Online real-time monitoring system and method for interior decoration frame

Abstract

The invention provides an on-line real-time monitoring system and method for an interior decoration frame, and the method comprises the steps: data collection: collecting the displacement, inclination angle and speed data of a movable door-type frame at a set frequency through a sensor; data preprocessing: abnormal values in the collected data are removed, and normalization processing is carried out; a neural network model is constructed and trained, and an MLP neural network is constructed and trained; performing safety state prediction and early warning, and inputting real-time data into the trained model to predict a safety state; and inputting the real-time data and the predicted safety state into a pre-designed BIM model for visual display, and analyzing and evaluating the safety risk of the movable portal frame in the current construction state by using spatial information and real-time monitoring data in the BIM model. The physical parameter information of the interior decoration frame is detected through the sensor, safety prediction is performed based on the neural network, visual display and safety analysis and evaluation are performed based on the BIM, and the safety of a construction site is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building construction monitoring, and particularly to an in - interior decoration scaffold online real - time monitoring system and method. Background Technique

[0002] With the application of high - formwork safety monitoring technology in the construction industry, the safety performance of the overall formwork support system has been further guaranteed. However, during the interior decoration stage, due to the uneven quality of the multi - layer mobile portal frames used indoors, and at the same time, during the use process, the online safety monitoring technology has not been realized. Once the support frame becomes unstable or collapses, it is extremely likely to cause production safety accidents, resulting in losses to the lives of operators and machinery.

[0003] With the further strengthening of the requirements for safety management in the construction industry and the requirements for refined management in the industry, how to improve the real - time safety management of mobile frames used indoors and realize the online monitoring modular technology is of extremely important significance for safety management and the realization of economic benefits.

[0004] Currently, for the existing high - formwork online monitoring platform of the same type of technology, during the concrete pouring process, it can monitor the dynamic changes of the erected formwork in real - time, and then transmit the data to the alarm system to achieve alarm and early warning. However, no system has been retrieved that studies the real - time monitoring and realizes alarm triggering during the use of multi - layer mobile portal frames used indoors during the building decoration stage. Therefore, the present in - interior decoration scaffold (mobile portal frame) online real - time monitoring system clarifies the above problems. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an in - interior decoration scaffold online real - time monitoring system and method to solve the deficiencies of the prior art described in the background technique.

[0006] To solve the above - mentioned technical problem, an embodiment of the present invention provides the following technical solution: An in - interior decoration scaffold online real - time monitoring system, comprising: a mobile portal frame vertical rod, a data monitoring module, a data control module, a 4G wireless transmission module, an audible and visual alarm device, and a server. The data monitoring module is fixed on the mobile portal frame vertical rod. The data monitoring module is electrically connected to the data control module through a first data transmission line. The data control module is electrically connected to the 4G wireless transmission module through a second data transmission line. The 4G wireless transmission module is wirelessly communicatively connected to the audible and visual alarm device and the server;

[0007] The data monitoring module includes a wire - type displacement sensor, a 4G wireless inclination sensor, and a three - axis displacement and velocity sensor. The data control module includes a controller and a display. The audible and visual alarm device includes a wireless receiving module and an audible and visual alarm circuit.

[0008] The present invention also provides an on-line real-time monitoring method for interior decoration scaffolds, comprising the following steps:

[0009] Data acquisition: Displacement, inclination and speed data of the mobile gantry scaffold are acquired through sensors at a set frequency.

[0010] Data preprocessing: Outliers in the acquired data are removed and normalization processing is performed.

[0011] Neural network model construction and training: An MLP neural network is constructed and trained, including forward propagation, calculation of the loss function, backpropagation to update parameters, and evaluation of the model using a validation set.

[0012] Safety status prediction and early warning: The real-time data is input into the trained model to predict the safety status.

[0013] The real-time data and the predicted safety status are input into a pre-designed BIM model for visual display, and using the spatial information and real-time monitoring data in the BIM model, the safety risks of the mobile gantry scaffold in the current construction state are analyzed and evaluated, and the construction progress is adjusted, the material stacking is optimized, and the personnel activity area is planned according to the analysis and evaluation results.

[0014] Further, the removing outliers in the acquired data and performing normalization processing are specifically as follows:

[0015] Removing outliers: Setting the upper and lower limit thresholds of the data, setting the upper limit as U and the lower limit as L. If the data d satisfies d > U or d < L, it is regarded as an outlier and removed.

[0016] Normalization processing is respectively performed on the displacement, inclination and speed data, and the data is mapped to the interval [0, 1]. The formula is where d n is the normalized data and d is the original data.

[0017] Further, for the neural network model construction and training, constructing an MLP neural network is specifically as follows:

[0018] Constructing a neural network: Selecting a multi-layer perceptron MLP neural network, constructing the network architecture of the input layer, hidden layer and output layer; among them, there are seven nodes in the input layer, corresponding respectively to the displacement amounts x, y, z measured by the displacement sensors, the inclination θ measured by the inclination sensor, and the speeds ν x 、ν y 、ν z ; there are 10 nodes in the hidden layer, adopting the ReLU activation function; there is 1 node in the output layer, and the output value is in [0, 1];

[0019] Initializing the weights ω ij and the bias parameter b j, where i represents the serial number of the previous layer node and j represents the serial number of the next layer node, and the learning rate and the number of training iterations are set.

[0020] Furthermore, the training includes forward propagation, calculating the loss function, backpropagation to update parameters, and evaluating the model using the validation set. Specifically:

[0021] Collect a large number of historical monitoring data and the corresponding safety status labels as the training sample set, and divide it into a training set and a validation set according to the ratio of 80% and 20%.

[0022] Input the preprocessed data x in the training set i into the neural network, and calculate the predicted output through forward propagation The forward propagation calculation process is as follows: for the hidden layer, for the output layer ω j is the weight from the hidden layer to the output layer, and b is the bias of the output layer;

[0023] Calculate the loss function value according to the predicted output and the actual marked safety status. The loss function uses the mean squared error loss function where m is the number of samples, y i is the actual safety status value, is the predicted safety status value;

[0024] Use the backpropagation algorithm to calculate the gradient according to the loss function value and Update the weight and bias parameters of the neural network, specifically to gradually reduce the loss function value;

[0025] During the training process, every k times of training rounds, use the validation set to evaluate the model, observe the loss function value and accuracy metrics on the validation set. When the metrics no longer improve or reach the preset number of training iterations, stop training and save the trained neural network model; where k = 100.

[0026] Furthermore, when the predicted safety status value predicted by the model is greater than the set relative safety threshold, it is determined that the scaffolding is in a dangerous state and an alarm signal is output.

[0027] Furthermore, it also includes model update and optimization, collecting new data to add to the sample set, regularly retraining the model, and adjusting the neural network structure and hyperparameters to optimize the performance.

[0028] Furthermore, the set learning rate is 0.001 and the number of training iterations is 1000.

[0029] Furthermore, the analysis and evaluation of the safety risk of the mobile gantry in the current construction state are specifically as follows:

[0030] Construct risk assessment function R = f(D, E, L) = ω d ×D+ω e ×E+ω l ×L, where D is the frame deformation parameter, E is the surrounding environment impact parameter and L is the construction load, ω d 、ω e 、ω l are the corresponding weighting coefficients respectively; among which:

[0031] (1) The calculation formula of the frame deformation parameter D is:

[0032] D=D d +D a +D v , D d Displacement deformation influence factor, D a is the inclination deformation influencing factor, D v is the velocity deformation influencing factor;

[0033] k d is the displacement weight coefficient;

[0034] D a =θ×k a ,k a is the displacement weight coefficient;

[0035] k v is the speed weight coefficient;

[0036] (2) The calculation formula of the surrounding environment impact parameter E is:

[0037] E=E m +E a , E m is the material influence factor, E a Vibration impact factor

[0038] where d m Indicates the distance between the material and the rack, m is the weight of the material, d m0 is the safety distance threshold, m0 is the reference material weight, k m is the material impact weight coefficient;

[0039] E a =a×k a , a is the vibration acceleration, k a is the vibration influence coefficient

[0040] (3) The calculation formula for the construction load L is L = (n × ω + M) × k L, where n is the number of personnel in the construction area, ω is the average weight per person, M is the weight of construction tools and materials, and k L is the construction load weight coefficient.

[0041] Furthermore, inputting the real-time data and predicted safety status into a pre-designed BIM model for visual display specifically means: in the BIM model, the deformation degree of different parts of the framework is displayed in different colors or dynamic effects, with green indicating normal, yellow indicating warning, and red indicating danger.

[0042] The beneficial effects of the above technical solutions of the present invention are as follows:

[0043] 1. By setting a data monitoring module including a wire-pulling displacement sensor, a 4G wireless inclination sensor, and a three-axis displacement velocity sensor on the vertical pole of the mobile portal frame, it provides rich data support for comprehensively and accurately grasping the state of the framework, effectively avoiding the omission of potential safety hazards caused by untimely monitoring or incomplete data.

[0044] 2. The constructed MLP neural network model is trained with a large amount of historical monitoring data, and can learn the complex non-linear relationship between the framework deformation data and the safety status, so as to accurately predict the safety status of the framework. When the predicted safety status value is greater than the set relative safety threshold, an alarm signal is output in a timely manner, and an audible and visual alarm is issued through an audible and visual alarm device. At the same time, detailed warning information and phone calls are sent to the mobile phones of operators and managers to ensure that relevant personnel can take measures quickly and nip the accident risk in the bud, greatly improving the safety of the construction site.

[0045] 3. Inputting the real-time monitoring data and predicted safety status into a pre-designed BIM model for visual display enables on-site managers and construction workers to clearly understand the real-time safety status of the framework at a glance, and can quickly judge the safety of the framework without complex data interpretation, greatly improving the intuitiveness and convenience of safety management. Using the spatial information and real-time monitoring data in the BIM model, combined with a pre-set risk assessment function, comprehensively considering various factors such as framework deformation parameters, surrounding environment impact parameters, and construction loads, a comprehensive and scientific analysis and assessment of the safety risk of the mobile portal frame in the current construction state is carried out. It can accurately determine high-risk areas and potential potential safety hazard points, providing a basis for taking targeted measures, helping to optimize resource allocation and management decisions during the construction process, improving construction efficiency, and reducing construction delays and cost increases caused by safety problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a schematic structural diagram of the in-fitting frame online real-time monitoring system of the present invention;

[0047] Figure 2Flow chart of the on-line real-time monitoring method for the interior decoration frame of the present invention. Specific embodiments

[0048] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0049] As Figure 1 shown, the present invention provides an on-line real-time monitoring system for an interior decoration frame, including: a vertical rod 1 of a mobile gantry, a data monitoring module 2, a data control module 3, a 4G wireless transmission module 4, an audible and visual alarm device 7, and a server 8. The data monitoring module 2 is fixed on the vertical rod 1 of the mobile gantry. The data monitoring module 2 is electrically connected to the data control module 3 through a first data transmission line 5. The data control module 3 is electrically connected to the 4G wireless transmission module 4 through a second data transmission line 6. The 4G wireless transmission module 7 is wirelessly communicatively connected to the audible and visual alarm device 7 and the server 8;

[0050] The data monitoring module 2 includes a wire-pulling displacement sensor 21, a 4G wireless inclination sensor 21, and a three-axis displacement velocity sensor 21. The data control module 3 includes a controller 31 and a display 32. The audible and visual alarm device 7 includes a wireless receiving module 71 and an audible and visual alarm circuit 72.

[0051] The data monitoring module 2 transmits the sensing and monitoring data to the data control module 3 through the first data transmission line 5. The data control module 3 can set the data switch boundary value. The received monitoring data can be displayed on the data control module 3 in real time. If the monitoring data exceeds the set boundary limit, the switch signal will be transmitted to the 4G wireless transmission module 4 through the second data transmission line 6. The 4G wireless transmission module 4 wirelessly transmits the signal to the audible and visual alarm device 7 and the server 8 in real time. The transmission distance and obstacles are not limited. If the monitoring data exceeds the limit value abnormally, the audible and visual alarm device 7 will emit an audible and visual alarm signal in real time, and at the same time send information to the mobile phones of operators and managers and call the abnormal alarm phone, continuously providing alarm and early warning information to inform the dangerous state information. The monitoring and early warning data are wirelessly transmitted to the server 8 in real time for safety analysis. The BIM system is installed on the server 8, which can visually display the monitoring data and perform analysis and evaluation.

[0052] As Figure 2 shown, the present invention also provides an on-line real-time monitoring method for an interior decoration frame, including the following steps:

[0053] S1. Data acquisition: Displacement, inclination, and velocity data of the mobile gantry are acquired through sensors at a set frequency. Through various sensors in the data monitoring module, data such as displacement, inclination, and velocity of the mobile gantry during use are continuously acquired at a set acquisition frequency (e.g., once every t seconds, where t = 0.1). Let the displacement measured by the displacement sensor be x, y, z (in three-dimensional directions), the inclination measured by the inclination sensor be θ, and the velocity measured by the velocity sensor be ν x , ν y , ν z .

[0054] S2. Data preprocessing: Remove outliers from the acquired data and perform normalization processing;

[0055] S3. Neural network model construction and training: Construct an MLP neural network and perform training, including forward propagation, calculating the loss function, backpropagation to update parameters, and evaluating the model using a validation set;

[0056] S4. Safety status prediction and early warning: Input real-time data into the trained model to predict the safety status;

[0057] S5. Input the real-time data and predicted safety status into a pre-designed BIM model for visual display, and use the spatial information and real-time monitoring data in the BIM model to analyze and evaluate the safety risks of the mobile gantry under the current construction conditions. Adjust the construction progress, optimize the material stacking, and plan the personnel activity areas according to the analysis and evaluation results.

[0058] In step S2, the removal of outliers from the acquired data and the normalization processing are specifically as follows:

[0059] Remove outliers: Set the upper and lower limit thresholds for the data. Let the upper limit be U and the lower limit be L. If the data d satisfies d > U or d < L, it is regarded as an outlier and removed;

[0060] Perform normalization processing on the displacement, inclination, and velocity data respectively, and map the data to the interval [0, 1]. The formula is where d n is the normalized data and d is the original data.

[0061] In step S3, the construction and training of the neural network model, constructing an MLP neural network, are specifically as follows:

[0062] Construct a neural network: Select a multi-layer perceptron (MLP) neural network and construct the network architecture of the input layer, hidden layer, and output layer. Among them, there are seven nodes in the input layer, corresponding to the displacement x, y, z measured by the displacement sensor, the inclination θ measured by the inclination sensor, and the velocity ν x , ν y, ν z ; The hidden layer has 10 nodes, using the ReLU activation function; the output layer has 1 node, and the output value is in [0, 1];

[0063] Initialize the weights ω of the neural network ij and the bias parameter b j , where i represents the serial number of the previous layer node, j represents the serial number of the next layer node, set the learning rate and the number of training iterations. The specifically set learning rate is 0.001, and the number of training iterations is 1000.

[0064] In step S3, the training includes forward propagation, calculating the loss function, backpropagation to update the parameters, and evaluating the model using the validation set. Specifically:

[0065] Collect a large number of historical monitoring data and the corresponding safety status labels as the training sample set, and divide them into a training set and a validation set according to the ratio of 80% and 20%;

[0066] Input the preprocessed data x in the training set i into the neural network, and calculate the predicted output through forward propagation The forward propagation calculation process is: for the hidden layer, for the output layer ω j is the weight from the hidden layer to the output layer, and b is the output layer bias;

[0067] Calculate the loss function value according to the predicted output and the actual labeled safety status. The loss function uses the mean squared error loss function where m is the number of samples, y i is the actual safety status value, is the predicted safety status value;

[0068] Use the backpropagation algorithm to calculate the gradient according to the loss function value and Update the weights and bias parameters of the neural network. Specifically to gradually reduce the loss function value;

[0069] During the training process, every k = 100 training rounds, use the validation set to evaluate the model, observe the loss function value and accuracy metrics on the validation set. When the metrics no longer improve or reach the preset number of training iterations, stop training and save the trained neural network model.

[0070] In step S4, when the predicted safety status value of the model is greater than the set relative safety threshold, it is determined that the frame is in a dangerous state, and an alarm signal is output. The relative safety threshold is 0.6. At the same time, detailed early warning information (including the predicted degree of danger, current frame deformation data, etc.) and a phone call are sent to the mobile phones of operators and managers through the 4G wireless transmission module to ensure that relevant personnel can take timely measures, such as stopping work, inspecting the frame, evacuating personnel, etc., to prevent the occurrence of safety accidents.

[0071] In addition, it also includes model update and optimization. New data is collected and added to the sample set, and the model is retrained regularly. The neural network structure and hyperparameters are adjusted to optimize the performance. As the usage time increases and the working environment changes, new monitoring data is continuously collected and marked with the safety status and then added to the sample set. The neural network model is retrained regularly (such as weekly or monthly) using the new sample set to update the weight and bias parameters of the model, so that the model can adapt to the performance changes of the frame and new working conditions, and continuously improve the accuracy and reliability of safety prediction. At the same time, the structure and hyperparameters of the neural network are optimized and adjusted according to the actual application situation. For example, try to increase the number of hidden layers or adjust the number of nodes, observe the changes in model performance (such as comparing indicators such as the accuracy of the validation set and the loss function value under different structures), and select the optimal model configuration to improve the overall performance of the safety prediction technical solution.

[0072] In step S5, the real-time data and the predicted safety status are input into a pre-designed BIM model for visual display, and the safety risks of the mobile gantry in the current construction state are analyzed and evaluated by using the spatial information and real-time monitoring data in the BIM model. Specifically, it includes:

[0073] Develop a special data interface program to ensure that the real-time data collected by the monitoring system (including the normalized data such as displacement, inclination angle, speed, etc. after preprocessing) can be accurately transmitted to the BIM model. This interface should be able to identify different types of data and associate them with the corresponding components in the BIM model (such as the vertical poles, horizontal bars, diagonal braces, etc. of the mobile gantry).

[0074] In the BIM model, set corresponding parameter attributes for each component to be monitored, and these attributes correspond to the received monitoring data. For example, set a displacement parameter for the vertical pole and an inclination angle parameter for the horizontal bar, etc. When the monitoring data is transmitted, the parameter values of these components are updated in real time through the data interface program.

[0075] According to the changes in component parameters, different visualization effects are set in the BIM model to intuitively display the deformation of the frame. The degree of deformation of different parts of the frame is displayed with different colors or dynamic effects. Green indicates normal, yellow indicates warning, and red indicates danger. For example, when the displacement of a certain pole is within the normal range, the pole is displayed in green; when the displacement approaches the warning value, its color is gradually changed to yellow; when the displacement exceeds the relative safety threshold, it is displayed in red, and a flashing effect can be set to highlight the dangerous parts. At the same time, for parameters such as inclination and speed, similar color gradients or dynamic effects (such as arrow length and direction to indicate speed magnitude and direction) can also be used to display their changes, so that on-site managers and construction personnel can understand the real-time safety status of the frame at a glance.

[0076] Furthermore, the BIM model's spatial information and real-time monitoring data were used to comprehensively analyze and assess the safety risks of the mobile gantry under its current construction status. By combining factors such as the building structure, the location of finishing materials, and the areas where construction personnel were active, a pre-defined risk assessment algorithm was used to identify high-risk areas and potential safety hazards.

[0077] According to the results of the safety risk assessment, the progress of interior decoration construction will be reasonably adjusted. If the safety risk of the mobile gantry in a certain area is high, the decoration work in that area will be suspended, and construction tasks in other areas with lower safety risks will be prioritized to avoid accidents caused by continuing construction under dangerous conditions. At the same time, the construction sequence will be re-planned to ensure that the force on the mobile gantry during subsequent construction is more reasonable, reducing the risk of deformation and instability of the frame caused by improper construction sequence. For example, for processes that may have a large load impact on the frame (such as lifting large materials, installing heavy equipment, etc.), adjust them to a time period when the frame deformation is small or after reinforcement.

[0078] The BIM model was used to examine the relative relationship between material stacking locations and the mobile gantry. Based on monitoring data showing areas of significant gantry deformation, material stacking was optimized. Heavier materials were moved away from areas of significant gantry deformation or to locations with greater load-bearing capacity. This prevented instability caused by increased localized loads on the gantry due to improper material stacking. Furthermore, the height and range of material stacking were rationally planned to ensure that they did not affect construction workers' sightlines and operating space, nor obstruct evacuation routes in emergencies.

[0079] Based on the BIM model and monitoring data, clearly demarcate the safe personnel activity areas and dangerous areas. Set obvious warning signs and isolation measures in the dangerous areas to prevent construction workers from straying into the dangerous areas. At the same time, dynamically adjust the personnel activity routes according to the deformation of the framework and the construction progress to ensure that construction workers can operate on the premise of safety and reduce the risk of injury to personnel that may be caused by the deformation of the framework. For example, set up a temporary protection passage on the side with a large deformation of the framework to guide personnel to pass from the safe side.

[0080] Utilize the visualization and simulation functions of the BIM model to formulate a detailed emergency plan for possible emergencies such as the instability or collapse of mobile portal frames. Simulate the personnel evacuation routes, rescue equipment access routes, etc. in different accident scenarios in the BIM model to evaluate the feasibility and effectiveness of the emergency plan. Optimize the emergency plan according to the simulation results to ensure that in the event of an actual accident, personnel evacuation and rescue work can be carried out quickly and orderly, minimizing casualties and property losses. For example, determine the location of the nearest safety exit, set up temporary shelters, etc., and mark and drill them in the BIM model.

[0081] According to the above content, construct a risk assessment function R = f(D, E, L) = ω d ×D + ω e ×E + ω l ×L, where D is the framework deformation parameter, E is the surrounding environment impact parameter, and L is the construction load, and ω d , ω e , ω l are the corresponding weighting coefficients respectively; where:

[0082] (1) The calculation formula for the framework deformation parameter D is:

[0083] D = D d + D a + D v , D d is the displacement deformation influence factor, D a is the inclination deformation influence factor, D v is the velocity deformation influence factor;

[0084] k d is the displacement weight coefficient;

[0085] D a = θ × k a , k a is the displacement weight coefficient;

[0086] k v is the velocity weight coefficient;

[0087] (2) The calculation formula for the surrounding environment impact parameter E is as follows:

[0088] E = E m + E a , where E m is the material impact factor, and E a is the vibration impact factor.

[0089] Among them, d m represents the distance between the material and the frame body, m is the material weight, d m0 is the safety distance threshold, m0 is the reference material weight, and k m is the material impact weight coefficient;

[0090] E a = a × k a , where a is the vibration acceleration and k a is the vibration impact coefficient.

[0091] (3) The calculation formula for the construction load L is L = (n × ω + M) × k L , where n is the number of personnel in the construction area, ω is the average weight per person, M is the weight of construction tools and materials, and k L is the construction load weight coefficient.

[0092] According to the collected data, after calculating D, E, and L, substitute them into the risk coefficient formula R = f(D, E, L) = ω d × D + ω e × E + ω l × L. According to the calculated risk coefficient R value, the safety risk of the mobile gantry can be evaluated and classified. For example:

[0093] When R is less than R1 (R1 is the low-risk threshold, such as 0.3), it is determined to be a low risk, the frame is in a safe state, and construction can proceed normally.

[0094] When R1 ≤ R < R2 (R2 is the medium-risk threshold, such as 0.6), it is determined to be a medium risk. It is necessary to increase the monitoring frequency, focus on relevant areas, and at the same time check whether there are any abnormalities in the construction process.

[0095] When R ≥ R2, it is determined to be a high risk. Immediately stop the construction in this area, take corresponding reinforcement measures or adjust the construction plan until the risk coefficient is reduced to the safe range.

[0096] In summary, the present invention can achieve the following technical effects:

[0097] 1. By setting up a data monitoring module including a wire-pulling displacement sensor, a 4G wireless inclination sensor, and a three-axis displacement and velocity sensor on the vertical pole of the mobile gantry, it is possible to collect the displacement, inclination, and velocity data of the frame in three dimensions in real time at a relatively high frequency (such as once every 0.1 seconds), providing rich data support for comprehensively and accurately grasping the frame state, and effectively avoiding the omission of potential safety hazards caused by untimely monitoring or incomplete data. Using 4G wireless communication technology, the rapid and stable transmission of data is realized, without being restricted by space and obstacles, ensuring that the monitoring data can be transmitted to the data control module, the audible and visual alarm device, and the server in real time, enabling on-site safety management personnel and relevant operators to obtain the real-time state information of the frame immediately and make timely responses.

[0098] 2. The constructed MLP neural network model is trained with a large amount of historical monitoring data, and can learn the complex non-linear relationship between the frame deformation data and the safety state, so as to accurately predict the safety state of the frame. When the predicted safety state value is greater than the set relative safety threshold (such as 0.6), an alarm signal is output in time, and an audible and visual alarm is issued through the audible and visual alarm device. At the same time, detailed warning information (including the predicted danger level, the current frame deformation data, etc.) and a phone call are sent to the mobile phones of operators and management personnel to ensure that relevant personnel can quickly take measures (such as stopping work, checking the frame, evacuating personnel, etc.), strangling the accident risk in the bud, and greatly improving the safety of the construction site. As time goes by and the working environment changes, by continuously collecting new data and adding them to the sample set and regularly retraining the model, as well as optimizing and adjusting the neural network structure and hyperparameters, the model can continuously adapt to the changes in the frame performance and new working conditions, and continuously improve the accuracy and reliability of safety prediction, providing a strong guarantee for long-term construction safety.

[0099] 3. The real-time monitoring data and the predicted safety state are input into the pre-designed BIM model for visual display. The deformation degree of different parts of the frame is displayed in an intuitive color (green indicates normal, yellow indicates warning, and red indicates danger) or dynamic effect (such as an arrow indicating the direction and magnitude of the velocity), enabling on-site management personnel and construction personnel to clearly understand the real-time safety state of the frame at a glance, and quickly judge the safety of the frame without complex data interpretation, greatly improving the intuitiveness and convenience of safety management. Using the spatial information and real-time monitoring data in the BIM model, combined with the pre-set risk assessment function, comprehensively considering various factors such as frame deformation parameters, surrounding environment impact parameters, and construction loads, a comprehensive and scientific analysis and assessment of the safety risk of the mobile gantry in the current construction state is carried out. It can accurately determine high-risk areas and potential safety hazard points, providing a basis for taking targeted measures, helping to optimize resource allocation and management decisions during the construction process, improving construction efficiency, and reducing construction delays and cost increases caused by safety problems.

[0100] First, according to the safety risk assessment results in the BIM model, reasonably adjust the indoor decoration construction progress. Suspend operations in areas with higher safety risks, prioritize the construction of low-risk areas, avoid continuing construction in dangerous situations to cause accidents, ensure the orderly progress of the construction process, reduce the work stoppage losses and economic compensation caused by accidents, and safeguard the overall progress and economic benefits of the decoration project. Secondly, optimize the material stacking position through the BIM model, transfer heavier materials to appropriate areas, avoid increasing the local load of the formwork and causing instability, and reasonably plan the stacking height and scope of materials. This not only ensures the safety of the formwork but also improves the space utilization rate of the construction site, reduces additional work such as secondary handling caused by unreasonable material stacking, reduces construction costs, and improves construction efficiency. Thirdly, clearly divide the safe and dangerous areas based on the BIM model, set warning signs and isolation measures, and dynamically adjust the personnel movement routes to ensure that construction personnel work under the premise of safety, reduce the risk of harm to personnel caused by formwork deformation, reduce potential work injury compensation costs, and also helps to improve the work efficiency of construction personnel and ensure the smooth progress of the construction schedule. Finally, utilize the visualization and simulation functions of the BIM model to formulate a detailed emergency plan for possible emergencies such as formwork instability or collapse, simulate the evacuation routes of personnel and the entry routes of rescue equipment in different accident scenarios in the model, and evaluate the feasibility and effectiveness of the plan. Optimize the emergency plan through simulation to ensure that in the event of an actual accident, personnel evacuation and rescue work can be carried out quickly and orderly, minimize casualties and property losses, and reduce the negative impact of the accident on the project and the enterprise.

[0101] The above is the preferred implementation mode of the present invention. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle described in the present invention, several improvements and retouches can still be made, and these improvements and retouches should also be regarded as the protection scope of the present invention.

Claims

1. An in - interior decoration rack online real - time monitoring system, characterized in that, Including: A vertical pole of a mobile gantry, a data monitoring module, a data control module, a 4G wireless transmission module, an audible and visual alarm device, and a server. The data monitoring module is fixed on the vertical pole of the mobile gantry. The data monitoring module is electrically connected to the data control module through a first data transmission line. The data control module is electrically connected to the 4G wireless transmission module through a second data transmission line. The 4G wireless transmission module is wirelessly communicatively connected to the audible and visual alarm device and the server; The data monitoring module includes a pull-wire displacement sensor, a 4G wireless inclination sensor, and a three-axis displacement and speed sensor. The data control module includes a controller and a display. The audible and visual alarm device includes a wireless receiving module and an audible and visual alarm circuit.

2. An on-line real-time monitoring method for an interior decoration rack, characterized in that, Including the following steps: Data acquisition, collecting displacement, inclination, and speed data of the mobile gantry through sensors at a set frequency; Data preprocessing, removing outliers from the collected data and performing normalization processing; Neural network model construction and training, constructing an MLP neural network, and training, including forward propagation, calculating the loss function, and backpropagating to update parameters, and using a validation set to evaluate the model; Safety status prediction and early warning, inputting real-time data into the trained model to predict the safety status; Inputting the real-time data and the predicted safety status into a pre-designed BIM model for visual display, and using the spatial information and real-time monitoring data in the BIM model to analyze and evaluate the safety risks of the mobile gantry under the current construction state, and adjusting the construction progress, optimizing the material stacking, and planning the personnel activity area according to the analysis and evaluation results.

3. The on-line real-time monitoring method for the interior decoration rack according to claim 2, wherein, The removing outliers from the collected data and performing normalization processing is specifically as follows: Removing outliers, setting the upper and lower limit thresholds of the data, setting the upper limit as U and the lower limit as L. If the data d satisfies d > U or d < L, it is regarded as an outlier and removed; Normalize the displacement, inclination, and velocity data separately, and map the data to the interval [0, 1]. The formula is where d n is the normalized data and d is the original data.

4. The on-line real-time monitoring method for the interior decoration rack according to claim 2, wherein, The constructing an MLP neural network in the neural network model construction and training is specifically as follows: Build a neural network, select the multi-layer perceptron (MLP) neural network, and construct the network architecture of the input layer, hidden layer, and output layer; among them, there are seven nodes in the input layer, corresponding to the displacement amounts x, y, z measured by the displacement sensors, the inclination angle θ measured by the inclination sensor, and the speed ν measured by the speed sensor x , ν y , ν z ; there are 10 nodes in the hidden layer, using the ReLU activation function; there is 1 node in the output layer, and the output value is in [0, 1]; Initialize the weights ω of the neural network ij and the bias parameter b j , where i represents the serial number of the previous layer node, j represents the serial number of the next layer node, and set the learning rate and the number of training iterations.

5. The online real-time monitoring method for the interior decoration frame according to claim 4, wherein The training, including forward propagation, calculating the loss function, and backpropagating to update parameters, and using a validation set to evaluate the model is specifically as follows: Collecting a large amount of historical monitoring data and corresponding safety status labels as a training sample set, and dividing them into a training set and a validation set according to the ratio of 80% and 20%; Input the preprocessed data x in the training set i into the neural network and calculate the predicted output through forward propagation The forward propagation calculation process is as follows: for the hidden layer, for the output layer ω j is the weight from the hidden layer to the output layer, and b is the output layer bias; Calculate the loss function value according to the predicted output and the actual marked safety state, and the mean squared error loss function is used as the loss function where m is the number of samples, y i is the actual safety state value, is the predicted safety state value; Using the backpropagation algorithm, calculate the gradient based on the loss function value and Update the weight and bias parameters of the neural network, specifically gradually reduce the loss function value; During the training process, every k times of training rounds, using the validation set to evaluate the model, observing the loss function value and accuracy rate indicators on the validation set. When the indicators no longer improve or reach the preset training iteration times, stop training and save the trained neural network model; where k = 100.

6. The on-line real-time monitoring method for the interior decoration rack according to claim 2, characterized in that, When the predicted safety status value predicted by the model is greater than the set relative safety threshold, it is determined that the frame is in a dangerous state and an alarm signal is output.

7. The online real-time monitoring method of the interior decoration rack according to claim 2, characterized in that It also includes model update and optimization, collecting new data to add to the sample set, regularly retraining the model, and adjusting the neural network structure and hyperparameters to optimize the performance.

8. The online real-time monitoring method for the interior decoration frame according to claim 4, characterized in that, The set learning rate is 0.001, and the training iteration times is 1000.

9. The online real-time monitoring method for the interior decoration rack according to claim 2, wherein The analyzing and evaluating the safety risks of the mobile gantry under the current construction state is specifically as follows: Construct a risk assessment function \(R = f(D, E, L)=\omega\) d \(\times D+\omega\) e \(\times E+\omega\) l \(\times L\), where \(D\) is the formwork deformation parameter, \(E\) is the surrounding environment impact parameter, and \(L\) is the construction load, and \(\omega\) d \(,\omega\) e \(,\omega\) l are the corresponding weighting coefficients respectively; Wherein: (1) The calculation formula for the frame deformation parameter D is: D = D d + D a + D v ,D d Displacement deformation influence factor, D a is the dip angle deformation influence factor, D v is the velocity deformation influence factor; k d is the displacement weight coefficient; D a = θ × k a , k a is the displacement weight coefficient; k v is the speed weight coefficient; (2) The calculation formula for the surrounding environment influence parameter E is: E = E m + E a , E m is the material influence factor, E a is the vibration influence factor Among them, d m represents the distance between the material and the rack body, m is the material weight, d m0 is the safety distance threshold, m0 is the reference material weight, k m is the material influence weight coefficient; E a = a × k a , where a is the vibration acceleration and k a is the vibration influence coefficient (3) The calculation formula for the construction load L is L = (n×ω + M)×k L , where n is the number of personnel in the construction area, ω is the average weight per person, M is the weight of construction tools and materials, and k L is the weight coefficient of the construction load.

10. The on-line real-time monitoring method for the interior decoration rack according to claim 2, characterized in that Visualizing the real-time data and predicted safety status by inputting them into a pre-designed BIM model specifically includes: showing the deformation degree of different parts of the formwork in different colors or dynamic effects in the BIM model, where green indicates normal, yellow indicates early warning, and red indicates danger.

Images