Elevator door control method and device, electronic equipment and medium

Abstract

The invention provides an elevator door control method and device, electronic equipment and a medium, and relates to the field of elevator safety, and the method comprises the steps that when it is determined that an obstacle exists between two door leaves of an elevator door, multi-source data of an elevator door area is obtained; the multi-source data are processed, and the current door leaf states of the two door leaves and obstacle information of obstacles between the door leaves are determined; a configured three-level detection strategy is adopted, the door leaf state and obstacle information are detected, and a current clamping detection result is determined; and based on the current clamping detection result, a target response strategy is determined, and the elevator door is controlled to operate according to the target response strategy. Through the advantages of improving the detection precision, enhancing the environmental adaptability, ensuring the real-time performance, reducing the false alarm rate and the like, the safety performance of the elevator is comprehensively improved, a more reliable and safer elevator taking environment is provided for passengers, and the elevator safety monitoring system has extremely high practical value and wide application prospects.

Description

Technical Field

[0001] The present application relates to the field of elevator safety, and more specifically, to a method, device, electronic equipment, and medium for controlling an elevator door. Background Art

[0002] Existing image processing-based elevator anti-pinch methods suffer from significant environmental adaptability deficiencies. Elevator operating environments are complex and highly variable, particularly with widely varying lighting conditions. In strong light environments (such as glass-curtain-wall elevators with direct sunlight hitting the door area), captured images are prone to severe overexposure, resulting in loss of information about the edges of door gaps and the outlines of small objects, making it difficult for image processing algorithms to accurately segment and identify potential obstacles. In low-light environments (such as underground garages or elevators operating at night), image noise increases, blurring details, and the confidence level of the object detection model drops by over 40% (measured data), while both false detection and missed detection rates increase. This degradation in algorithm performance due to light fluctuations prevents the system from consistently and reliably ensuring safe operation of elevator doors in alternating light and dark scenarios, posing a significant safety hazard. Summary of the Invention

[0003] The purpose of the embodiments of the present application is to provide an elevator door control method, device, electronic device and medium to solve the above-mentioned problems existing in the prior art and to improve the perception accuracy of external obstacles during the closing process of the elevator door.

[0004] In a first aspect, a method for controlling an elevator door is provided, which may include: When it is determined that there is an obstacle between the two door leaves of the elevator, multi-source data of the elevator door area is obtained; Processing the multi-source data to determine current door leaf states of the two door leaves and obstacle information of obstacles between the door leaves; Adopting the configured three-level detection strategy, the door leaf state and the obstacle information are detected to determine the current clamping detection result; Based on the current clamping detection result, a target response strategy is determined, and the elevator door is controlled to operate according to the target response strategy.

[0005] In one possible implementation, a camera and a TOF depth sensor are provided in the elevator car corresponding to the elevator door; The multi-source data includes image data corresponding to the elevator door area collected by a camera and depth data corresponding to the elevator door area collected by a TOF depth sensor.

[0006] In one possible implementation, determining the current door leaf states of the two door leaves includes: Segmenting the image data to obtain a plurality of sub-image data; Performing feature extraction on the plurality of sub-image data to determine at least one target sub-image data corresponding to the door gap of the elevator door; If there are multiple target sub-image data, the multiple target sub-image data are fused to obtain the door gap image data; Based on the door gap image data, the current door leaf state is determined, and the current door leaf state includes: a fully open state, a fully closed state, and a partially open state.

[0007] In one possible implementation, determining obstacle information of an obstacle between door leaves includes: When the current door leaf state is not fully open, a target sub-image data corresponding to the door gap or the door gap image data is binarized to obtain grayscale image data; Performing feature extraction on the depth data to obtain target depth data corresponding to the obstacle between the two door leaves of the elevator door; Using a multi-scale feature pyramid, the grayscale image data and the target depth data are fused to obtain a fusion result; Obstacle information is determined based on the fusion result; the obstacle information includes: the distance between the obstacle and each door leaf.

[0008] In one possible implementation, a configured three-level detection strategy is used to process the door leaf state and the obstacle information to determine a current clamping detection result, including: If the door leaf is not in a fully open state, the configured first detection strategy is adopted to analyze the image data and historical image data adjacent to the image data to determine the first motion mode of each door leaf and the second motion mode of the obstacle; When an abnormal pattern exists in the first motion pattern and / or the second motion pattern, a second detection strategy is adopted to analyze the image data and historical image data adjacent to the image data to determine an initial clamping detection result; If the initial clamping detection result is determined to be clamping, a third detection strategy is adopted to analyze the fusion result and the historical fusion result to determine a current clamping detection result.

[0009] In a possible implementation, if the current clamping detection result is that clamping exists, a target response strategy is determined based on the current door leaf state and the distance between the obstacle and each door leaf.

[0010] In one possible implementation, determining a target response strategy includes: If the current door leaf state is a partially open state and is in the closing process, and the distance between the obstacle and each door leaf is not greater than the preset distance, the target response strategy is to control the door leaf to stop the current movement and open the door leaf to the maximum opening at a first preset speed; If the current door leaf state is partially open and in the closing process, and the distance between the obstacle and each door leaf is greater than the preset distance, the target response strategy is to control the door leaf to pause movement for a preset period of time and monitor whether the obstacle moves. When the obstacle is removed, the door leaf is controlled to close at a second preset speed.

[0011] In a second aspect, a control device for an elevator door is provided, which may include: an acquisition unit, configured to acquire multi-source data of the elevator door area when it is determined that there is an obstacle between the two door leaves of the elevator door; a processing unit, configured to process the multi-source data to determine current door leaf states of the two door leaves and obstacle information of an obstacle between the door leaves; A detection unit, configured to detect the door leaf state and the obstacle information using a configured three-level detection strategy, and determine a current clamping detection result; A determination unit determines a target response strategy based on the current clamping detection result, and controls the elevator door to operate according to the target response strategy.

[0012] In a third aspect, an electronic device is provided, the electronic device including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; Memory for storing computer programs; The processor is configured to implement any of the method steps described in the first aspect when executing a program stored in the memory.

[0013] In a fourth aspect, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any of the method steps described in the first aspect is implemented.

[0014] This application provides an elevator door control method, comprising: acquiring multi-source data about the elevator door area when an obstacle is determined to exist between two door leaves; processing the multi-source data to determine the current door leaf states of the two door leaves and obstacle information about the obstacle between them; employing a configured three-level detection strategy to detect the door leaf states and obstacle information and determine a current pinch detection result; determining a target response strategy based on the current pinch detection result, and controlling the elevator door to operate according to the target response strategy. This application can identify obstacles with a minimum diameter of 2mm, achieving a three-fold improvement in detection accuracy compared to traditional light curtain technology. By utilizing the optional laser radar in the multi-source data acquisition module and a three-level anti-pinch detection mechanism, this application achieves accurate identification of obstacles of various sizes, significantly reducing the possibility of pinching accidents caused by undetected small objects and effectively ensuring passenger safety. In summary, this application comprehensively enhances elevator safety performance by improving detection accuracy, enhancing environmental adaptability, ensuring real-time performance, and reducing false alarm rates, providing passengers with a more reliable and safe elevator riding environment. The application has extremely high practical value and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 A system architecture diagram of an elevator door control method provided in an embodiment of the present application; Figure 2 A flowchart of a method for controlling an elevator door provided in an embodiment of the present application; Figure 3 A schematic structural diagram of an elevator door control device provided in an embodiment of the present application; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] The elevator door control method provided in the embodiment of the present application can be applied to Figure 1 In the system architecture shown in Figure 1 As shown, the system may include: an elevator operation system and a processor.

[0019] The elevator operation system consists of a sensor module and a control module. The sensor module includes various sensing devices for comprehensive monitoring of the elevator doors and their surroundings. These include multiple time-of-flight (TOF) depth sensors (which collect real-time 3D spatial information about the elevator door area), cameras (which capture color image data of the area in front of the elevator door), lidar (which accurately scans the area near the elevator door to assist in identifying the outline and location of obstacles), and pressure sensors. Each sensor type captures raw data and sends it to the processor, providing multi-dimensional information for subsequent analysis and decision-making. The camera is a 2-megapixel, wide-angle, high-definition camera that captures RGB images of the elevator door area at a frame rate of 30 fps. This captures clear, real-time visual information around the elevator door, providing a rich data source for subsequent image processing. A TOF depth sensor with an accuracy of ±5mm generates 3D point cloud data, supplementing the image data with depth information, enabling the system to more accurately perceive the spatial position and shape of objects. The optional lidar is specifically designed to assist in detecting small objects, addressing the poor detection performance of traditional infrared light curtains and significantly improving the detection capabilities of various objects.

[0020] In some embodiments, the TOF depth sensor can be installed at the center of the top of the elevator door, the bottom edge of the elevator door, and the middle side edge of the two elevator door leaves. A camera is installed at the center of the top of the elevator door or above the interior of the elevator car to provide high-resolution color image data to support target detection and recognition. The camera can help distinguish different types of obstacles and provide texture information. A lidar is installed at the center of the top of the elevator door or above the interior of the elevator car to scan the environment near the elevator door with high precision and assist in identifying the outline and location of obstacles. Lidar is particularly suitable for providing accurate distance measurement in complex environments.

[0021] The processor is used to receive various original data, and after filtering, edge detection, contrast enhancement and other operations on the original data, execute a control method for an elevator door provided by the present application.

[0022] The control module dynamically adjusts the operating status of the elevator based on the target response strategy output by the processor.

[0023] With the acceleration of urbanization, high-rise buildings are constantly emerging. As an essential means of vertical transportation for people to go up and down stairs, the safety of elevators is of paramount importance. However, accidents involving people being trapped by elevator doors are frequent, posing a serious threat to the lives of passengers. Although existing elevator anti-pinch technology has improved safety to a certain extent, it still has many shortcomings: Infrared light curtains are a widely used anti-pinch technology in elevators. They detect obstacles between elevator doors by emitting and receiving infrared light. However, this technology is less effective for detecting small or flexible objects, such as pet leashes and thin items. These objects block a small area of ​​light, making them prone to missed detection by the infrared light curtain. If missed, these small objects could become trapped in the elevator doors during closing, potentially causing injury.

[0024] As a contact-based detection device, the mechanical touch panel can detect obstacles to a certain extent. However, it only activates after the elevator door contacts the obstacle, by which time damage may have already occurred to trapped objects or people. Furthermore, due to factors such as the inertia of the elevator door, triggering the mechanical touch panel may cause secondary damage, further exacerbating injuries to trapped people.

[0025] Image processing-based elevator anti-pinch methods rely on cameras to capture image data of the area in front of the elevator door. However, elevators operate in complex and diverse environments, with widely varying lighting conditions. In strong light, images are prone to overexposure, resulting in loss of critical information; in low light, images become blurred, affecting the algorithm's object recognition accuracy. Therefore, in unstable lighting conditions, the performance of these image processing-based methods degrades significantly, making them unable to reliably guarantee the safe operation of elevator doors.

[0026] Traditional elevator door status recognition and anti-pinch algorithms perform poorly in distinguishing between normal door opening and closing and abnormal entrapment situations. During normal door opening and closing, factors such as lighting changes and the normal movement of surrounding objects can cause the algorithm to misidentify an abnormal entrapment situation, resulting in false alarms. This high false alarm rate not only impacts elevator operation efficiency and increases unnecessary maintenance costs, but also reduces passenger confidence in elevator safety.

[0027] Therefore, the present application provides a control method for an elevator door, which solves the above-mentioned problems existing in the prior art and can improve the perception accuracy of external obstacles during the closing process of the elevator door.

[0028] The preferred embodiments of the present application are described below in conjunction with the drawings in the specification. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application and are not used to limit the present application. In addition, the embodiments and features in the embodiments of the present application can be combined with each other if there is no conflict.

[0029] Figure 2 This is a flow chart of a method for controlling an elevator door provided in an embodiment of the present application. Figure 2 As shown, the method may include: Step S210: When it is determined that there is an obstacle between the two door leaves of the elevator door, multi-source data of the elevator door area is obtained.

[0030] Specifically, the method of determining whether there is an obstacle between the two door leaves of the elevator door may include: Obtain image data corresponding to the elevator door area captured by the camera; process the image to determine the light source conditions when the image data was captured; If the image data is taken under a normal light source environment, determining whether there is an obstacle between the two doors based on the image data; If the image data is captured in a strong or weak light environment, the depth data of consecutive frames collected by the TOF depth sensor is obtained to determine whether there is an obstacle between the two doors: When an obstacle is present between the two doors, multiple sensors installed inside the elevator car are triggered to collect data from the elevator door area, thereby obtaining multi-source data of the elevator door area. The multi-source data includes image data corresponding to the elevator door area collected by the camera and depth data corresponding to the elevator door area collected by the TOF depth sensor.

[0031] Step S220: Process the multi-source data to determine the current door leaf states of the two door leaves and the obstacle information of the obstacles between the door leaves.

[0032] Specifically, the current door leaf states of the two door leaves are determined, including: Processing the image data to determine the light source conditions when the image data was captured; A. If the image data is taken under a normal light source environment, the image data is segmented to obtain a plurality of sub-image data; Performing feature extraction on the plurality of sub-image data to determine at least one target sub-image data corresponding to the door gap of the elevator door; If there are multiple target sub-images, these are fused to obtain the door gap image data. Specifically, if the target sub-images are directly adjacent in space (grid segmentation), they can be directly concatenated to form a local region encompassing the entire door gap (or most of the door gap). If the segmentations overlap or are not directly adjacent, the strongest vertical edge segments (feature points) in each target sub-image can be extracted. These edge segments (feature points) are plotted according to their coordinates in the original image onto a blank mask image (mask image) that matches the original image size. The "bright line" area on this mask image represents the fused door gap location information, which is the door gap image data. Alternatively, automatic door gap region calibration technology based on U-Net can be used. U-Net is a convolutional neural network architecture with excellent performance in image segmentation. By training on a large number of elevator door images, the network can accurately identify the door gap location and automatically calibrate the door gap region. Even if there is a certain position deviation in the camera during installation, this technology can accurately determine the position of the door gap by learning and analyzing the image features, eliminating the impact of position deviation on subsequent analysis.

[0033] Based on the door gap image data, the current door state is determined. Current door states include fully open, fully closed, and partially open. Specifically, an improved YOLOv5s network architecture is employed, consisting of an image preprocessing module, an improved CSPDarknet53 module, a classification branch module, a regression branch module, and an output module. The image preprocessing module converts the input elevator door image into a format suitable for network processing and enhances image features to improve the model's generalization capability. The improved CSPDarknet53 module efficiently extracts basic image features, providing a rich feature representation for subsequent network layers. The Bi-FPN module (replacing the original FPN) enhances multi-scale feature fusion, enabling the network to effectively utilize feature information at different levels, improving the detection accuracy of elevator door states and surrounding objects of varying sizes. The classification branch module predicts the elevator door state category (fully open, fully closed, or partially open (30%, 60%, and 90%)) based on the feature map output by the neck network. The regression branch module predicts the position and size of detected target objects (such as obstacles) to determine whether the objects are in a dangerous area. The output layer integrates the results of the classification and regression branches and outputs the final detection results, including the elevator door status, detected objects, and their locations. This can also be understood as first replacing the original Feature Pyramid Network (FPN) with a Bidirectional Feature Pyramid Network (Bi-FPN). By introducing additional bidirectional connections, Bi-FPN enhances the information flow between features at different scales, making the network more effective at fusing multi-scale features and better capturing the features of objects of varying sizes. Furthermore, a Network Attention Mechanism (NAM) is introduced. NAM automatically learns the importance weights of different features, giving greater attention to the features of small objects, thereby improving the detection accuracy of small objects. During training, the improved network is trained using a large amount of elevator door image data containing different door states (fully open, fully closed, partially open 30%, partially open 60%, and partially open 90%). The network parameters are adjusted to enable it to accurately classify different elevator door states. During actual operation, image data processed by the adaptive image processing module is input into the trained network, which then outputs the specific state category of the elevator door.

[0034] B. If the image data was captured in a strong or weak light environment, the distance between the two elevator door panels is determined by the depth data provided by the TOF depth sensor corresponding to the middle side edges of the two elevator door panels, thereby determining the current door panel status. Specifically, edges perpendicular to the direction of door movement are enhanced (the elevator door gap is a vertical edge, and the horizontal gradient is calculated); a direction-controllable Gabor filter is used to suppress non-vertical noise (such as human shadows and debris); the depth data of the previous N frames after median filtering is subtracted from the current frame depth data to construct the depth residual data of the continuous frames; persistent stable edges (real door gaps) are retained, and transient interference (such as passing passengers) is automatically filtered out. This method can accurately lock the edge of the door gap in dynamic and cluttered environments, avoiding misjudgments due to temporary occlusion or reflections.

[0035] Because a door gap is essentially a "depression" in physical depth between two doors, by constructing depth data for the door leaf area, the gap naturally appears as a "groove" with a sudden change in depth. Its width (opening degree) can be precisely quantified and is unaffected by light or smoke.

[0036] In some embodiments, if the image data is taken in a strong light or weak light environment, the improved Retinex algorithm is used to perform illumination compensation on the collected image data. The algorithm can adaptively adjust the brightness and contrast of the image according to the illumination distribution of the image. Within the illumination range of 0-100000 lux, whether in an overly bright strong light environment or a dim weak light environment, the details of the image can be enhanced to facilitate subsequent processing. In the specific implementation process, the image data under different illumination conditions is normalized to an appropriate illumination level by transforming the pixel values ​​of the image data. Among them, the improved Retinex algorithm includes: using multi-scale Gaussian filtering (the scale is adaptively selected according to the local illumination variance) to estimate the illumination component L of the input image data I, and calculating the reflection component in the logarithmic domain. The R component is corrected using a chroma gain coefficient to restore color distortion caused by light separation. HDR fusion is enabled to prevent overexposure in areas above 80,000 lux, and noise suppression constraints are added to areas below 1 lux. The output image data's luminance range is linearly mapped to a processable range (e.g., 0–255) to ensure consistent detail visibility under varying lighting conditions. This algorithm significantly improves the recognition of key features such as door edges and textures within the 0–100,000 lux illumination range.

[0037] Obstacle information to determine obstacles between door leaves, including: When the current door leaf state is not fully open, the target detection level corresponding to the current operating speed of the two door leaves is determined according to the corresponding relationship between the configured different operating speeds and different detection levels, including: If the operating speed of the two door leaves is less than the configured first speed threshold, the target detection level corresponding to the current operating speed is level 1 detection. The target detection method corresponding to level 1 detection is to control devices such as cameras, time-of-flight depth sensors, and lidar to collect relevant data from the elevator door area for a first preset duration. In a specific example, the camera speed is reduced to 15fps, the resolution of the time-of-flight depth sensor is halved (for example, 320×240→160×120), the lidar point cloud density is reduced by 30%, and the sampling time is extended to 200ms (to improve the ability to capture details in static scenes).

[0038] If the operating speed of the two door leaves is not less than the configured first speed threshold, the target detection level corresponding to the current operating speed is level 2 detection. In a specific example, the camera is overclocked to 60fps, the TOF depth sensor resolution is restored to full resolution, the lidar starts high-frequency mode (15Hz→30Hz), and the sampling time is shortened to 50ms (meeting the real-time requirements in fast-motion scenarios).

[0039] Afterwards, a target sub-image data or door gap image data corresponding to the door gap is binarized to obtain grayscale image data.

[0040] Perform feature extraction on the depth data to obtain the target depth data corresponding to the obstacle between the two door leaves of the elevator door; A multi-scale feature pyramid is used to fuse grayscale image data and target depth data to obtain a fusion result. This method enables the fused features to contain both overall information at a large scale and detailed information at a small scale, thereby giving full play to the texture information of image data and the spatial information advantages of depth images, generating more comprehensive and representative feature data, and providing strong support for subsequent door status classification and anti-pinch detection.

[0041] Based on the fusion results, obstacle information is determined; obstacle information includes: the distance of the obstacle from each door leaf and the obstacle category; obstacle categories include flexible obstacles (rope, scarf, paper, etc.) and rigid obstacles.

[0042] In one embodiment, the velocity embedding module can be used to integrate door leaf speed information into the target detection network. Based on YOLOv8, a velocity feature branch is added and concatenated with the main feature map, and the attention mechanism is used to enhance the spatiotemporal features related to speed. To address the issue of camera image blur in high-speed scenarios, a deblurring algorithm based on a physical model is developed. The algorithm calculates the camera motion velocity using the displacement field of two adjacent frames of depth data, generates a point spread function (PSF), and combines non-local mean filtering with Wiener filtering to deblur a single frame within 5ms, restoring obstacle edge details. This method improves obstacle detection accuracy by 23% after deblurring at a door leaf speed of 30 cm / s.

[0043] Step S230: Use the configured three-level detection strategy to detect the door leaf status and obstacle information to determine the current clamping detection result.

[0044] Specifically, if the door leaf is not in a fully open state, the configured first detection strategy is used to analyze the image data and the historical image data adjacent to the image data to determine the first motion mode of each door leaf and the second motion mode of the obstacle; specifically, the first detection strategy is to use the optical flow method to analyze the position changes of the door leaf in continuous frames to determine its motion mode. For example, if the door leaf closes normally, it should follow a smooth, linear trajectory; if there are anomalies (such as sudden stops or reverse movement), it may indicate a problem. Similarly, the motion trajectory of any objects that appear in the door leaf path is tracked. These objects may be passengers' clothing or other personal belongings. If the object's motion trajectory shows signs of being squeezed or obstructed by the door leaf, there may be a risk of pinching.

[0045] When abnormal patterns are detected in the first and / or second motion patterns, a second detection strategy is employed to analyze the image data and adjacent historical image data to determine the initial pinching detection result. Specifically, body motion recognition is performed using a spatiotemporal graph convolutional network (ST-GCN). ST-GCN is a neural network specifically designed to process spatiotemporal data and is suitable for human motion recognition. Image sequence data representing potential pinching risks, transmitted from the first-level detection, is input into the ST-GCN network. The network learns the spatiotemporal motion characteristics of human limbs to identify potential pinching motion patterns. For example, when an arm or other body part exhibits unnatural blocking or twisting motions, the network can accurately identify and categorize it as a pinching motion, further confirming the pinching risk.

[0046] If the initial clamping detection result indicates a clamping event, the third detection strategy is used to analyze the fusion results and historical fusion results to determine the current clamping detection result. Specifically, features such as edge density, color consistency, and motion trajectory can be extracted from the fusion results and historical fusion results. A classifier (such as a support vector machine (SVM) or random forest (RF)) can then be trained to distinguish between flexible and rigid obstacles.

[0047] Step S240: Determine a target response strategy based on the current clamping detection result, and control the elevator door to operate according to the target response strategy.

[0048] Specifically, when the current clamping detection result is clamping, the target response strategy is determined based on the current door leaf state, the distance of the obstacle from each door leaf, and the obstacle type: If the current door leaf state is partially open and the distance between the obstacle and each door leaf during the closing process is not greater than the preset distance, the target response strategy is to control the door leaf to stop the current movement and open the door leaf to the maximum opening at the first preset speed; in this method, the process of opening the door leaf to the maximum opening at the first preset speed includes: a graded flexible door opening mechanism. During the door opening process, the door opening speed is dynamically adjusted according to the pressure changes fed back by the pressure sensor in real time. If the pressure continues to increase, it means that the obstacle may be stuck tighter. At this time, the door opening speed is reduced to avoid causing greater damage to the obstacle; if the pressure decreases, the speed is appropriately increased to speed up the rescue process. At the same time, combined with visual data, if the obstacle is identified as easily-torn items such as clothing, a pulsed door opening method is adopted, that is, each time the door is opened a short distance, it pauses for 0.5 seconds before continuing to open the door to reduce the risk of damage to items.

[0049] If the door is partially open and in the process of closing, and the obstacle is greater than a preset distance from each door, the target response strategy is to pause the door movement for a preset duration and monitor whether the obstacle has moved. Once the obstacle has cleared, the door is closed at a second preset speed. This process may include adding predictive analysis to the preset pause and obstacle movement monitoring. An LSTM (Long Short-Term Memory) model trained with historical data is used to predict the obstacle's movement trend. If the model predicts that the obstacle is likely to clear within the next few seconds, even if the obstacle has not moved, the door closing action can be slowly initiated in advance, gradually closing the door at the second preset speed to reduce waiting time. Furthermore, elevator door status information is linked to the elevator car control system and building management system. If an obstacle is detected for an extended period (e.g., exceeding 30 seconds), an alarm is automatically sent to the elevator operator and a prompt is displayed in the elevator car, guiding passengers to manually intervene or wait for rescue.

[0050] The present application provides a method for controlling elevator doors, which includes obtaining multi-source data of the elevator door area when it is determined that there is an obstacle between the two door leaves of the elevator door; processing the multi-source data to determine the current door leaf status of the two door leaves and the obstacle information of the obstacle between the door leaves; using a configured three-level detection strategy to detect the door leaf status and the obstacle information to determine the current clamping detection result; based on the current clamping detection result, determining the target response strategy, and controlling the elevator door to operate with the target response strategy. The present application can identify obstacles with a minimum diameter of 2mm, and the detection accuracy is increased by 3 times compared to traditional light curtain technology. The present application realizes the accurate identification of obstacles of various sizes through the optional configuration of laser radar in the multi-source data acquisition module and the three-level detection mechanism of anti-pinch detection, greatly reducing the possibility of pinching accidents caused by small objects not being detected, and effectively ensuring the personal safety of passengers. In summary, this application comprehensively improves the safety performance of elevators by improving detection accuracy, enhancing environmental adaptability, ensuring real-time performance, and reducing false alarm rates, providing passengers with a more reliable and safe elevator riding environment. It has extremely high practical value and broad application prospects.

[0051] Corresponding to the above method, the embodiment of the present application also provides a control device for an elevator door, such as Figure 3 As shown, the device includes: An acquisition unit 310 is configured to acquire multi-source data of the elevator door area when it is determined that there is an obstacle between the two door leaves of the elevator door; a processing unit 320 for processing the multi-source data to determine the current door states of the two door leaves and obstacle information of an obstacle between the door leaves; The detection unit 330 is configured to detect the door leaf state and the obstacle information using the configured three-level detection strategy to determine the current clamping detection result; The determination unit 340 determines a target response strategy based on the current clamping detection result, and controls the elevator door to operate according to the target response strategy.

[0052] The functions of the various functional units of the elevator door control device provided in the above embodiment of the present application can be achieved through the above method steps. Therefore, the specific working process and beneficial effects of the various units in the elevator door control device provided in the embodiment of the present application will not be repeated here.

[0053] The present application also provides an electronic device, such as Figure 4 As shown, it includes a processor 410 , a communication interface 420 , a memory 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other via the communication bus 440 .

[0054] Memory 430, for storing computer programs; The processor 410 is configured to execute the program stored in the memory 430 by performing the following steps: When it is determined that there is an obstacle between the two door leaves of the elevator, multi-source data of the elevator door area is obtained; Processing the multi-source data to determine current door leaf states of the two door leaves and obstacle information of obstacles between the door leaves; Adopting the configured three-level detection strategy, the door leaf state and the obstacle information are detected to determine the current clamping detection result; Based on the current clamping detection result, a target response strategy is determined, and the elevator door is controlled to operate according to the target response strategy.

[0055] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. This communication bus can be divided into address buses, data buses, and control buses. For ease of illustration, the figure uses only one thick line, but this does not mean that there is only one bus or only one type of bus.

[0056] The communication interface is used for communication between the above electronic device and other devices.

[0057] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage. Alternatively, the memory may be at least one storage device located away from the processor.

[0058] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.

[0059] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments to solve the problems can be found in Figure 2 The various steps in the embodiment shown are implemented, therefore, the specific working process and beneficial effects of the electronic device provided by the embodiment of the present application are not repeated here.

[0060] In another embodiment provided by the present application, a computer-readable storage medium is further provided, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes an elevator door control method described in any one of the above embodiments.

[0061] In another embodiment provided by the present application, a computer program product including instructions is further provided, which, when executed on a computer, enables the computer to execute the elevator door control method described in any one of the above embodiments.

[0062] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the embodiments of the present application may take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware. Furthermore, the embodiments of the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0063] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0064] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0066] Unless otherwise defined, the technical or scientific terms used in this application should have the usual meanings understood by persons of ordinary skill in the field to which the invention belongs. The words "first", "second" and similar terms used in this application do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprise" mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect", "couple" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0067] Although preferred embodiments have been described in the present application, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the present application is intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0068] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the embodiments of the present application and their equivalents, the embodiments of the present application are also intended to include these modifications and variations.

Claims

1. A method for controlling an elevator door, characterized in that: The method comprises: When it is determined that there is an obstacle between the two door leaves of the elevator, multi-source data of the elevator door area is obtained; Processing the multi-source data to determine current door leaf states of the two door leaves and obstacle information of obstacles between the door leaves; Adopting the configured three-level detection strategy, the door leaf state and the obstacle information are detected to determine the current clamping detection result; Based on the current clamping detection result, a target response strategy is determined, and the elevator door is controlled to operate according to the target response strategy.

2. The method according to claim 1, wherein A camera and a TOF depth sensor are provided in the elevator car corresponding to the elevator door; The multi-source data includes image data corresponding to the elevator door area collected by a camera and depth data corresponding to the elevator door area collected by a TOF depth sensor.

3. The method according to claim 2, wherein Determine the current leaf status of both door leaves, including: Segmenting the image data to obtain a plurality of sub-image data; Performing feature extraction on the plurality of sub-image data to determine at least one target sub-image data corresponding to the door gap of the elevator door; If there are multiple target sub-image data, the multiple target sub-image data are fused to obtain the door gap image data; Based on the door gap image data, the current door leaf state is determined, and the current door leaf state includes: a fully open state, a fully closed state, and a partially open state.

4. The method according to claim 3, wherein Obstacle information to determine obstacles between door leaves, including: When the current door leaf state is not fully open, a target sub-image data corresponding to the door gap or the door gap image data is binarized to obtain grayscale image data; Performing feature extraction on the depth data to obtain target depth data corresponding to the obstacle between the two door leaves of the elevator door; Using a multi-scale feature pyramid, the grayscale image data and the target depth data are fused to obtain a fusion result; Obstacle information is determined based on the fusion result; the obstacle information includes: the distance between the obstacle and each door leaf.

5. The method according to claim 4, wherein The configured three-level detection strategy is used to process the door leaf status and the obstacle information to determine the current clamping detection result, including: If the door leaf is not in a fully open state, the configured first detection strategy is adopted to analyze the image data and historical image data adjacent to the image data to determine the first motion mode of each door leaf and the second motion mode of the obstacle; When an abnormal pattern exists in the first motion pattern and / or the second motion pattern, a second detection strategy is adopted to analyze the image data and historical image data adjacent to the image data to determine an initial clamping detection result; If the initial clamping detection result is determined to be clamping, a third detection strategy is adopted to analyze the fusion result and the historical fusion result to determine a current clamping detection result.

6. The method according to claim 5, wherein If the current clamping detection result is that clamping exists, a target response strategy is determined based on the current door leaf state and the distance between the obstacle and each door leaf.

7. The method according to claim 6, wherein Determine target response strategies, including: If the current door leaf state is a partially open state and is in the closing process, and the distance between the obstacle and each door leaf is not greater than the preset distance, the target response strategy is to control the door leaf to stop the current movement and open the door leaf to the maximum opening at a first preset speed; If the current door leaf state is partially open and in the closing process, and the distance between the obstacle and each door leaf is greater than the preset distance, the target response strategy is to control the door leaf to pause movement for a preset period of time and monitor whether the obstacle moves. When the obstacle is removed, the door leaf is controlled to close at a second preset speed.

8. An elevator door control device, characterized in that: The device comprises: an acquisition unit, configured to acquire multi-source data of the elevator door area when it is determined that there is an obstacle between the two door leaves of the elevator door; a processing unit, configured to process the multi-source data to determine current door leaf states of the two door leaves and obstacle information of an obstacle between the door leaves; A detection unit, configured to detect the door leaf state and the obstacle information using a configured three-level detection strategy, and determine a current clamping detection result; A determination unit determines a target response strategy based on the current clamping detection result, and controls the elevator door to operate according to the target response strategy.

9. An electronic device, characterized in that: The electronic device includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; A processor, configured to implement the method steps described in any one of claims 1 to 7 when executing a program stored in a memory.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps according to any one of claims 1 to 7 are implemented.

Images