Intelligent monitoring and early warning system for curtain wall unit plate hoisting

Abstract

The embodiment of the present application relates to the technical field of building engineering, and discloses a curtain wall unit plate hoisting intelligent monitoring and early warning system, which comprises: a hoisting device for hoisting a target curtain wall unit plate; a winch for receiving control instructions of a control device to move the hoisting device based on the control instructions; an IMU sensor for acquiring the motion state of the target curtain wall unit plate in real time; a starry night intelligent safety helmet worn by a construction site operator for real-time monitoring and reporting the personnel motion state including the position of the operator; a detection device for generating a corrected system state vector based on the motion state, the personnel motion state and an estimated system state vector; the predicted position and the predicted speed at each time based on the corrected system state vector are used to predict the early warning time of performing a first-level early warning, and the early warning time is sent to the control device to control the winch to stop working. The winch is triggered to stop moving by predicting the early warning time of performing a first-level early warning, thereby ensuring the safety of high-altitude operation.

Description

Technical Field

[0001] This invention relates to the field of building engineering technology, and in particular to an intelligent monitoring and early warning system for the hoisting of curtain wall unit panels. Background Technology

[0002] In the hoisting and installation of curtain walls for high-rise buildings, traditional methods rely primarily on manual command and experience-based operation. They lack real-time, precise quantitative monitoring of changes in the posture, speed, and position of the hoisting units. Especially at high altitudes, in strong winds, or under complex conditions, curtain wall panels are prone to swaying, twisting, or shifting, potentially leading to collisions with the building structure or other equipment, posing significant safety risks. While some existing hoisting systems are equipped with sensors and control devices, they are mostly limited to single signal acquisition or simple threshold alarms, resulting in low sensing accuracy, delayed system response, and difficulty in predicting dynamic movement trends. Furthermore, the sensing and control processes are disconnected, lacking effective data fusion and closed-loop feedback mechanisms, making it impossible to promptly identify potential hazards and take proactive intervention measures. Summary of the Invention

[0003] The purpose of this invention is to provide at least one intelligent monitoring and early warning system for the hoisting of curtain wall unit panels, which can at least solve the technical problem of significant safety risks during the installation process and the inability to identify potential dangers and take proactive intervention measures in a timely manner. It can at least effectively improve the real-time monitoring capability and safety response speed of the hoisting process, realize the transformation from passive protection to proactive early warning, and ensure the technical effect of ensuring the safety of high-altitude operations.

[0004] To address the aforementioned technical problems, at least one embodiment of this application provides an intelligent monitoring and early warning system for the hoisting of curtain wall unit panels, comprising: hoisting equipment for hoisting target curtain wall unit panels; a winch for receiving control commands from a control device to move the hoisting equipment based on the control commands, thereby moving the target curtain wall unit panels; an IMU sensor installed on the hoisting equipment for acquiring the motion state of the target curtain wall unit panels in real time and sending the motion state to a detection device; and a Xingxuan intelligent safety helmet worn by construction site workers for real-time monitoring of their own position and sending the personnel's motion state to a detection device. The detection device is configured to receive the motion state sent by the IMU sensor, receive the personnel motion state sent by the Xingxuan smart safety helmet, generate a corrected system state vector based on the motion state, the personnel motion state, and the estimated system state vector, and predict the predicted position and predicted speed of the target curtain wall unit panel at each moment within a preset time period based on the corrected system state vector; the detection device is also configured to predict the warning time for executing a first-level warning based on the predicted position and predicted speed at each moment within the preset time period, and send the warning time to the control device to control the winch to stop working.

[0005] This solution integrates hoisting equipment, winches, IMU sensors, Xingxuan smart safety helmets, and detection equipment to construct a closed-loop intelligent hoisting control system. This system achieves high-precision state perception, dynamic prediction, and proactive safety control during the hoisting of curtain wall unit panels. IMU sensors collect the motion state of the target curtain wall unit in real time. The detection equipment combines measured data with estimated system states for correction, generating a high-precision corrected system state vector. Based on this vector, it predicts the position and velocity trends over future time periods. Simultaneously, the system integrates personnel position information from Xingxuan smart safety helmets to construct a comprehensive state representation encompassing multiple elements of "mechanism-structure-person," significantly enhancing situational awareness in complex construction environments. The detection equipment can identify potential risk trajectories in advance, and the system can predict the timing of a Level 1 warning and promptly send instructions to the control equipment, triggering the winch to stop. This solution effectively improves real-time monitoring capabilities and safety response speed during the hoisting process, achieving a shift from passive protection to proactive warning. It significantly reduces the risk of collisions caused by swaying, deviation, malfunction, or misoperation, ensuring the safety, stability, and construction accuracy of high-altitude operations.

[0006] In some examples, the system further includes: a display device for displaying an overall view of the moving target curtain wall unit panel, the overall view including a three-dimensional view of the building where the target curtain wall unit panel is installed, the vertical transport cableway and loop track for the winch to transport the target curtain wall unit panel, and the real-time position of the target curtain wall unit panel; the display device is also used to display a key focus view of the moving target curtain wall unit panel, the key focus view including a view of the first floor where the target curtain wall unit panel is placed.

[0007] In some examples, when the detection device is used to predict the warning time for executing a Level 1 warning based on the predicted position and the predicted speed at each time within the preset time period, it is specifically used to: predict whether to issue a Level 1 warning based on the predicted position and the predicted speed at each time within the preset time period; if yes, then predict the warning time for executing a Level 1 warning based on the predicted position and the predicted speed at each time within the preset time period; if no, then predict whether to issue a Level 2 warning based on the predicted position and the predicted speed at each time within the preset time period; if no, then no warning is issued.

[0008] In some examples, the system further includes: if it is determined that a secondary warning is required, the detection device is also used to generate a voice warning and send the voice warning to the Xingxuan smart safety helmet; the Xingxuan smart safety helmet is used to receive the voice warning sent by the detection device and broadcast the voice warning to remind the worker wearing the Xingxuan smart safety helmet that there is a risk of collision between the target curtain wall unit panel and the obstacle.

[0009] In some examples, when the detection device is used to predict whether to issue a Level 1 warning based on the predicted position and predicted speed at each moment within the preset time period, it is specifically used to: obtain a preset prediction covariance, a random variable, and a preset static safety distance; determine a dynamic safety distance based on the predicted speed at each moment within the preset time period; obtain at least one obstacle position; determine the danger zone corresponding to the predicted position at each moment based on the obstacle positions, the preset static safety distance, and the dynamic safety distance; predict the collision probability of the predicted position being in the danger zone based on the predicted position, the random variable, and the danger zone; select the maximum collision probability from the collision probabilities corresponding to each predicted position; if the maximum collision probability is not less than a preset first collision threshold, it is considered that a Level 1 warning is required.

[0010] In some examples, when the detection device is used to predict the warning time for executing a Level 1 warning based on the predicted position and the predicted speed at each time within the preset time period, it is specifically used to: obtain the maximum deceleration of the winch; determine the object distance between each obstacle position and the predicted position at each time within the preset time period; determine the minimum braking distance based on the predicted speed and the maximum deceleration at each time; if there is no object distance greater than the minimum braking distance among at least one object distance at each time, then determine that time as the warning time for executing the Level 1 warning.

[0011] In some examples, when the detection device is used to predict whether to issue a secondary warning based on the predicted position and the predicted speed at each moment within the preset time period, it is specifically used to: if the maximum collision probability is less than the preset first collision threshold and not less than the preset second collision threshold, then it is considered that a secondary warning needs to be issued.

[0012] In some examples, the system further includes: a path planning device for acquiring digital twin information and determining an optimal global path based on the digital twin information and the MARL algorithm, and sending the optimal global path to a detection device; the digital twin information includes all entity information of the building on which the target curtain wall unit panel is installed; the detection device for receiving the optimal global path sent by the path planning device and predicting whether there is a position on the optimal global path where the distance to an obstacle is less than a safe distance; if so, it acquires the current position of the target curtain wall unit panel and determines a self-healing path based on the current position and the target position, so that the control device guides the winch to move the target curtain wall unit panel based on the optimal global path and / or the self-healing path; the safe distance is the sum of a preset static safe distance and a dynamic safe distance.

[0013] In some examples, the path planning device, when used to determine the optimal global path based on the digital twin information and the MARL algorithm, specifically performs the following steps: Based on the digital twin information, it constructs the state space and action space of each executing agent; obtains a preset multi-objective reward function, which comprehensively considers hoisting progress, safety, hoisting efficiency, and motion comfort, and is used to calculate the immediate reward obtained by the executing agent performing a specific action in the current state, wherein the current state is the state space and the specific action belongs to the action space; uses the MARL algorithm to train each executing agent based on its current state, specific action, and immediate reward to solve for the optimal cooperative strategy; and generates the optimal global path from the starting position to the target position according to the optimal cooperative strategy.

[0014] In some examples, the detection device, when used to determine a self-healing path based on the current location and the target location, specifically involves: using the current location as the starting node for dynamic replanning, constructing digital twin information containing multiple candidate nodes; for each candidate node, employing a heuristic search algorithm to calculate its comprehensive cost value, the comprehensive cost value including the cumulative actual cost from the starting node to the candidate node, the heuristically predicted cost from the candidate node to the target location, and the obstacle avoidance cost from the candidate node to the obstacle; based on each comprehensive cost value, traversing the digital twin information to generate an optimal alternative path from the starting node to the target location; and using the optimal alternative path as the self-healing path. Attached Figure Description

[0015] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0016] Figure 1 This is a schematic diagram of the structure of an intelligent monitoring and early warning system for the hoisting of curtain wall unit panels provided in one embodiment of this application;

[0017] Figure 2 This is a structural schematic diagram of the Xingxuan smart safety helmet provided in another embodiment of this application;

[0018] Figure 3 This is a schematic diagram of the structure of a display device provided in one embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.

[0020] To facilitate understanding of the embodiments of this application, the relevant content of the intelligent monitoring and early warning system for curtain wall unit panel hoisting will be introduced first.

[0021] In the hoisting and installation of curtain walls for high-rise buildings, traditional methods rely primarily on manual command and experience-based operation. They lack real-time, precise quantitative monitoring of changes in the posture, speed, and position of the hoisting units. Especially at high altitudes, in strong winds, or under complex conditions, curtain wall panels are prone to swaying, twisting, or shifting, potentially leading to collisions with the building structure or other equipment, posing significant safety risks. While some existing hoisting systems are equipped with sensors and control devices, they are mostly limited to single signal acquisition or simple threshold alarms, resulting in low sensing accuracy, delayed system response, and difficulty in predicting dynamic movement trends. Furthermore, the sensing and control processes are disconnected, lacking effective data fusion and closed-loop feedback mechanisms, making it impossible to promptly identify potential hazards and take proactive intervention measures.

[0022] To address the significant safety risks during the hoisting process and the inability to promptly identify potential hazards and take proactive intervention measures, this invention proposes an intelligent monitoring and early warning system for curtain wall unit panel hoisting. The implementation details of this embodiment of the intelligent monitoring and early warning system for curtain wall unit panel hoisting are described below. The following content is for illustrative purposes only and is not essential for implementing this solution.

[0023] Example 1:

[0024] The intelligent monitoring and early warning system for curtain wall unit panel hoisting in this embodiment has the following specific process: Figure 1 As shown, it includes:

[0025] Lifting equipment 110 is used for lifting the target curtain wall unit panels.

[0026] Here, "curtain wall unit panel" refers to a prefabricated curtain wall module in the factory, and "target curtain wall unit panel" refers to the curtain wall module that needs to be hoisted and installed at the designated location. Hoisting equipment can be a spreader beam or a hook.

[0027] The winch 120 is used to receive control commands from the control device 100 to move the hoisting equipment based on the control commands, thereby moving the target curtain wall unit panel.

[0028] The IMU sensor 130 is installed at the hoisting equipment to acquire the motion status of the target curtain wall unit panel in real time and send the motion status to the detection equipment.

[0029] The motion state refers to the motion state of the target curtain wall unit panel, including its position, velocity, acceleration, and attitude angle. This motion state is acquired based on IMU sensors. The IMU sensors provide relatively accurate real-time motion state of the target curtain wall unit panel.

[0030] The Xingxuan Smart Safety Helmet 000 is worn by construction site workers to monitor their movement status in real time, including their own location, and sends the movement status to the detection device 140.

[0031] The detection device 140 is used to receive the motion state sent by the IMU sensor, receive the personnel motion state sent by the Xingxuan smart safety helmet, generate a corrected system state vector based on the motion state, personnel motion state and the estimated system state vector, and predict the predicted position and predicted speed of the target curtain wall unit panel at each moment within a preset time period based on the corrected system state vector.

[0032] The predicted system state vector refers to the prediction of the target curtain wall unit panel's state vector at the current moment by using IMU pre-integration to predict the historical system state vector. The historical system state vector can be the system state vector of the target curtain wall unit panel at the previous moment. The preset time period refers to a time period starting from the current moment, lasting a preset duration, and ending after the current moment. The predicted position and predicted velocity correspond one-to-one with the moments within the preset time period; that is, the predicted position and predicted velocity refer to the predicted position and velocity of the target curtain wall unit panel at a specified moment within the preset time period. The predicted position and predicted velocity also correspond one-to-one, representing the predicted position and velocity of the target curtain wall unit panel at a specific moment. Personnel movement status includes the wearer's position, movement velocity, and acceleration detected by the Xingxuan intelligent safety helmet.

[0033] The detection device 140 is also used to predict the warning time for executing the first-level warning based on the predicted position and predicted speed at each time within a preset time period, and send the warning time to the control device 100 to control the winch to stop working.

[0034] The highest level of warning is Level 1, which requires emergency braking. The warning time is the moment when emergency braking must be initiated.

[0035] Specifically, when the warning time is determined, a first-level braking signal is sent to the control equipment to control the winch to stop working.

[0036] For example, the braking signal is integrated with the existing winch control signal through hardware logic circuitry to ensure the highest priority and reliability. See the following formula:

[0037]

[0038] in, This indicates a logical OR. This is a braking signal. This is the original winch control signal; This is the final winch control signal; once This is a first-level braking signal, regardless of What value should be output? Both will cut off power to achieve an emergency stop of the winch.

[0039] Therefore, by integrating hoisting equipment, winches, IMU sensors, and detection equipment, a closed-loop intelligent hoisting control system was constructed, achieving high-precision state perception, dynamic prediction, and proactive safety control of the curtain wall unit panel hoisting process. IMU sensors collect the motion state of the target curtain wall unit in real time, and the detection equipment combines the measured data with the estimated system state for fusion correction, generating a high-precision corrected system state vector, and predicting the position and speed change trends over future time periods based on this vector. By identifying potential risk trajectories in advance, the system can predict the warning time for executing a level-one warning and promptly send instructions to the control equipment, triggering the winch to stop. This solution effectively improves the real-time monitoring capability and safety response speed of the hoisting process, greatly enhancing the safety and intelligence level of the curtain wall hoisting process. It achieves a shift from passive protection to proactive early warning, significantly reducing the collision risk caused by swaying, offset, malfunction, or misoperation, ensuring the safety, stability, and construction accuracy of high-altitude operations.

[0040] Specifically, when the detection equipment is used to generate a corrected system state vector based on the motion state, the personnel motion state, and the estimated system state vector, it is used to: obtain the time-varying non-line-of-sight error estimation state; determine the system state vector based on the time-varying non-line-of-sight error estimation state, the personnel motion state, and the motion state; and generate a corrected system state vector based on the system state vector and the estimated system state vector.

[0041] The system state vector includes the time-varying non-line-of-sight error estimation state, the current personnel movement state, and the current real-time movement state of the target curtain wall unit panel. The non-line-of-sight error estimation state is used to actively estimate the error impact of environmental changes on the positioning of the target curtain wall unit panel, also known as... .

[0042] For example, the system state vector can be represented by the following formula:

[0043]

[0044] in, The position, velocity, and acceleration of the target curtain wall unit panel in the world coordinate system; The attitude angle; This is the TOA NLOS error state. Let k be the k-th moment, which is the current moment. This refers to the position, velocity, and acceleration of a person in motion within the world coordinate system.

[0045] Specifically, the detection device 140 is also used to acquire the estimated system state vector. The acquisition method involves acquiring the historical rotation matrix, historical personnel velocity, historical personnel position, historical velocity, historical position, current acceleration, current personnel acceleration, and IMU bias from the previous moment. Here, "previous moment" refers to the moment before the current moment. Based on the historical rotation matrix, historical personnel velocity, historical personnel position, historical velocity, historical position, current acceleration, current personnel acceleration, and IMU bias, the predicted position and velocity, as well as the predicted personnel position and velocity, are predicted at the current moment. The estimated system state vector is then generated based on the predicted position, predicted velocity, acceleration, predicted personnel position, predicted personnel velocity, personnel acceleration, attitude angle, and non-line-of-sight error estimation state.

[0046] The IMU zero bias includes the zero bias of the accelerometer and the zero bias of the gyroscope.

[0047] For example, based on the historical rotation matrix, historical personnel velocity, historical personnel position, historical velocity, historical position, current acceleration, current personnel acceleration, and IMU zero bias, the predicted position and velocity at the current moment, as well as the predicted personnel position and velocity, can be predicted using the following formula:

[0048] ,

[0049] ,

[0050] ,

[0051]

[0052] in, For historical rotation matrix, It is the gravity vector. This is the zero bias of the accelerometer in the IMU zero bias. For acceleration; For historical speed; For historical position; This is the time interval between the previous moment and the current moment. For predicting speed; To predict the location. For the positions of historical personnel; To predict the speed of personnel; To predict the location of personnel. For historical personnel speed. Accelerate personnel

[0053] For example, the predicted system state vector can be expressed as the following formula:

[0054]

[0055] in, For the aforementioned .

[0056] Therefore, it can be seen that the IMU kinematic prediction method based on this pre-integration can provide high-frequency, short-time, and high-precision state prediction.

[0057] In addition, the detection equipment is also used to obtain the angular velocity at the current moment, and predict the rotation matrix at the current moment based on the historical rotation matrix, angular velocity, and IMU zero bias.

[0058] For example, the rotation matrix at the current moment can be predicted based on the historical rotation matrix, angular velocity, and IMU bias, as shown in the following formula:

[0059]

[0060] in, For rotation matrix, For historical rotation matrix; Angular velocity, This refers to the zero bias of the gyroscope in the IMU zero bias. This is the time interval between the previous moment and the current moment.

[0061] Furthermore, when the detection device 140 generates a corrected system state vector based on the system state vector and the estimated system state vector, it specifically performs the following steps: inputting the system state vector into the target observation model to obtain a theoretical TOA observation vector. The target observation model models the system state vector by explicitly including the non-line-of-sight error estimation state, calculates the theoretical TOA observation values ​​of each base station and the target curtain wall unit panel under the non-line-of-sight error estimation state, and generates a theoretical TOA observation vector; obtaining a predicted TOA observation vector based on the estimated system state vector; generating an innovation sequence based on the theoretical TOA observation vector and the predicted TOA observation vector; determining the Mahalanobis distance based on a preset innovation covariance matrix and the innovation sequence; obtaining a corrected observation matrix based on the Mahalanobis distance; and obtaining the corrected system state vector based on the corrected observation matrix, the innovation covariance matrix, and the estimated system state vector.

[0062] For example, the system state vector is input into the target observation model to obtain the theoretical TOA observation vector, as shown in the following formula:

[0063]

[0064] in, This refers to the vector composed of the TOA observations of all base stations, also known as the theoretical TOA observation vector. The TOA observation value is the measured value of the time it takes for a signal to propagate from the base station to the target curtain wall unit panel. The nonlinear observation function takes the expanded system state vector as input. The output is the theoretical TOA observation vector. This refers to the observation noise vector.

[0065] For example, the predicted TOA observation vector is represented as . To predict the system state vector.

[0066] For example, an innovation sequence is generated based on the theoretical TOA observation vector and the predicted TOA observation vector, as shown in the following formula:

[0067]

[0068] in, For innovation sequences, under line-of-sight conditions, the innovation sequences follow a zero-mean Gaussian distribution. For the theoretical TOA observation vector, To predict the TOA observation vector.

[0069] For example, the Mahalanobis distance is determined based on a preset innovation covariance matrix and innovation sequence, as shown in the following formula:

[0070]

[0071] in, To innovate the covariance matrix; For innovation sequence; This is the Mahalanobis distance.

[0072] For example, the corrected observation matrix can be obtained based on Mahalanobis distance, as shown in the following formula:

[0073]

[0074] in, The observation matrix; It is the expansion factor. The preset non-line-of-sight error variation frequency; This is the preset frequency of change in sight distance error. This is a preset value.

[0075] For example, the corrected system state vector can be obtained based on the corrected observation matrix, the innovative covariance matrix, and the predicted system state vector, as shown in the following formula:

[0076]

[0077]

[0078] in, For innovation sequence; To correct the system state vector; The Mann gain matrix; To predict the system state vector. It is a Jacobian matrix; The Mann gain matrix is... The observation matrix; Let be the prior state covariance.

[0079] For example, the system state vector is modified. , can be represented as:

[0080]

[0081] in, This is the corrected position for the current moment. This is the corrected speed at the current moment; This is the corrected acceleration for the current moment; This is the corrected attitude angle for the current moment. This is the corrected non-line-of-sight error estimate for the current moment.

[0082] Therefore, the corrected system state vector comprehensively considers the actual measurement data (system state vector) and the predicted data (estimated system state vector), and updates the current system state to obtain a more accurate and reliable corrected system state vector than the system state vector or the estimated system state vector.

[0083] Specifically, when the detection device 140 is used to predict the predicted position and predicted velocity of the target curtain wall unit panel at each moment within a preset time period based on the corrected system state vector, it is specifically used to: predict the predicted position of the moment based on the moment, the current moment, the corrected position of the current moment, the corrected velocity, and the corrected acceleration at each moment within the preset time period, and predict the predicted velocity of the moment based on the corrected velocity, the moment, the current moment, and the corrected acceleration.

[0084] For example, the predicted position of a given time is predicted based on the time, the current time, the corrected position at the current time, the corrected velocity, and the corrected acceleration, as shown in the following formula:

[0085]

[0086] in, The distance from the current time The duration is And the predicted position corresponding to a time later than the current time; This is the corrected position at the current moment. The correction speed at the current moment; The corrected acceleration for the current moment; This represents the time interval between the current moment and the previous moment.

[0087] For example, the predicted velocity at a given time can be calculated based on the corrected velocity, the time, the current time, and the corrected acceleration, as shown in the following formula:

[0088]

[0089] in, The distance from the current time The duration is And the prediction speed is later than the time corresponding to the current moment; The correction speed at the current moment; The corrected acceleration for the current moment; This is the time interval between the current time and the previous time. Δt is greater than or equal to 0 and less than or equal to the duration of the preset time interval.

[0090] Specifically, when the detection device 140 is used to predict the warning time for implementing a first-level warning based on the predicted position and predicted speed at each moment within a preset time period, it is specifically used to: predict whether to implement a first-level warning based on the predicted position and predicted speed at each moment within the preset time period; if yes, then predict the warning time for implementing a first-level warning based on the predicted position and predicted speed at each moment within the preset time period; if no, then predict whether to implement a second-level warning based on the predicted position and predicted speed at each moment within the preset time period; if no, then no warning is issued.

[0091] Among them, Level 1 warning has a higher priority than Level 2 warning, and Level 1 warning is more urgent than Level 2 warning. The warning measures adopted for Level 1 warning are more mandatory than those adopted for Level 2 warning.

[0092] Furthermore, when the detection device 140 is used to predict whether to issue a Level 1 warning based on the predicted position and predicted speed at each moment within a preset time period, it specifically performs the following steps: acquiring a preset prediction covariance, random variables, and a preset static safety distance; determining a dynamic safety distance based on the predicted speed at each moment within the preset time period; acquiring the position of at least one obstacle; determining the danger zone corresponding to the predicted position at each moment based on the obstacle positions, the preset static safety distance, and the dynamic safety distance; predicting the collision probability of the predicted position being in the danger zone based on the predicted position, random variables, and danger zone; selecting the maximum collision probability from the collision probabilities corresponding to each predicted position; and considering that a Level 1 warning is required if the maximum collision probability is not less than a preset first collision threshold.

[0093] For example, the dynamic safety distance is determined based on the predicted velocity at a given time, as shown in the following formula:

[0094]

[0095] in, for Dynamic safety distance at all times; for Predicting speed at any given moment; for The modulus of the predicted velocity at any given time; A preset positive real coefficient is used to adjust the size of the dynamic safety distance.

[0096] For example, the location of at least one obstacle is obtained; based on the locations of each obstacle, a preset static safety distance, and a dynamic safety distance, the danger zone corresponding to the predicted location at a given time is determined, as shown in the following formula:

[0097]

[0098] in, for Danger zone at all times; for The distance between the predicted position at any given time and one of the at least one obstacle position. for Predicted location at any given time.

[0099] For example, based on the predicted location, random variables, and the danger zone, the probability of a collision where the predicted location is in the danger zone can be predicted using the following formula:

[0100]

[0101] in, For a moment The probability of collision at that time; Let be the covariance matrix. Let t be the predicted position at time t. Let t be the predicted location point at time t. by For the mean, The variance is a Gaussian distribution.

[0102] Specifically, the detection device 140, when used to predict the warning time for executing a Level 1 warning based on the predicted position and predicted speed at each moment within a preset time period, is specifically used for: obtaining the maximum deceleration of the winch; determining the object distance between the position of each obstacle and the predicted position at each moment within the preset time period; determining the minimum braking distance based on the predicted speed and maximum deceleration at each moment; and determining that if there is no object distance greater than the minimum braking distance among at least one object distance at each moment, then the moment is determined to be the warning time for executing a Level 1 warning.

[0103] For example, the minimum braking distance is determined based on the predicted velocity and maximum deceleration at a given time, as shown in the following formula:

[0104]

[0105] in, for Minimum braking distance at any given time; for Predicting speed at any given moment; For maximum deceleration, Less than 0.

[0106] Furthermore, it should be noted that the minimum braking distance refers to the shortest distance required to reduce the predicted speed to 0 at that moment, according to the kinematic formula:

[0107]

[0108] It can be seen that when for Predicting speed at any given moment; When the maximum deceleration is reached, let This will allow you to obtain the minimum braking distance required.

[0109] For example, if at least one object distance at a given time does not contain an object distance greater than the minimum braking distance, then the time is determined as the warning time for executing a Level 1 warning, as shown in the following formula:

[0110]

[0111] in, for The predicted position at any given time is the distance between the object at one of the at least one obstacle position and the object at that obstacle position. for The minimum object distance among at least one object distance at time t. express A level-one warning signal is issued at all times; express We will not issue a Level 1 warning signal at any time.

[0112] Specifically, when the detection device 140 is used to predict whether to issue a secondary warning based on the predicted position and predicted speed at each moment within a preset time period, it is specifically used to: if the maximum collision probability is less than a preset first collision threshold and not less than a preset second collision threshold, then it is considered that a secondary warning is required.

[0113] For example, if the maximum collision probability is not less than a preset first collision threshold, it is considered that a first-level warning is required; if the maximum collision probability is less than the preset first collision threshold but not less than a preset second collision threshold, it is considered that a second-level warning is required; if the maximum collision probability is less than the preset second collision threshold, it is considered that no warning is required. See the following formula for details:

[0114]

[0115] in, for . This indicates a Level 1 warning; This indicates a Level II warning. The maximum collision probability; This is the second collision threshold; This is the first collision threshold.

[0116] In some examples, the system also includes the Xingxuan Smart Safety Helmet: if a secondary warning is determined to be required, the detection equipment is also used to generate a voice warning and send the voice warning to the Xingxuan Smart Safety Helmet; the Xingxuan Smart Safety Helmet is used to receive the voice warning sent by the detection equipment and broadcast the voice warning to remind the worker wearing the Xingxuan Smart Safety Helmet that there is a risk of collision between the target curtain wall unit panel and the obstacle.

[0117] For example, such as Figure 2 When a level-two warning is required, the corresponding voice warning can be sent to the Xingxuan smart safety helmet worn by the worker for voice warning broadcast. Specifically, the voice warning can be "Beware of collision".

[0118] In some examples, the detection device is also used to: detect whether the speed in the motion state sent by the IMU sensor exceeds a preset speed; if so, generate a first-level braking signal and send the first-level braking signal to the control device to control the winch to stop working.

[0119] In some examples, the system also includes: a display device for displaying an overall view of the moving target curtain wall unit panel, including a three-dimensional view of the building where the target curtain wall unit panel is installed, the vertical transport cableway and ring track for the winch to transport the target curtain wall unit panel, and the real-time position of the target curtain wall unit panel; and a display device for displaying a key focus view of the moving target curtain wall unit panel, including a view of the first floor where the target curtain wall unit panel is placed.

[0120] Furthermore, such as Figure 3 As shown, the display device provides a global view and automatic switching function. The left screen of the device can display an overall view including the vertical transport cableway and the circular track, while the right screen can display the first floor for easy observation of the lifting point. Furthermore, it uses an IMU (Inertial Measurement Unit) to sense whether the lifting equipment is stationary or in motion. When a change in the lifting equipment's status is detected, it automatically pushes the corresponding local view to the right screen or highlights it in the center of the screen, based on the part that requires the most attention (e.g., the spreader or hook). In addition, it monitors the speed of the lifting equipment in real time. When the speed exceeds a set threshold, the winch stops immediately, and an alarm message is displayed in the center of the screen. (For example, if the winch speed exceeds the preset speed by 2 m / s, a winch stop warning is displayed, along with a display showing that the spreader speed is greater than 2 m / s).

[0121] Specifically, operators can quickly select the target floor for the hoisting operation (such as "floor 25") by touching the screen, using buttons, or speaking. Once selected, the system will highlight or flash the floor in the 3D visualization model or floor plan as a visual guide.

[0122] Therefore, by integrating an intelligent display and interactive system, the visualization level and operational safety of curtain wall hoisting operations are significantly improved. The display equipment adopts a split-screen layout, with the left side showing the overall situation of the vertical transport cableway and the ring track, and the right side monitoring the first-floor hoisting area in real time. Combined with the IMU's precise perception of the dynamic and static status of hoisting equipment (such as the spreader pole), the system can automatically push key local images to the main display area or highlight them when the status changes, ensuring that the operator always focuses on the most important work links and effectively reducing the risk of human error. At the same time, the system monitors the operating speed of the winch and spreader pole in real time. Once it exceeds the preset threshold (such as 2m / s), it immediately triggers an emergency stop command and pops up an alarm message (such as "Spreader pole speed exceeds limit, emergency braking has been initiated") in a prominent position in the center of the screen, realizing proactive safety protection. In addition, the operator can quickly set the target floor through the touch screen, buttons, or voice. The system will then highlight or flash the corresponding floor in the 3D model or floor plan, providing clear visual guidance and avoiding misoperation. Overall, the system realizes an integrated human-machine collaboration mechanism of "dynamic focusing, intelligent early warning, rapid setting, and linkage control", which greatly improves the situational awareness, response efficiency and operational safety of the hoisting process, and provides strong support for efficient, accurate and reliable operation in complex high-altitude construction scenarios.

[0123] In some examples, the system also includes: a path planning device for acquiring digital twin information and determining the optimal global path based on the digital twin information and the MARL algorithm, and sending the optimal global path to the detection device; the digital twin information includes all entity information of the building on which the target curtain wall unit panel is installed; the detection device for receiving the optimal global path sent by the path planning device and predicting whether there are positions on the optimal global path where the distance to the obstacle is less than the safe distance. If so, it acquires the current position of the target curtain wall unit panel and determines a self-healing path based on the current position and the target position, so that the control device guides the winch to move the target curtain wall unit panel based on the optimal global path and / or the self-healing path; the safe distance is the sum of the preset static safe distance and the dynamic safe distance.

[0124] MARL stands for Multi-Agent Reinforcement Learning. It is a machine learning method that enables multiple agents to learn and collaborate in a shared environment to achieve common or individual goals.

[0125] Specifically, digital twin information is a time-varying graph structure containing information about all entities, which can be represented as:

[0126]

[0127] in, It is a geometric model used to represent the collection of static structures such as buildings, ring tracks, and floors where target curtain wall unit panels are installed. This is a dynamic state set, representing the states of all intelligent agents (e.g., winches, hoisting equipment, IMU sensors, etc.). ,in For the first The state of an intelligent agent (such as a carrying pole or an electric hoist). This is a dynamic obstacle set, representing the state of all moving obstacles (workers, equipment). . Environmental rules include physical rules (such as kinematics) and safety rules (such as safe distances). )wait.

[0128] Specifically, when the path planning equipment is used to determine the optimal global path based on digital twin information and the MARL algorithm, it is used to: construct the state space and action space of each executing agent based on the digital twin information; obtain a preset multi-objective reward function, which comprehensively considers hoisting progress, safety, hoisting efficiency, and motion comfort, and is used to calculate the immediate reward obtained by the executing agent in the current state for performing a specific action, where the current state belongs to the state space and the specific action belongs to the action space; use the MARL algorithm to train each executing agent based on its current state, specific action, and immediate reward to solve for the optimal cooperative strategy; and generate the optimal global path from the starting position to the target position according to the optimal cooperative strategy.

[0129] Among them, digital twin information As a state space The set of control commands issued by all controllable intelligent agents in the digital twin information (such as the target speed sent to the winch). ) as action space The reward function is expressed as follows: Used to guide an agent to learn the optimal strategy.

[0130] For example, the reward function is carefully designed for multi-objective optimization, as shown in the following formula:

[0131]

[0132] in, For instant rewards; ; ; This represents the vector change from the current position to the target position. ; Let represent the state of the agent at time step t. and obstacles The distance between them; The preset penalty collision intensity; Safety rewards are used to measure safety, taking into account both penalties for collisions and approaching obstacles. To improve hoisting efficiency, we encourage rapid completion. Δt is the difference between the actual time taken to reach the current position and the ideal time. ; For improved comfort, it is used to penalize abrupt movements and ensure smooth motion. The degree of urgency; These are the weighting coefficients for each objective.

[0133] Specifically, the MARL algorithm is used to train each agent based on its current state, specific actions, and immediate rewards to solve for the optimal cooperative policy. Based on the optimal cooperative policy, an optimal global path from the starting position to the target position is generated. This includes: using the MARL algorithm, iterative training is performed based on the current state of each agent, the specific actions taken, and the immediate rewards obtained. By optimizing the long-term accumulated expected reward, the optimal cooperative policy is solved. After the policy converges, policy deduction is performed starting from the starting position to gradually generate a state-action sequence, ultimately forming an optimal global path from the starting position to the target position.

[0134] Furthermore, the collaborative strategy corresponding to the largest long-term cumulative expected reward is taken as the optimal collaborative strategy.

[0135] For example, the maximum long-term cumulative expected reward = ,in, Discount factor; For immediate rewards. The optimal cooperative strategy is represented as: .

[0136] For example, a position on the optimal global path where the distance to an obstacle is less than the safe distance can be expressed by the following formula:

[0137]

[0138] in, This indicates that it exists. obstacle; A dynamic obstacle set; To meet the conditions; This refers to a position on the optimal global path relative to an obstacle. For safe distance.

[0139] Specifically, when the detection equipment is used to determine a self-healing path based on the current location and the target location, it is used to: take the current location as the starting node for dynamic replanning and construct digital twin information containing multiple candidate nodes; for each candidate node, use a heuristic search algorithm to calculate its comprehensive cost value, which includes the cumulative actual cost from the starting node to the candidate node, the heuristic predicted cost from the candidate node to the target location, and the obstacle avoidance cost from the candidate node to the obstacle; based on each comprehensive cost value, traverse the digital twin information to generate the optimal alternative path from the starting node to the target location; and use the optimal alternative path as the self-healing path.

[0140] In a digital twin environment, a self-healing path refers to replanning a path around sudden obstacles and back to the original target location, starting from the current position. The self-healing path is determined based on a self-healing algorithm that combines discrete graph search and continuous motion dynamics.

[0141] For example, the comprehensive cost value calculated using a heuristic search algorithm can be expressed as the following formula:

[0142] ,

[0143]

[0144] in, Overall value To accumulate actual costs; for ; The cost of avoiding obstacles. It can be the Riemann distance from the current position to the target position. The distance; This is the minimum collision distance.

[0145] For example, based on the comprehensive value of each digital twin, traversing the digital twin information to generate the optimal alternative path from the starting node to the target location can be expressed as the following formula:

[0146]

[0147] in, After establishing a self-healing pathway, This represents the current position, and goal represents the target position. This involves using digital twin information. By traversing the digital twin information, the path with the minimum total cost from the current location to the target location is found.

[0148] In addition, after the detection equipment obtains the self-healing path, if the self-healing path can be implemented, it will send the self-healing path to the control equipment; if it cannot be implemented, it will notify the human to intervene.

[0149] Therefore, this technical solution achieves virtual-real linkage and dynamic mapping of the entire curtain wall hoisting process by constructing a high-fidelity digital twin model synchronized with the physical hoisting system. This model not only synchronizes the real-time operating status of key entities such as hoisting equipment, spreader poles, and winches, but also integrates real-time spatial information of all obstacles, construction equipment, and personnel in the work environment, constructing a dynamically updated and precisely calculable virtual work scenario. Based on this, a multi-agent reinforcement learning (MARL) algorithm is introduced to model key execution units such as spreader poles and winches as collaborative agents. Under the unified scheduling of the cloud-based "digital twin brain," globally optimal path planning and intelligent collaborative control are achieved. When encountering sudden obstacles (such as mobile devices or personnel intrusion), the system no longer relies on simple emergency braking, but instead performs real-time risk assessment and path replanning based on the digital twin environment, automatically triggering a "self-healing" mechanism to dynamically adjust the hoisting path and the actions of each execution mechanism, achieving an organic unity of obstacle avoidance and task continuity. This technology significantly improves the intelligence level of hoisting operations under complex working conditions, enhances the system's adaptability, safety, and operational efficiency, and achieves a leap from "passive response" to "proactive decision-making and adaptive execution".

[0150] In summary, the intelligent monitoring and early warning system for curtain wall unit panel hoisting provided in this solution includes: hoisting equipment for hoisting the target curtain wall unit panel; a winch for receiving control commands from the control equipment to move the hoisting equipment and thus the target curtain wall unit panel; an IMU sensor installed on the hoisting equipment to acquire the motion status of the target curtain wall unit panel in real time and send the motion status to the detection equipment; a Xingxuan smart safety helmet worn by construction site workers to monitor their own position and movement status in real time and send the movement status to the detection equipment; the detection equipment receives the motion status from the IMU sensor and the movement status from the Xingxuan smart safety helmet, generates a corrected system state vector based on the motion status, the movement status of the workers, and the estimated system state vector, and predicts the predicted position and speed of the target curtain wall unit panel at each moment within a preset time period based on the corrected system state vector; the detection equipment also predicts the warning time for executing a level one warning based on the predicted position and speed at each moment within the preset time period and sends the warning time to the control equipment to control the winch to stop working.

[0151] By integrating hoisting equipment, winches, IMU sensors, Xingxuan smart safety helmets, and detection equipment, a closed-loop intelligent hoisting control system was constructed, achieving high-precision state perception, dynamic prediction, and proactive safety control of the curtain wall unit panel hoisting process. IMU sensors collect the motion state of the target curtain wall unit in real time. The detection equipment combines measured data with estimated system state for fusion correction, generating a high-precision corrected system state vector, and predicting the position and velocity change trends over future time periods based on this vector. Simultaneously, the system integrates personnel position information from Xingxuan smart safety helmets, constructing a comprehensive state representation encompassing multiple elements of "mechanism-structure-person," significantly enhancing situational awareness in complex construction environments. The detection equipment can identify potential risk trajectories in advance, and the system can predict the timing of a Level 1 warning and promptly send instructions to the control equipment, triggering the winch to stop. This solution effectively improves the real-time monitoring capability and safety response speed of the hoisting process, realizing a shift from passive protection to proactive early warning, significantly reducing the collision risk caused by swaying, deviation, malfunction, or misoperation, and ensuring the safety, stability, and construction accuracy of high-altitude operations.

[0152] Specifically, by constructing a closed-loop process encompassing "perception-early warning-decision-planning-control-self-healing," the intelligence level and overall performance of curtain wall hoisting operations have been significantly improved. Compared to traditional emergency modes that rely on manual intervention or simple sudden stops, the system no longer passively interrupts operations when encountering unexpected obstacles. Instead, it generates the optimal self-healing path in real time based on a digital twin environment and multi-agent reinforcement learning (MARL) algorithms, enabling dynamic detours and automatic resumption of operation. This significantly reduces the number of task interruptions and enhances the resilience and continuity of the construction process. Through global collaborative optimization, conflicts and efficiency bottlenecks caused by local decisions of single equipment are avoided, achieving efficient coordination among multiple execution units such as winches and spreader poles, and shortening the overall hoisting cycle. Ultimately, the system not only ensures the inherent safety of high-altitude operations but also propels the hoisting process towards higher efficiency, smoother operation, and full automation, providing a highly reliable technical solution for intelligent construction in complex building scenarios.

[0153] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.

Claims

1. A smart monitoring and early warning system for the hoisting of curtain wall unit panels, characterized in that, include: Lifting equipment, used for lifting target curtain wall unit panels; A winch is used to receive control commands from the control equipment to move the hoisting equipment based on the control commands, thereby moving the target curtain wall unit panel; An IMU sensor, installed at the hoisting equipment, is used to acquire the motion status of the target curtain wall unit panel in real time and send the motion status to the detection equipment; The Xingxuan Smart Safety Helmet is worn by construction site workers to monitor their movement status in real time, including their own location, and sends the movement status to the detection device. The detection device is used to receive the motion state sent by the IMU sensor, receive the personnel motion state sent by the Xingxuan smart safety helmet, generate a corrected system state vector based on the motion state, the personnel motion state and the estimated system state vector, and predict the predicted position and predicted speed of the target curtain wall unit panel at each moment within a preset time period based on the corrected system state vector. The detection device is also used to predict the warning time for executing the first-level warning based on the predicted position and the predicted speed at each time within the preset time period, and send the warning time to the control device to control the winch to stop working; Specifically, when the detection device is used to predict the warning time for executing a Level 1 warning based on the predicted position and the predicted speed at each moment within the preset time period, it is used for: Based on the predicted position and predicted speed at each moment within the preset time period, it is determined whether to issue a Level 1 warning. If so, the warning time for issuing a Level 1 warning is determined based on the predicted position and predicted speed at each moment within the preset time period. If not, then based on the predicted position and predicted speed at each moment within the preset time period, it is determined whether to issue a secondary warning; if not, no warning is issued. The detection device, when used to predict whether to issue a Level 1 warning based on the predicted position and the predicted speed at each moment within the preset time period, is specifically used for: Obtain the preset prediction covariance, random variables, and preset static safety distance; For each moment within a preset time period, a dynamic safety distance is determined based on the predicted speed at that moment; Obtain the location of at least one obstacle; The danger zone corresponding to the predicted location at the given time is determined based on the location of each obstacle, the preset static safety distance, and the dynamic safety distance. Based on the predicted location, the random variable, and the danger zone, predict the probability of a collision where the predicted location is in the danger zone; Select the highest collision probability from the collision probabilities corresponding to each predicted location; If the maximum collision probability is not less than the preset first collision threshold, it is considered that a level one warning is required.

2. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 1, the system further includes: The display device is used to display the overall picture of the moving target curtain wall unit panel. The overall picture includes a three-dimensional view of the building where the target curtain wall unit panel is installed, the vertical transport cableway and ring track for the winch to transport the target curtain wall unit panel, and the real-time position of the target curtain wall unit panel. The display device is also used to display a key focus screen of the moving target curtain wall unit panel, the key focus screen including the screen corresponding to the first floor where the target curtain wall unit panel is placed.

3. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 1, characterized in that, The system also includes: If it is determined that a level 2 warning is required, the detection device is also used to generate a voice warning and send the voice warning to the Xingxuan Smart Safety Helmet. The Xingxuan smart safety helmet is also used to receive the voice warning sent by the detection device and broadcast the voice warning to remind the worker wearing the Xingxuan smart safety helmet that there is a risk of collision between the target curtain wall unit panel and the obstacle.

4. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 1, characterized in that, The detection device, when used to predict the warning time for executing a level-one warning based on the predicted position and the predicted speed at each moment within the preset time period, is specifically used for: Obtain the maximum deceleration of the winch; For each moment within a preset time period, determine the distance between the position of each obstacle and the predicted position at that moment; The minimum braking distance is determined based on the predicted speed at the time and the maximum deceleration. If at least one object distance at the specified time does not exceed the minimum braking distance, then the specified time is determined to be the warning time for executing the Level 1 warning.

5. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 1, characterized in that, The detection device, when used to predict whether to issue a secondary warning based on the predicted position and the predicted speed at each moment within the preset time period, is specifically used for: If the maximum collision probability is less than the preset first collision threshold and not less than the preset second collision threshold, it is considered that a level two warning is required.

6. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 1, characterized in that, The system also includes: A path planning device is used to acquire digital twin information, determine the optimal global path based on the digital twin information and the MARL algorithm, and send the optimal global path to a detection device; the digital twin information includes all entity information of the building on which the target curtain wall unit panel is installed; The detection device is used to receive the optimal global path sent by the path planning device, and predict whether there is a position on the optimal global path where the distance to the obstacle is less than the safe distance. If so, it obtains the current position of the target curtain wall unit panel, and determines a self-healing path based on the current position and the target position, so that the control device guides the winch to move the target curtain wall unit panel based on the optimal global path and / or the self-healing path; the safe distance is the sum of a preset static safe distance and a dynamic safe distance.

7. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 6, characterized in that, The path planning device, when used to determine the optimal global path based on the digital twin information and the MARL algorithm, is specifically used for: Based on the digital twin information, the state space and action space of each executing agent are constructed; Obtain a preset multi-objective reward function, which comprehensively considers hoisting progress, safety, hoisting efficiency and motion comfort, and is used to calculate the immediate reward obtained by the agent performing a specific action in the current state, wherein the current state is the state space, and the specific action belongs to the action space; The optimal cooperative strategy is solved by training based on the current state, specific action and immediate reward of each agent. Based on the optimal cooperation strategy, an optimal global path from the starting position to the target position is generated.

8. The intelligent monitoring and early warning system for curtain wall unit panel hoisting according to claim 6, characterized in that, The detection device, when used to determine the self-healing path based on the current position and the target position, is specifically used for: Using the current position as the starting node for dynamic replanning, a digital twin information containing multiple candidate nodes is constructed; For each candidate node, a heuristic search algorithm is used to calculate its comprehensive cost value, which includes the cumulative actual cost from the starting node to the candidate node, the heuristically predicted cost from the candidate node to the target location, and the obstacle avoidance cost from the candidate node to the obstacle. Based on the comprehensive cost value of each item, the digital twin information is traversed to generate the optimal alternative path from the starting node to the target location; The optimal alternative path is taken as the self-healing path.

Images