapted to different door models and usage scenarios through simple parameter adjustments, and has broad application prospects. Attached Figure Description

[0019] Figure 1 This is a schematic block diagram illustrating the overall architecture of the present invention; Figure 2 This is a schematic block diagram of the multimodal sensing fusion module of the present invention; Figure 3 This is a network architecture block diagram of the door operation health assessment model of the present invention; Figure 4 This is a network architecture diagram of the user behavior intent prediction model of the present invention; Figure 5 This is a flowchart illustrating the workflow of the hierarchical adaptive decision engine of this invention. Figure 6 This is a block diagram illustrating the control principle of the closed-loop operation control module of the present invention. Figure 7 This is a flowchart illustrating the workflow of the anomaly handling and security assurance mechanism of the present invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the implementation scheme and embodiments: Implementation Plan like Figure 1 As shown, this invention provides an AI-adaptive control intelligent gate operation algorithm, including a multimodal perception fusion module, a gate operation health assessment module, a user behavior intent prediction module, a hierarchical adaptive decision engine, and a closed-loop operation control module. The multimodal perception fusion module is connected to the gate operation health assessment module, the user behavior intent prediction module, and the hierarchical adaptive decision engine, respectively. The gate operation health assessment module and the user behavior intent prediction module are connected to the hierarchical adaptive decision engine, respectively. The hierarchical adaptive decision engine is connected to the closed-loop operation control module, and the closed-loop operation control module is connected to both the gate operation health assessment module and the user behavior intent prediction module, forming a closed-loop feedback system.

[0021] like Figure 2As shown, the multimodal perception fusion module includes a door state perception unit, an environment perception unit, and a user behavior perception unit. The door state perception unit includes a current sensor, a voltage sensor, a speed sensor, a temperature sensor, a position sensor, a speed sensor, and an acceleration sensor, which are used to collect motor current, voltage, speed, temperature, and door position, speed, and acceleration data, respectively. The environment perception unit includes a millimeter-wave radar, an infrared sensor, a temperature and humidity sensor, a wind sensor, and a light sensor, which are used to collect door area obstacle, light intensity, temperature and humidity, and wind data, respectively. The user behavior perception unit includes a depth camera, an RGB camera, a human body sensor, and a biometric recognition sensor, which are used to collect user position, movement speed, posture, orientation, and biometric data, respectively. The multimodal perception fusion module also includes a data preprocessing unit and a feature fusion unit. The data preprocessing unit is used to perform preprocessing operations such as filtering, denoising, and normalization on the raw data collected by each sensor. The feature fusion unit is used to perform spatiotemporal alignment and feature-level fusion on the preprocessed data to extract the fused feature vector.

[0022] like Figure 3 As shown, the gate operation health assessment model adopts a deep residual network architecture, including an input layer, multiple residual blocks, a global average pooling layer, a fully connected layer, and an output layer. The input layer receives the gate operation status fusion feature vector output by the multimodal perception fusion module. The residual blocks are used to extract deep features of the gate operation status. The global average pooling layer is used to perform dimensionality reduction on the feature map. The fully connected layer is used to perform further nonlinear transformation on the features. The output layer outputs the wear degree, aging degree, and failure risk probability of each component of the gate, and comprehensively obtains the overall health index of the gate.

[0023] like Figure 4 As shown, the user behavior intent prediction model adopts an architecture combining temporal convolutional networks and an attention mechanism. It includes an input layer, multiple temporal convolutional layers, an attention mechanism layer, a fully connected layer, and an output layer. The input layer receives the user behavior feature fusion vector output by the multimodal perception fusion module. The temporal convolutional layers extract the temporal features of user behavior. The attention mechanism layer weights the features at different times, highlighting features more important for user intent judgment. The fully connected layer performs further nonlinear transformations on the features. The output layer outputs the predicted results, such as the user's travel intent, travel direction, travel speed, and whether they are carrying items or accompanied by others. like Figure 5 As shown, the workflow of the hierarchical adaptive decision engine includes the following steps: a. Receive user behavior intent prediction results and generate basic operation control strategies; b. Receive the gate health index and compensate the basic control strategy for health. c. Receive environmental perception data and optimize the compensated control strategy for environmental adaptation. d. Generate the final gate operation control strategy and output it to the closed-loop operation control module.

[0024] like Figure 6 As shown, the control principle of the closed-loop operation control module is as follows: First, it receives the gate operation control strategy output by the hierarchical adaptive decision engine, sets the target parameters for gate operation, and then drives the gate actuator to operate. At the same time, the actual operating status data of the gate is collected in real time through the gate status sensing unit. The actual operating status data is compared with the target parameters to calculate the control deviation. The control parameters are dynamically adjusted according to the control deviation by the PID controller and output to the gate actuator to achieve precise control of the gate operation. At the same time, the actual operating data is fed back to the gate operation health assessment module and the user behavior intention prediction module to realize online updating and optimization of the model.

[0025] like Figure 7 As shown, the workflow of the exception handling and security assurance mechanism includes the following steps: a. Real-time monitoring of door operation status, door area environment, and user behavior; b. Determine if there are any abnormal situations, including abnormal door operation, obstruction by obstacles, and user safety risks; c. If any abnormal situation occurs, the safety protection action will be triggered immediately to stop the door from running or to reverse its operation; d. Record abnormal event information, including the time of occurrence, type of abnormality, and abnormal data; e. Send alarm information to the management platform to notify administrators to handle it promptly.

[0026] The working principle of this implementation scheme is as follows: After the system starts, the multimodal perception fusion module synchronously collects data on the gate's own operating status, the gate area's environmental perception data, and user behavior characteristics. This data is preprocessed and fused to extract a fused feature vector. The gate's operational health assessment module receives this fused feature vector and uses a trained deep learning network to calculate the wear and tear, aging, and failure probability of each component of the gate, ultimately obtaining an overall gate health index. The user behavior intent prediction module receives the fused user behavior feature vector and, using a model combining a temporal convolutional network and an attention mechanism, analyzes the temporal changes in user behavior to predict the user's passage intent and mode of transport.

[0027] The hierarchical adaptive decision engine receives the gate health index, the user behavior intention prediction result, and the environmental perception data. First, it generates a basic operation control strategy based on the user behavior intention prediction result. Then, it performs parameter compensation on the basic control strategy based on the gate health index. Finally, it performs environmental adaptation optimization on the compensated control strategy based on the environmental perception data, and finally generates the optimal gate operation control strategy.

[0028] The closed-loop operation control module receives the control strategy output by the hierarchical adaptive decision engine, drives the gate actuator to operate, and simultaneously collects the actual operating status data of the gate in real time. It compares the data with the preset operating target, calculates the control deviation, and dynamically adjusts the control parameters through the PID controller to achieve precise control of the gate operation. At the same time, the actual operating data is fed back to the gate operation health assessment model and the user behavior intention prediction model to realize online updates and optimization of the models, continuously improving the accuracy and adaptability of the models.

[0029] During the operation of the gate, the anomaly handling and safety protection mechanism monitors the gate's operating status, the gate area environment, and user behavior in real time. When an anomaly is detected, it immediately triggers safety protection actions, stops the gate's operation or reverses its operation, records the abnormal event information, and sends alarm information to the management platform to ensure the personal safety of users and the safety of the gate equipment.

[0030] This embodiment achieves AI adaptive control of door operation by constructing a multimodal perception fusion module, a door operation health assessment module, a user behavior intent prediction module, a hierarchical adaptive decision engine, and a closed-loop operation control module. It can monitor the door operation status in real time, accurately predict user behavior intent, and dynamically adjust door operation parameters, thereby improving the safety, reliability, and user experience of door operation.

[0031] This embodiment achieves comprehensive perception of the gate's operating environment and user behavior through multimodal perception fusion, real-time monitoring of the gate's own status through gate health assessment, predictive control of the gate through user behavior intent prediction, optimal control of gate operating parameters through a hierarchical adaptive decision engine, and precise control of gate operation and online optimization of the model through closed-loop operation control. This algorithm has good versatility and scalability, and can be applied to various types of gate control systems. It effectively solves the problems existing in the prior art and has important practical application value. Example 1

[0032] This embodiment takes a smart entrance door for a residential building as an example to illustrate in detail the specific application of the AI ​​adaptive control algorithm for smart door operation of the present invention.

[0033] The residential intelligent entrance door system includes a door body, a motor drive mechanism, a transmission mechanism, multimodal sensing sensors, a controller, and a power module. The multimodal sensing sensors include a current sensor, a voltage sensor, a speed sensor, and a temperature sensor installed on the motor; a position sensor, a speed sensor, and an acceleration sensor installed on the door body; a millimeter-wave radar, an infrared sensor, a temperature and humidity sensor, and a light sensor installed on the door frame; and a depth camera and an RGB camera installed above the door body. The controller incorporates the AI ​​adaptive control algorithm of this invention, which is responsible for processing sensor data, generating control strategies, and driving the motor.

[0034] After system installation and debugging, initial model training is performed first. Operating data of the door under normal conditions is collected, including motor current, voltage, speed, temperature, and door position, speed, and acceleration, serving as initial training samples for the door's operational health assessment model. Simultaneously, behavioral data of different users in different scenarios is collected, including user position, speed, posture, and orientation during the processes of approaching, preparing to open, passing through, and leaving the door, serving as initial training samples for the user behavior intent prediction model. Through offline training, the initial door operational health assessment model and user behavior intent prediction model are obtained.

[0035] In practical use, the multimodal perception fusion module simultaneously collects door operation status data, environmental perception data, and user behavior data. When a user walks from inside the room to the door to leave, the depth camera and RGB camera collect the user's image data and extract features such as the user's position, movement speed, posture, and orientation. The user behavior intent prediction model analyzes these features, predicts that the user intends to leave, and predicts that the user will arrive at the door in 3 seconds based on the user's movement speed.

[0036] After receiving the user intent prediction result, the hierarchical adaptive decision engine generates a basic control strategy, setting the door to open in 2 seconds, with an opening speed of 0.5m / s, an opening angle of 90 degrees, and an opening time of 5 seconds. At the same time, the door health assessment module calculates the current health index of the door based on the real-time collected door operation data, which is 85 points, indicating a good condition. Therefore, the health compensation layer does not need to adjust the basic control strategy. The environmental perception unit detects that the current indoor and outdoor temperature is suitable, the wind is weak, and the lighting is sufficient. Therefore, the environmental adaptation layer also does not need to adjust the control strategy. The final generated control strategy is consistent with the basic control strategy.

[0037] The closed-loop operation control module drives the motor according to the control strategy, and the door opens at the set speed and angle. During the door opening process, the position sensor and speed sensor collect the door's operating status data in real time and compare it with the preset operating target. The PID controller dynamically adjusts the motor's drive voltage and current to ensure that the door operates smoothly and accurately. When the user passes through the door, the user behavior intention prediction model detects that the user has left the door area, and the hierarchical adaptive decision engine generates a closing control strategy, and the door closes smoothly at the set speed.

[0038] During use, as the door's operating time increases, the transmission mechanism will experience some wear. The door's operational health assessment model analyzes real-time collected data such as motor current and speed, finding that the motor current increases and the speed decreases under the same load, indicating slight wear in the transmission mechanism. The door's health index drops to 70 points. At this point, the health compensation layer adjusts the door's operation control strategy, reducing the opening and closing speeds to 0.4 m / s while increasing anti-pinch sensitivity to ensure the door's safety and stability. Simultaneously, the system generates maintenance reminders and pushes them to the user via a mobile app, reminding them to perform timely door maintenance.

[0039] When encountering strong winds, the wind sensor in the environmental perception unit detects that the outdoor wind force is strong, reaching level 5. The environmental adaptation layer adjusts the door operation control strategy, further reducing the door opening speed to 0.3m / s, while increasing the output torque of the motor to improve the door's wind resistance and prevent the door from being blown by strong winds and causing collisions or damage.

[0040] When an obstacle is detected during the closing process of the door, the infrared sensor and millimeter-wave radar simultaneously detect the presence of an obstacle in the door area. The anomaly handling and safety protection mechanism immediately triggers the safety protection action, stops the door from closing and reverses to the fully open state, and at the same time issues a voice prompt "Obstacle detected, please clear it before closing the door". After the obstacle is cleared, the door automatically closes again.

[0041] This embodiment utilizes an AI-adaptive intelligent door operation algorithm. The intelligent entrance door can monitor its own operating status in real time, promptly detect component wear and tear and automatically adjust operating parameters. It can accurately predict the user's passage intentions, start door operation in advance to shorten response time, automatically adjust door operation strategies according to environmental changes to adapt to different usage scenarios, and promptly detect and handle abnormal situations to ensure user safety. This algorithm effectively improves the intelligence level and user experience of residential intelligent entrance doors, demonstrating excellent application results. Example 2

[0042] This embodiment uses an automatic sliding door in a commercial building as an example to illustrate in detail the specific application of the AI ​​adaptive control intelligent door operation algorithm of the present invention.

[0043] The automatic sliding door system for commercial buildings includes a double-leaf sliding door, two motor drive mechanisms, a synchronous transmission mechanism, multimodal sensing sensors, a main controller, a backup controller, and a power module. The multimodal sensing sensors include current sensors, voltage sensors, speed sensors, and temperature sensors installed on the two motors; position sensors, speed sensors, and acceleration sensors installed on the door; two sets of millimeter-wave radars, multiple sets of infrared beam sensors, temperature and humidity sensors, and wind sensors installed above the door frame; and depth cameras and RGB cameras installed on both sides of the door area. The main controller incorporates the AI ​​adaptive control algorithm of this invention and is responsible for the main control tasks. The backup controller serves as a redundancy backup, taking over the control tasks when the main controller fails, ensuring continuous operation of the system.

[0044] After the system installation and debugging are completed, the initial training of the model is carried out. Operational data of the door under different load conditions (empty, half-loaded, full-loaded, etc.) and under different health states are collected as training samples for the door operation health assessment model. Simultaneously, a large amount of passage data from different groups of people in different scenarios is collected, including single-person passage, multi-person parallel passage, passage with luggage, passage with strollers, and passage in wheelchairs, as training samples for the user behavior intent prediction model. Through offline training, a door operation health assessment model and a user behavior intent prediction model suitable for automatic sliding doors in commercial buildings are obtained.

[0045] In the daily operation of commercial buildings, the multimodal perception fusion module continuously and synchronously collects door operation status data, environmental perception data, and user behavior data. During periods of low foot traffic, when a user approaches the door, depth cameras and RGB cameras collect image data of the user, extracting features such as the user's position, movement speed, posture, and orientation. The user behavior intent prediction model analyzes these features to predict the user's intention and mode of passage. If it is predicted that the user is a single person passing through normally, the hierarchical adaptive decision engine generates a basic control strategy, setting the door opening speed to 0.6 m / s, the opening width to 1.2 meters, and the opening time to 3 seconds.

[0046] When multiple people are detected passing through in parallel, the user behavior intent prediction model predicts the number of people and the required passage width based on the location and movement trajectory of the multiple people. If it is predicted that 3 people are passing through in parallel, the required passage width is 1.8 meters. The hierarchical adaptive decision engine adjusts the door opening width to 1.8 meters and extends the opening time to 5 seconds to ensure that multiple people can pass through smoothly.

[0047] When a user pushing a stroller or sitting in a wheelchair is detected, the user behavior intent prediction model analyzes the user's posture and movement characteristics to identify the user's special passage needs. The hierarchical adaptive decision engine adjusts the door opening speed to 0.4m / s, the opening width to 2.0 meters, and maintains the opening time to 8 seconds. At the same time, it reduces the door closing speed to ensure that users pushing strollers or sitting in wheelchairs can pass through the door safely and smoothly.

[0048] During peak hours, such as rush hour, dense crowds appear at entrances. The user behavior intent prediction model analyzes the density and movement trends of people at entrances to predict the duration and passage demand of the crowds. A hierarchical adaptive decision engine adjusts the gate operation strategy to keep the gate fully open until the crowd density drops below a preset threshold, avoiding frequent gate opening and closing, improving passage efficiency, and reducing mechanical wear on the gate.

[0049] The door operation health assessment module monitors the operating status of the two motors and transmission mechanism in real time. Due to the high frequency of use of automatic sliding doors in commercial buildings, the wear rate of door components is relatively fast. When an abnormal increase in temperature, increased current, and decreased speed are detected in one of the motors, the door operation health assessment model determines that the motor has an overheating fault risk, and the door health index drops to 60 points. At this point, the health compensation layer adjusts the door operation strategy, reducing the door's operating speed and load, while distributing more load to the other normal motor to prevent further damage to the faulty motor. Simultaneously, the system sends an alarm message to the property management platform, notifying maintenance personnel to conduct timely repairs.

[0050] When it rains, the temperature and humidity sensors in the environmental sensing unit detect that the air humidity is high and the ground may be slippery. The environmental adaptation layer adjusts the door's operating strategy, appropriately reducing the opening and closing speed of the door and extending the time it remains open to prevent users from slipping and falling on the wet ground, while also preventing the door from opening and closing quickly and bringing rainwater into the room.

[0051] When the main controller fails, the backup controller immediately takes over the control task and continues to run the AI ​​adaptive control algorithm of this invention to ensure the normal operation of the gate. At the same time, the system sends a main controller failure alarm information to the property management platform to notify maintenance personnel to replace the main controller in a timely manner.

[0052] This embodiment utilizes an AI-adaptive intelligent door operation algorithm, enabling automatic sliding doors in commercial buildings to adapt to various passage requirements, including single-person passage, multi-person parallel passage, passage for special groups, and dense passage during peak hours. This improves passage efficiency and user experience. The algorithm can monitor the operating status of each door component in real time, promptly detect fault risks, and take corresponding protective measures, extending the door's lifespan. It can automatically adjust its operating strategy according to environmental changes and adapt to different weather conditions. The use of a redundant main / backup controller design enhances system reliability and continuous operation. This algorithm is particularly suitable for automatic door systems in commercial buildings with high usage frequency and complex passage scenarios, offering significant application advantages.

[0053] The present invention has been described in detail above. However, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, any modifications or improvements that do not depart from the spirit of the present invention are within the scope of protection of the present invention.

Claims

1. An AI-adaptive control intelligent door operation algorithm, characterized in that, Includes the following steps: S1. Construct a multimodal perception fusion module to simultaneously collect data on the door's own operating status, the door area's environmental perception data, and user behavior characteristics. S2. Establish a gate operation health assessment model, and dynamically calculate the current health index of the gate based on historical operation data and real-time collected data; S3. Generate a user behavior intent prediction model. Through time-series analysis and semantic understanding of user behavior feature data, predict the user's travel intent and travel method. S4. Construct a hierarchical adaptive decision engine, combining the gate health index, user behavior intention prediction results, and environmental perception data to generate the optimal gate operation control strategy. S5. Execute closed-loop operation control, drive the gate actuator to operate according to the generated control strategy, and provide real-time feedback on the operating status to dynamically adjust the control parameters.

2. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, The multimodal perception fusion module includes a door status perception unit, an environment perception unit, and a user behavior perception unit. The door status perception unit collects data on motor current, voltage, speed, temperature, and door position, speed, and acceleration. The environment perception unit collects data on obstacles in the door area, light intensity, temperature, humidity, and wind force. The user behavior perception unit collects data on user position, movement speed, posture, orientation, and biometric features.

3. The AI ​​adaptive control intelligent door operation algorithm according to claim 2, characterized in that, The multimodal perception fusion module employs a spatiotemporal alignment and feature-level fusion method to synchronize and register heterogeneous data collected by different sensors in time and space, extract fused feature vectors, and input them into the subsequent evaluation and prediction model.

4. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, The door's operational health assessment model is constructed using a deep learning network. It uses historical operational data of the door as training samples to learn the operational characteristic patterns of the door under different health conditions. It calculates the wear and tear, aging, and failure probability of each component of the door in real time, and obtains the overall health index of the door.

5. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, The user behavior intent prediction model adopts an architecture that combines temporal convolutional networks and attention mechanisms. It analyzes the temporal changes of user behavior feature data, extracts key features of user behavior, and combines them with historical user behavior databases to predict whether the user needs to pass through, the direction of passage, the speed of passage, and whether the user is carrying items or accompanying others.

6. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, The hierarchical adaptive decision engine includes a basic decision layer, a health compensation layer, and an environment adaptation layer. The basic decision layer generates a basic operation control strategy based on the prediction results of user behavior intentions. The health compensation layer compensates the parameters of the basic control strategy based on the door health index. The environment adaptation layer further optimizes and adjusts the control strategy based on environmental perception data.

7. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, The health compensation layer dynamically adjusts the motor drive power, door running speed, acceleration, and anti-pinch sensitivity parameters based on the door health index. When the door health index is lower than a preset threshold, it automatically reduces the door running speed and increases the anti-pinch sensitivity, while generating maintenance reminder information.

8. The AI ​​adaptive control intelligent door operation algorithm according to claim 6, characterized in that, The environmental adaptation layer dynamically adjusts the door's opening angle, opening speed, and the duration of keeping it open based on wind force, obstacle distribution, and pedestrian density information from environmental perception data. In strong wind environments, it automatically reduces the door's opening speed and increases windproof damping, while in densely populated environments, it extends the duration of keeping the door open.

9. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, The closed-loop operation control system collects real-time operating status data of the gate actuator, compares it with preset operating targets, calculates control deviations, and dynamically adjusts control parameters through a PID controller to achieve precise control of the gate operation. At the same time, the actual operating data is fed back to the gate operation health assessment model and the user behavior intention prediction model to achieve online updating and optimization of the models.

10. The AI ​​adaptive control intelligent door operation algorithm according to claim 1, characterized in that, It also includes anomaly handling and security mechanisms. When abnormal door operation, obstruction, or user safety risks are detected, security protection actions are immediately triggered to stop the door operation or reverse it, record abnormal event information, and send alarm information to the management platform.

Images