High-place operation safety belt failure early warning device and method

Abstract

The invention relates to the technical field of safety protection, and provides a high-place operation safety belt failure early warning device and method.The real-time state of a safety rope can be evaluated through cooperation of a monitoring main machine and a monitoring auxiliary machine, and the real-time state of the safety rope is divided into a normal state, a low-hanging high-use state, a virtual-hanging state and a straightening state; the real-time state of the safety rope is sent to the handheld monitoring terminal and the supervision terminal through the monitoring host by using the Lora communication technology, so that safety supervision personnel can quickly master the safety condition of a construction site, timely take necessary safety measures and timely find potential safety hazards; by monitoring and feeding back the real-time state of the safety rope in real time, the occurrence rate of falling accidents in high-altitude operation is remarkably reduced, and the safety of constructors is improved; the handheld monitoring terminal has the function of monitoring the plurality of monitoring hosts at the same time, so that the safety supervision efficiency is enhanced, and the safety personnel can better manage the safety states of the plurality of construction personnel.

Description

Technical Field

[0001] The present invention relates to the technical field of safety protection, and in particular to a device and method for early warning failure of a safety belt for high-altitude operations. Background Art

[0002] In order to ensure the safe, stable and efficient operation of the power system, it is necessary to carry out overhead line maintenance work regularly. When carrying out overhead line maintenance work, operators are required to work at high altitudes. In order to ensure the safety of operators, they need to wear safety ropes when performing operations to prevent falling. However, in actual operations, some operators fail to use the safety ropes correctly, resulting in operators falling and causing safety accidents.

[0003] In order to supervise the operators to use the safety rope correctly, additional personnel are needed to supervise the operators in real time. However, during the supervision process, since the operators are working at high altitudes and the supervisors are generally supervising on the ground, it is difficult for the supervisors to see the working status of the safety rope and whether the safety rope is hung loosely. This poses a great safety hazard, is time-consuming and labor-intensive, and cannot guarantee the supervision effect.

[0004] In order to solve the above problems, some safety rope monitoring devices have appeared in the prior art. For example, based on image recognition technology, it is used to determine whether the safety belt is worn correctly. Although it solves some regulatory requirements for easily obtaining complete work images, for many complex high-altitude work scenes, such as the presence of obstructions, long distances, etc., there are still problems such as long detection time and low detection accuracy. It is difficult to discover safety hazards in time and cannot prevent falling accidents well.

[0005] Based on this, the present invention provides a safety belt failure warning device and method for working at heights, so as to shorten the detection time, improve the detection accuracy, timely discover safety hazards, and ensure the safety of personnel. Summary of the invention

[0006] The purpose of the present invention is to solve the deficiencies of the prior art and to provide a safety belt failure warning device and method for working at heights, so as to shorten the detection time, improve the detection accuracy, timely discover safety hazards, and ensure the safety of personnel.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A safety belt failure warning device for high-altitude operations, comprising: a wearing portion, a safety rope with one end fixedly connected to the wearing portion, and a hook detachably connected to the other end of the safety rope;

[0009] The wearable part is provided with a back buckle, and a monitoring host is provided on the outside of the back buckle; a monitoring auxiliary machine is detachably connected to the outside of the safety rope; the monitoring auxiliary machine is used to detect the spatial shape data of the safety rope, and the monitoring auxiliary machine detects the spatial shape data of the safety rope through an array displacement sensor; the array displacement sensor can collect the spatial shape data of the safety rope from the bottom of the hook to the back buckle connection point; the monitoring auxiliary machine is electrically connected to the monitoring host;

[0010] The monitoring host is used to determine the real-time status of the safety rope according to the safety rope spatial shape data output by the monitoring auxiliary machine, and send the real-time status information of the safety rope to the monitoring terminal.

[0011] Preferably, the array displacement sensor is integrated into a wear-resistant cable, and the wear-resistant cable is tightly wrapped with a safety rope.

[0012] Preferably, the real-time status of the safety rope includes: normal status, low hanging and high use status, virtual hanging status, and straightened status.

[0013] Preferably, the real-time status assessment of the safety rope is based on the following method:

[0014] Step 1: The monitoring host reads the deployed MobileNetV2 model;

[0015] Step 2: The monitoring auxiliary machine detects the spatial shape data of the safety rope through the array displacement sensor and transmits the data to the monitoring host;

[0016] Step 3: The monitoring host inputs the spatial morphology data of the safety rope into the MobileNetV2 model for classification, and divides the real-time status of the safety rope into: normal state, low hanging and high use state, virtual hanging state, and straightened state;

[0017] Step 4: Send the real-time status information of the safety rope to the monitoring terminal.

[0018] Preferably, in step 2, the monitoring auxiliary machine detects the spatial shape data of the safety rope through the array displacement sensor by the following method:

[0019] Step 2.1: The monitoring auxiliary machine captures the spatial coordinates x, y, and z of each part of the safety rope in three-dimensional space in real time through an array displacement sensor, and based on the acquired spatial coordinates x, y, and z of each part of the safety rope in three-dimensional space, uses an interpolation algorithm to fill in the gaps between the data to form a continuous dot matrix distribution;

[0020] Step 2.2: Generate a vector diagram using the generated dot matrix data to represent the overall shape of the safety rope. The features of the vector diagram include the length, curvature and tilt angle of the safety rope.

[0021] Step 2.3: Standardize the generated vector image.

[0022] Preferably, the MobileNetV2 acquisition method is:

[0023] Step 1: Data collection: collect safety rope status images in different states;

[0024] Step 2: Data labeling and processing:

[0025] Step 2.1: Label content: Use the LabelStudio tool to label and build a database containing information such as the labeler, labeling time, and image ID; label each image with a bounding box, identify the state of the safety rope, and classify it into: normal state, low hanging and high use state, virtual hanging state, and straightened state;

[0026] Step 2.2: Vector map extraction: Use edge detection method to extract the vector map of the safety rope and generate feature point information as input for subsequent models;

[0027] Step 2.3: Unify the input format: All images are uniformly resized to 224×224 pixels, and the pixel values ​​are normalized to the [0,1] interval using the normalization method;

[0028] Step 2.4: Data enhancement: Increase the diversity and robustness of samples by adding Gaussian noise, rotation, flipping and other techniques;

[0029] Step 3: Feature extraction and model selection:

[0030] Select MobileNetV2 as the base model, load the MobileNetV2 model pre-trained on ImageNet, use the weights = 'imagenet' parameter, and set the previous convolutional layer to an untrained layer;

[0031] Add a custom fully connected layer in the last layer, and set the number of output nodes to 4, corresponding to the normal state, low hanging and high use state, virtual hanging state and straightening state respectively;

[0032] Apply Softmax activation function for classification tasks;

[0033] Step 4: Data division and model training:

[0034] Step 4.1: Data division: Divide the extracted vector map dataset into training set, validation set and test set in a ratio of 7:2:1;

[0035] Step 4.2: Model training: Use the Adam optimizer and cross entropy loss function, set an appropriate learning rate; introduce the Dropout layer and early stopping strategy, monitor the validation set loss, and stop training early when it no longer decreases; use grid search to optimize hyperparameters, including learning rate, batch size, and Dropout ratio;

[0036] Step 5: Model validation and performance evaluation:

[0037] Use the validation set to test the classification performance of the model, record the classification accuracy, recall rate and F1 score, and set the target accuracy of the test set to more than 90%;

[0038] Generate a confusion matrix to analyze the performance of the model in different state categories to facilitate subsequent model adjustments;

[0039] Step 6: Model deployment: Deploy the trained model to the monitoring host so that it can monitor the safety rope status in real time and provide feedback.

[0040] Preferably, in the MobileNetV2 acquisition method, step 1 data collection and step 2 data labeling and processing can also be performed by the following method:

[0041] Step 1: Build a simulation testing platform;

[0042] Step 2: Use a safety rope with an array displacement sensor to simulate the state of the safety rope. The simulation is classified into: normal state, low hanging and high use state, virtual hanging state, and straightened state. By reading and analyzing the output data of the array displacement sensor, the spatial morphological data of the safety rope under the corresponding simulation state is obtained respectively, and the classified data is unified into a unified input format. All images are uniformly adjusted to 224×224 pixels, and the pixel values ​​are normalized to the [0,1] interval using a standardized method.

[0043] Preferably, the monitoring host communicates via a Lora wireless communication handheld monitoring terminal, and the handheld monitoring terminal is capable of receiving monitoring information sent by multiple monitoring hosts.

[0044] Another object of the present invention is to provide a method for early warning of failure of a safety belt for working at heights, using the above-mentioned early warning device for failure of a safety belt for working at heights, comprising the following steps:

[0045] S1: Workers working at heights should put on their clothes properly;

[0046] S2: Turn on the detection host and monitoring auxiliary machine, enter the real-time monitoring state, and conduct real-time status evaluation of the safety rope;

[0047] S2.1: The monitoring auxiliary machine periodically detects the spatial shape data of the safety rope through the array displacement sensor and transmits the data to the monitoring host;

[0048] S2.2: The monitoring host inputs the spatial morphology data of the safety rope into the MobileNetV2 model for classification, and divides the real-time status of the safety rope into: normal state, low hanging and high use state, virtual hanging state, and straightened state;

[0049] S3: The monitoring host sends the corresponding real-time status of the safety rope to the handheld monitoring terminal and the supervision terminal according to the real-time status evaluation result of the safety rope.

[0050] The present invention discloses a safety belt failure warning device and method for high-altitude operations, which have the following beneficial effects.

[0051] The present invention can evaluate the real-time status of the safety rope through the cooperation of the monitoring host and the monitoring auxiliary machine, and divide the real-time status of the safety rope into: normal state, low hanging and high use state, virtual hanging state, and straightened state. Then, the monitoring host uses Lora communication technology to send the real-time status of the safety rope to the handheld monitoring terminal and the supervision terminal, so that the safety supervision personnel can quickly grasp the safety status of the construction site, take necessary safety measures in time, and discover potential safety hazards in time.

[0052] By monitoring the real-time status of the safety rope and providing real-time feedback, the incidence of falling accidents during high-altitude operations has been significantly reduced, improving the safety of construction workers. The handheld monitoring terminal can monitor multiple monitoring hosts at the same time, enhancing the efficiency of safety supervision and enabling safety personnel to better manage the safety status of multiple construction workers. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a structural schematic diagram of the failure warning device of the safety belt for working at heights of the present invention.

[0054] Figure 2 The present invention is a flow chart of the method for early warning of failure of a safety belt for working at height.

[0055] In the attached figure: 1. Wearable part; 11. Back buckle; 2. Safety rope; 3. Hook; 4. Monitoring host; 5. Monitoring auxiliary machine; 51. Array displacement sensor; 6. Handheld monitoring terminal. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0057] Reference to "embodiment" herein means that the specific features, structures or characteristics described in conjunction with the embodiment may be included in at least one embodiment of the present application. The term "embodiment" appearing in various places in the specification does not necessarily refer to the same embodiment, nor does it particularly limit its independence or association with other embodiments. In principle, in the present application, as long as there is no technical contradiction or conflict, the various technical features mentioned in the embodiments can be combined in any way to form a corresponding implementable technical solution.

[0058] Unless otherwise defined, the technical terms used in this document have the same meanings as those generally understood by those skilled in the art to which this application belongs; the use of relevant terms in this document is only for describing specific embodiments and is not intended to limit this application.

[0059] In the description of this application, the term "and / or" is an expression used to describe the logical relationship between objects, indicating that three relationships may exist, for example, A and / or B, which means: A exists, B exists, and A and B exist at the same time. In addition, the character " / " in this article generally indicates that the objects before and after are in an "or" logical relationship.

[0060] Unless otherwise expressly specified or limited, in the description of the embodiments of the present application, the terms such as "install", "connect", "connect", "fix", "set", etc. used should be understood in a broad sense. For example, the "connection" can be a fixed connection, a detachable connection, or an integrated setting; it can be a mechanical connection, an electrical connection, or a communication connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium; it can be the internal connection of two elements or the interaction relationship between two elements. For technicians in the technical field to which the present application belongs, the specific meanings of the above terms in the embodiments of the present application can be understood according to the specific circumstances.

[0061] Example 1

[0062] Please refer to Figure 1 , a safety belt failure warning device for working at heights, comprising: a wearing part 1, a safety rope 2 having one end fixedly connected to the wearing part 1, and a hook 3 detachably connected to the other end of the safety rope 2;

[0063] In this embodiment, the wearable part 1 is provided with a back buckle 11, and a monitoring host 4 is provided on the outside of the back buckle 11; a monitoring sub-machine 5 is detachably connected to the outside of the safety rope 2; the monitoring sub-machine 5 is used to detect the spatial shape data of the safety rope 2, and the monitoring sub-machine 5 detects the spatial shape data of the safety rope 2 through an array displacement sensor 51; the array displacement sensor 51 can collect the spatial shape data of the safety rope 2 from the bottom of the hook 3 to the connection point of the back buckle 11; the monitoring sub-machine 5 is electrically connected to the monitoring host 4; as an option, the monitoring sub-machine 5 is connected to the monitoring host 4 through a signal line, and the monitoring sub-machine 5 and the monitoring host 4 share the same battery.

[0064] In the present application, the wearable part 1 is not specifically limited, and it is only required that the wearable part 1 can be stably connected to the safety rope 2, and the monitoring host 4 can be stably fixed on the wearable part 1. In the present embodiment, the number of safety ropes 2 is not limited, and each safety rope 2 is connected to a monitoring sub-machine 5.

[0065] The monitoring host 4 is used to determine the real-time status of the safety rope 2 according to the spatial shape data of the safety rope 2 output by the monitoring auxiliary machine 5, and send the real-time status information of the safety rope 2 to the monitoring terminal.

[0066] Preferably, in this embodiment, the monitoring host 4 communicates via the Lora wireless communication handheld monitoring terminal 6, and the handheld monitoring terminal 6 can receive monitoring information sent by multiple monitoring hosts 4, so that supervisors can simultaneously monitor multiple monitoring hosts 4. The monitoring host 4 uses the Lora communication technology to send the real-time status of the safety rope 2 to the handheld monitoring terminal 6 and the supervisory terminal.

[0067] In this embodiment, the handheld monitoring terminal 6 is a device used by safety supervisors at the bottom of the construction site. It has a display screen for real-time viewing of the real-time status of the safety rope 2 and the corresponding monitoring host 4 information. As an option, the corresponding operator information can be input into the corresponding monitoring host 4 before performing high-altitude operations. The monitoring host 4 transmits the personnel information back when transmitting the real-time status of the safety rope 2. By monitoring the real-time status of the safety rope 2 and providing real-time feedback, the incidence of falling accidents in high-altitude operations is significantly reduced, and the safety of construction workers is improved; the handheld monitoring terminal 6 can monitor multiple monitoring hosts 4 at the same time, which enhances the efficiency of safety supervision and enables safety personnel to better manage the safety status of multiple construction workers. As an option, in this implementation, the monitoring host 4 sends the power information of the monitoring host 4 to the handheld monitoring terminal 6 in real time.

[0068] Preferably, in the present embodiment, the array displacement sensor 51 is integrated into a wear-resistant cable, and the wear-resistant cable is tightly wrapped with the safety rope 2 to reduce external interference and improve durability.

[0069] Preferably, in the present embodiment, the real-time states of the safety rope 2 include: normal state, low-hanging and high-use state, virtual hanging state, and straightened state; the normal state is the normal use state of the safety rope 2, which meets the high-hanging resistance requirement, that is, the safety belt is hung at a place higher than the standing position of the person, and the place where the personnel work is lower than the position where the safety belt is hung; the low-hanging and high-use state is the opposite of the high-hanging resistance; the virtual hanging state is when the safety belt hook 3 is partially in a placed state, for example, placed on stairs or shelves; the straightened state is the state when all safety ropes 2 are straightened to indicate that the construction personnel have slipped.

[0070] Example 2

[0071] Based on Example 1, please refer to Figure 2 In this embodiment, the real-time status evaluation of the safety rope 2 is based on the following method:

[0072] Step 1: Monitor host 4 to read the deployed MobileNetV2 model;

[0073] Step 2: The monitoring auxiliary machine 5 detects the spatial shape data of the safety rope 2 through the array displacement sensor 51, and transmits the data to the monitoring host machine 4;

[0074] Preferably, in step 2, the monitoring auxiliary machine 5 detects the spatial shape data of the safety rope 2 through the array displacement sensor 51 by the following method:

[0075] Step 2.1: The monitoring auxiliary machine 5 captures the spatial coordinates x, y, z of each part of the safety rope 2 in the three-dimensional space in real time through the array displacement sensor 51, and based on the acquired spatial coordinates x, y, z of each part of the safety rope 2 in the three-dimensional space, uses an interpolation algorithm to fill the gaps between the data to form a continuous dot matrix distribution;

[0076] Step 2.2: Generate a vector diagram using the generated dot matrix data to represent the overall shape of the safety rope 2. The features of the vector diagram include the length, bending degree and tilt angle of the safety rope 2.

[0077] Step 2.3: Standardize the generated vector image.

[0078] Step 3: The monitoring host 4 inputs the spatial morphological data of the safety rope 2 into the MobileNetV2 model for classification, and divides the real-time status of the safety rope 2 into: normal status, low hanging and high use status, virtual hanging status, and straightened status;

[0079] Step 4: Send the real-time status information of the safety rope 2 to the monitoring terminal.

[0080] Preferably, in this embodiment, the MobileNetV2 acquisition method is:

[0081] Step 1: Data collection: Collect images of safety rope 2 in different states; directly using the images of safety rope 2 as the data set can save the time of data set collection. You can directly use the images of safety rope 2 on the Internet, which can ensure the diversity of the data set and the number of samples in the data set can be large enough to make the subsequent model training more accurate.

[0082] Step 2: Data labeling and processing:

[0083] Step 2.1: Label content: Use the LabelStudio tool to label and build a database containing information such as the labeler, labeling time, and image ID; label each image with a bounding box, identify the state of the safety rope 2, and classify it into: normal state, low hanging and high use state, virtual hanging state, and straightened state;

[0084] Step 2.2: Vector map extraction: Use edge detection method to extract the vector map of safety rope 2 and generate feature point information as input for subsequent models;

[0085] Step 2.3: Unify the input format: All images are uniformly resized to 224×224 pixels, and the pixel values ​​are normalized to the [0,1] interval using the normalization method;

[0086] Step 2.4: Data enhancement: Increase the diversity and robustness of samples by adding Gaussian noise, rotation, flipping and other techniques;

[0087] Step 3: Feature extraction and model selection:

[0088] Select MobileNetV2 as the base model, load the MobileNetV2 model pre-trained on ImageNet, use the weights = 'imagenet' parameter, and set the previous convolutional layer to an untrained layer;

[0089] Add a custom fully connected layer in the last layer, and set the number of output nodes to 4, corresponding to the normal state, low hanging and high use state, virtual hanging state and straightening state respectively;

[0090] Apply Softmax activation function for classification tasks;

[0091] Step 4: Data division and model training:

[0092] Step 4.1: Data division: Divide the extracted vector map dataset into training set, validation set and test set in a ratio of 7:2:1;

[0093] Step 4.2: Model training: Use the Adam optimizer and cross entropy loss function, set an appropriate learning rate; introduce the Dropout layer (set to 0.5) and the early stopping strategy, monitor the validation set loss, and stop training early when it no longer decreases; use grid search to optimize hyperparameters, including learning rate, batch size, and Dropout ratio;

[0094] Step 5: Model validation and performance evaluation:

[0095] Use the validation set to test the classification performance of the model, record the classification accuracy, recall rate and F1 score, and set the target accuracy of the test set to more than 90%;

[0096] Generate a confusion matrix to analyze the performance of the model in different state categories to facilitate subsequent model adjustments;

[0097] Step 6: Model deployment: Export the trained model to ONNX or TensorFlow Saved Model format, and then deploy the trained model to the monitoring host 4 so that it can monitor the status of the safety rope 2 in real time and provide feedback.

[0098] Example 3

[0099] Based on Example 2, as a preference, in the MobileNetV2 acquisition method in this embodiment, step 1 data collection and step 2 data annotation and processing can also be performed by the following method:

[0100] Step 1: Build a simulation testing platform;

[0101] Step 2: Use the safety rope 2 with the array displacement sensor 51 to simulate the state of the safety rope 2. The simulation is classified into: normal state, low hanging and high use state, virtual hanging state, and straightened state; obtain the spatial morphological data of the safety rope 2 under the corresponding simulation state by reading and analyzing the output data of the array displacement sensor 51, and input the obtained classified data in a unified format, adjust all images to 224×224 pixels, and use the standardization method to normalize the pixel values ​​to the [0,1] interval.

[0102] Using the safety rope 2 with the array displacement sensor 51 to simulate the state of the safety rope 2 and then collecting the corresponding database can make the trained model more suitable for the use of the high-altitude operation safety belt failure warning device of the present application, and can make the real-time state of the safety rope 2 more accurate.

[0103] Example 4

[0104] Based on Examples 1-3, please refer to Figure 2 This embodiment provides a method for early warning of failure of a safety belt for working at height, comprising the following steps:

[0105] S1: The person working at height should put on the wearing part 1;

[0106] S2: Turn on the detection host and the monitoring auxiliary machine 5, enter the real-time monitoring state, and conduct real-time status evaluation of the safety rope 2;

[0107] S2.1: The monitoring auxiliary machine 5 regularly detects the spatial shape data of the safety rope 2 through the array displacement sensor 51, and transmits the data to the monitoring host 4;

[0108] S2.2: The monitoring host 4 inputs the spatial morphological data of the safety rope 2 into the MobileNetV2 model for classification, and divides the real-time status of the safety rope 2 into: normal status, low hanging and high use status, virtual hanging status, and straightened status;

[0109] S3: The monitoring host 4 sends the corresponding real-time status of the safety rope 2 to the handheld monitoring terminal 6 and the supervision terminal according to the real-time status evaluation result of the safety rope 2.

[0110] Through the cooperation of the monitoring host 4 and the monitoring auxiliary machine 5, the real-time status of the safety rope 2 can be evaluated, and the real-time status of the safety rope 2 can be divided into: normal state, low hanging high use state, virtual hanging state, straightened state, and then the real-time status of the safety rope 2 is sent to the handheld monitoring terminal 6 and the supervision terminal through the monitoring host 4 using Lora communication technology, so that the safety supervisor can quickly grasp the safety status of the construction site and take necessary safety measures in time. Potential safety hazards can be discovered in time.

[0111] The above are only preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The substitution may be a substitution of a part of the structure, device, method step, or a complete technical solution. The equivalent replacement or change according to the technical solution and the inventive concept of the present invention shall be covered within the protection scope of the present invention.

Claims

1. A safety belt failure warning device for working at heights, characterized in that: include: A wearable part, a safety rope fixedly connected to the wearable part at one end, and a hook detachably connected to the other end of the safety rope; The wearable part is provided with a back buckle, and a monitoring host is provided on the outside of the back buckle; a monitoring auxiliary machine is detachably connected to the outside of the safety rope; the monitoring auxiliary machine is used to detect the spatial shape data of the safety rope, and the monitoring auxiliary machine detects the spatial shape data of the safety rope through an array displacement sensor; the array displacement sensor can collect the spatial shape data of the safety rope from the bottom of the hook to the back buckle connection point; the monitoring auxiliary machine is electrically connected to the monitoring host; The monitoring host is used to determine the real-time status of the safety rope according to the safety rope spatial shape data output by the monitoring auxiliary machine, and send the real-time status information of the safety rope to the supervision terminal.

2. The high-altitude operation safety belt failure warning device according to claim 1 is characterized in that The array displacement sensor is integrated into the wear-resistant cable, and the wear-resistant cable is tightly wrapped with the safety rope.

3. The high-altitude operation safety belt failure warning device according to claim 1, characterized in that: The real-time status of the safety rope includes: normal status, low hanging and high use status, virtual hanging status, and straightened status.

4. The high-altitude operation safety belt failure warning device according to claim 3 is characterized in that: The real-time status assessment of the safety rope is based on the following method: Step 1: The monitoring host reads the deployed MobileNetV2 model; Step 2: The monitoring auxiliary machine detects the spatial shape data of the safety rope through the array displacement sensor and transmits the data to the monitoring host; Step 3: The monitoring host inputs the spatial morphology data of the safety rope into the MobileNetV2 model for classification, and divides the real-time status of the safety rope into: normal state, low hanging and high use state, virtual hanging state, and straightened state; Step 4: Send the real-time status information of the safety rope to the monitoring terminal.

5. The high-altitude work safety belt failure warning device according to claim 4, characterized in that: In step 2, the monitoring auxiliary machine detects the spatial shape data of the safety rope through the array displacement sensor by the following method: Step 2.1: The monitoring auxiliary machine captures the spatial coordinates x, y, and z of each part of the safety rope in three-dimensional space in real time through an array displacement sensor, and based on the acquired spatial coordinates x, y, and z of each part of the safety rope in three-dimensional space, uses an interpolation algorithm to fill in the gaps between the data to form a continuous dot matrix distribution; Step 2.2: Generate a vector diagram using the generated dot matrix data to represent the overall shape of the safety rope. The features of the vector diagram include the length, curvature and tilt angle of the safety rope. Step 2.3: Standardize the generated vector image.

6. The high-altitude work safety belt failure warning device according to claim 4, characterized in that: The MobileNetV2 acquisition method is: Step 1: Data collection: collect safety rope status images in different states; Step 2: Data labeling and processing: Step 2.1: Label content: Use the Label Studio tool to label and build a database containing information such as the labeler, labeling time, and image ID; label each image with a bounding box, identify the state of the safety rope, and classify it into: normal state, low hanging and high use state, virtual hanging state, and straightened state; Step 2.2: Vector map extraction: Use edge detection method to extract the vector map of the safety rope and generate feature point information as input for subsequent models; Step 2.3: Unify the input format: All images are uniformly resized to 224×224 pixels, and the pixel values ​​are normalized to the [0,1] interval using the normalization method; Step 2.4: Data enhancement: Increase the diversity and robustness of samples by adding Gaussian noise, rotation, flipping and other techniques; Step 3: Feature extraction and model selection: Select MobileNetV2 as the base model, load the MobileNetV2 model pre-trained on ImageNet, use the weights = 'imagenet' parameter, and set the previous convolutional layer to an untrained layer; Add a custom fully connected layer in the last layer, and set the number of output nodes to 4, corresponding to the normal state, low hanging and high use state, virtual hanging state and straightening state respectively; Apply Softmax activation function for classification tasks; Step 4: Data division and model training: Step 4.1: Data division: Divide the extracted vector map dataset into training set, validation set and test set in a ratio of 7:2:1; Step 4.2: Model training: Use the Adam optimizer and cross entropy loss function, set an appropriate learning rate; introduce the Dropout layer and early stopping strategy, monitor the validation set loss, and stop training early when it no longer decreases; use grid search to optimize hyperparameters, including learning rate, batch size, and Dropout ratio; Step 5: Model validation and performance evaluation: Use the validation set to test the classification performance of the model, record the classification accuracy, recall rate and F1 score, and set the target accuracy of the test set to more than 90%; Generate a confusion matrix to analyze the performance of the model in different state categories to facilitate subsequent model adjustments; Step 6: Model deployment: Deploy the trained model to the monitoring host so that it can monitor the safety rope status in real time and provide feedback.

7. The high-altitude work safety belt failure warning device according to claim 6, characterized in that: In the MobileNetV2 acquisition method, step 1 data collection and step 2 data annotation and processing can also be performed by the following methods: Step 1: Build a simulation testing platform; Step 2: Use a safety rope with an array displacement sensor to simulate the state of the safety rope. The simulation is classified into: normal state, low hanging and high use state, virtual hanging state, and straightened state. By reading and analyzing the output data of the array displacement sensor, the spatial morphological data of the safety rope under the corresponding simulation state is obtained respectively, and the classified data is unified into a unified input format. All images are uniformly adjusted to 224×224 pixels, and the pixel values ​​are normalized to the [0,1] interval using a standardized method.

8. The high-altitude work safety belt failure warning device according to claim 1, characterized in that: The monitoring host communicates via the Lora wireless communication handheld monitoring terminal, and the handheld monitoring terminal can receive monitoring information sent by multiple monitoring hosts.

9. A method for early warning of failure of a safety belt for working at heights, using the device for early warning of failure of a safety belt for working at heights as claimed in any one of claims 1 to 8, characterized in that: The steps include: S1: The workers who are working at heights should put on their clothes properly; S2: Turn on the detection host and monitoring auxiliary machine, enter the real-time monitoring state, and conduct real-time status assessment of the safety rope; S2.1: The monitoring auxiliary machine periodically detects the spatial shape data of the safety rope through the array displacement sensor and transmits the data to the monitoring host; S2.2: The monitoring host inputs the spatial morphological data of the safety rope into the MobileNetV2 model for classification, and divides the real-time status of the safety rope into: normal state, low hanging and high use state, virtual hanging state, and straightened state; S 3: The monitoring host sends the corresponding real-time status of the safety rope to the handheld monitoring terminal and the supervision terminal according to the real-time status evaluation result of the safety rope.

Images