Mobile formwork intelligent monitoring method and system based on digital twinborn and multi-mode perception

Abstract

The invention provides a mobile formwork intelligent monitoring method and system based on digital twinning and multi-modal perception, and belongs to the technical field of building construction intelligence, and the method comprises the steps: building a multi-modal sensor network on a mobile formwork entity; establishing a digital twin platform: constructing a BIM model mapped with a mobile formwork entity, pre-burying a virtual monitoring point in the BIM model, performing matching calibration with a sensor of the multi-modal sensor network, and fusing multi-source sensor data acquired by the multi-modal sensor network in real time; calculating a construction health index (SHI) in real time based on the multi-source sensor data; on the basis of the multi-source sensor data, the instability probability is predicted in real time through an AE-LSTM mixed model; meanwhile, predicting a weight factor of a construction health index SHI; and according to the SHI and the instability probability, jointly evaluating whether a mobile formwork adjustment instruction is generated or not and sending out an early warning. The safety state of the movable formwork is controlled in real time, and the problem that the instability trend cannot be monitored in real time is solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of intelligent construction technology, and particularly relates to a mobile formwork intelligent monitoring method and system based on digital twinning and multi-modal perception, which is combined with a building information model (BIM), an Internet of Things (IoT) and artificial intelligence (AI). BACKGROUND

[0002] The mobile formwork cast-in-place box girder technology is an advanced bridge construction technology, and its core is to cast a concrete box girder structure on a bridge site through a movable formwork system. This technology combines the advantages of mechanized construction and cast-in-place technology and is widely used in modern bridge construction. Although the construction of the mobile formwork cast-in-place box girder is a relatively mature technology, there are still many problems in the construction, such as: 1. The mobile formwork structure is risky, the stress and deformation are complex, and the instability trend cannot be monitored in real time; 2. The formwork position adjustment under complex conditions (such as small curve radius hole passing) depends on manual experience, and the error accumulation leads to linear deviation; 3. The concrete vibration compaction degree and formwork displacement lack real-time monitoring, and quality defects occur frequently; the mobile formwork moves for a long time, and the stress state of the formwork cannot be controlled in real time, which easily causes problems such as honeycomb and pitted surface; 4. The data of the design, construction and supervision parties are fragmented, there is no unified risk judgment standard, the decision-making is lagging behind, and it is difficult to realize the cooperation of all parties. SUMMARY

[0003] The purpose of the present application is to provide a mobile formwork intelligent monitoring method and system based on digital twinning and multi-modal perception, which is combined with a building information model (BIM), an Internet of Things (IoT) and artificial intelligence (AI), realizes real-time collection of various data of the mobile formwork cast-in-place box girder, and controls the safety state of the mobile formwork in real time through the collected data, thereby enhancing the construction safety of the mobile formwork, speeding up the construction efficiency of the cast-in-place box girder and improving the construction quality of the cast-in-place box girder.

[0004] In order to achieve the above purpose, the technical scheme of the present application is as follows: A mobile formwork intelligent monitoring method based on digital twinning and multi-modal perception, comprising: S1, establishing a multi-modal sensor network on a mobile formwork entity; S2, establishing a digital twinning platform: constructing a BIM model mapped with the mobile formwork entity, pre-burying virtual monitoring points in the BIM model, matching and calibrating the sensors of the multi-modal sensor network, and real-time fusing multi-source sensor data collected by the multi-modal sensor network; S3, based on the multi-source sensor data, calculating a construction health index (SHI) in real time; S4, based on the multi-source sensor data, the instability probability is predicted in real time by an AE-LSTM hybrid model; meanwhile, the weight factors of the parameters of the construction health index SHI are predicted for the next calculation of the construction health index SHI; S5, according to the construction health index SHI and the predicted instability probability, it is jointly evaluated whether to generate a mobile formwork adjustment instruction and issue a warning.

[0005] Further, in step S1, the establishment of the multi-modal sensor network comprises: S101, optical fiber grating sensors are arranged in the main beam span of the mobile formwork and the fulcrum to monitor the stress of the main beam; S102, inclination sensors are arranged at the four corners of the top plate of the mobile formwork to monitor the lateral inclination of the formwork; S103, an acoustic sensor array is arranged inside the formwork of the mobile formwork to monitor the compaction degree of the concrete; S104, a laser displacement meter is arranged at the bottom of the support leg of the mobile formwork to monitor the settlement of the support leg; S105, a temperature and humidity anemometer is arranged at the top of the formwork of the mobile formwork to monitor the environmental parameters; S106, an industrial camera is arranged at the key joint of the mobile formwork to monitor the formwork joint misalignment.

[0006] Further, in step S2, the method for pre-burying virtual monitoring points in the BIM model comprises: S201, the correlation between the GIS coordinates of each sensor in the multi-modal sensor network and the positions of the mobile formwork components in the BIM model is established to realize the spatial binding of the monitoring data and the BIM model; S202, monitoring point families for different parameters are created to mark and distinguish different sensor types and annotate key information of the sensors; S203, the sensor data stream is associated with the mobile formwork components in the BIM model in real time by using the MQTT protocol.

[0007] Further, the calculation method of the construction health index SHI in step S3 comprises: ; Wherein, SHI represents the construction health index; represents the stress safety factor, represents the weight factor of the stress safety factor; represents the deformation deviation, represents the weight factor of the deformation deviation; represents the environmental risk value, represents the weight factor of the environmental risk value; 、 、 Real-time calculation according to multi-source sensor data; Real-time update through AE-LSTM hybrid model prediction.

[0008] Further, the step S4 of predicting the instability probability in real time through the AE-LSTM hybrid model comprises: S401, fixing the LSTM module, training the AE module through historical multi-source sensor data, so that the AE module effectively extracts key features of the multi-source sensor data; S402, loading the AE module, accessing the LSTM module, training through historical data containing instability events, and outputting instability probability, the instability events being instability values corresponding to future k time steps of historical multi-source sensor data; S403, predicting the instability probability through the AE module and the LSTM module based on real-time received multi-source sensor data.

[0009] Further, the training of the AE-LSTM hybrid model further comprises: Taking historical multi-source sensor data as input data, corresponding to the feature engineering of SHI parameter input data; in the historical multi-source sensor data, the stress value time series change rate of the data corresponding to the SHI stress safety factor parameter is calculated; the spatial gradient distribution is formed according to the deformation data of the data corresponding to the SHI deformation deviation parameter; the coupling oscillation frequency of the environmental parameters is calculated according to the data corresponding to the SHI environmental risk value parameter; the stress value time series change rate, spatial gradient distribution and coupling oscillation frequency of environmental parameters are taken as feature vectors for training; The AE-LSTM hybrid model introduces an attention layer for dynamically adjusting the weights of each feature vector; The output is the weight factor of each parameter of the construction health index SHI.

[0010] Further, the step S5 comprises: Setting instability probability threshold and SHI threshold, when the instability probability predicted by the AE-LSTM hybrid model is greater than the instability probability threshold, and the SHI calculated according to the weight factor predicted by the AE-LSTM hybrid model is greater than the SHI threshold, a mobile formwork adjustment instruction is generated and a warning is issued.

[0011] Further, generating a mobile formwork adjustment instruction and issuing a warning comprises: Marking the risk area on the digital twin platform, formulating a hydraulic correction scheme for the mobile formwork actuator, and pushing an emergency plan warning.

[0012] Further, the AE-LSTM hybrid model makes a prediction every 30 seconds, the SHI calculation is performed every 1 second, and the weight factor predicted by the current AE-LSTM hybrid model is used for each SHI calculation until the next AE-LSTM hybrid model prediction weight factor is updated.

[0013] Another aspect of the present application also provides a mobile formwork intelligent monitoring system based on digital twinning and multi-modal perception, comprising: A multi-modal sensor network is established on the mobile formwork entity. A digital twinning platform is constructed, which maps a BIM model to the mobile formwork entity, and pre-buries virtual monitoring points in the BIM model, matches and calibrates the sensors of the multi-modal sensor network, and fuses multi-source sensor data collected by the multi-modal sensor network in real time. An SHI calculation module calculates the construction health index SHI in real time based on the multi-source sensor data. An AE-LSTM hybrid model prediction module predicts the instability probability in real time based on the multi-source sensor data through an AE-LSTM hybrid model, and simultaneously predicts the weight factor of each parameter of the construction health index SHI for the next calculation of the construction health index SHI. An early warning evaluation module jointly evaluates whether to generate a mobile formwork adjustment instruction and issue a warning according to the construction health index SHI and the predicted instability probability.

[0014] Further, the multi-modal sensor network comprises: Fiber Bragg grating sensors are arranged in the main beam span and support points of the mobile formwork to monitor the stress of the main beam. Inclination sensors are arranged at the four corners of the top plate of the mobile formwork to monitor the lateral inclination of the formwork. An acoustic sensor array is arranged inside the formwork of the mobile formwork to monitor the concrete vibration density. A laser displacement meter is arranged at the bottom of the support leg of the mobile formwork to monitor the support leg settlement. A temperature and humidity anemometer is arranged at the top of the mobile formwork to monitor the environmental parameters. Industrial cameras are arranged at the key joints of the mobile formwork to monitor the formwork joint misalignment.

[0015] Further, the digital twinning platform comprises: A space binding unit establishes the association between the GIS coordinates of each sensor in the multi-modal sensor network and the location of the mobile formwork component in the BIM model, and realizes the spatial binding of the monitoring data and the BIM model. A monitoring marking unit creates a monitoring point family for different parameters, marks different sensor types, and annotates key sensor information. Real-time association unit: real-time association of sensor data stream and BIM model mobile formwork component by using MQTT protocol.

[0016] Further, the SHI calculation module comprises: ; Wherein, SHI represents construction health index; represents stress safety factor, represents the weight factor of stress safety factor; represents deformation deviation, represents the weight factor of deformation deviation; represents environmental risk value, represents the weight factor of environmental risk value; 、 、 According to multi-source sensor data real-time calculation; Real-time update by AE-LSTM hybrid model prediction.

[0017] Further, the AE-LSTM hybrid model prediction module comprises: The first training unit: fix the LSTM module, train the AE module by historical multi-source sensor data, so that the AE module can effectively extract the key features of multi-source sensor data; The second training unit: load the AE module, access the LSTM module, and train by historical data containing instability events, with the output being instability probability, and the instability event being the instability value corresponding to the future k time steps of historical multi-source sensor data; The prediction unit predicts the instability probability by the AE module and the LSTM module based on the real-time received multi-source sensor data.

[0018] Still further, the AE-LSTM hybrid model prediction module further comprises: The historical multi-source sensor data is used as input data, and the feature engineering of the input data corresponding to each parameter of the construction health index SHI is performed; in the historical multi-source sensor data, the time series change rate of the stress value is calculated for the data corresponding to the stress safety factor parameter of the SHI; the spatial gradient distribution is formed according to the deformation data for the data corresponding to the deformation deviation parameter of the SHI; the coupling oscillation frequency of the environmental parameters is calculated for the data corresponding to the environmental risk value parameter of the SHI; the time series change rate of the stress value, the spatial gradient distribution, and the coupling oscillation frequency of the environmental parameters are used as feature vectors for training; The AE-LSTM hybrid model introduces an attention layer for dynamically adjusting the weight of each feature vector; The output is the weight factor of each parameter of the construction health index SHI.

[0019] Further, the early warning evaluation module comprises: The instability probability threshold and the SHI threshold are set, when the instability probability predicted by the AE-LSTM hybrid model is greater than the instability probability threshold, and the SHI calculated according to the weight factor predicted by the AE-LSTM hybrid model is greater than the SHI threshold, a mobile support frame adjustment instruction is generated and an early warning is issued.

[0020] Further, the early warning evaluation module comprises: The digital twin platform calibrates the risk area, formulates the hydraulic correction scheme of the mobile support frame actuator, and pushes the emergency plan warning.

[0021] Further, the early warning evaluation module further comprises that the AE-LSTM hybrid model predicts every 30 seconds, the SHI is calculated every 1 second, and the weight factor predicted by the current AE-LSTM hybrid model is used for each SHI calculation until the weight factor is updated after the next AE-LSTM hybrid model prediction.

[0022] Compared with the prior art, the present application has the following beneficial effects: 1、The present application realizes the synchronous real-time monitoring of multiple key parameters of the mobile support frame cast-in-place box girder construction by setting a multi-modal sensor network, and provides necessary data support for digital twinning and AI intelligent decision-making; 2、The present application realizes intelligent management of the whole life cycle of construction by establishing a digital twin platform to dynamically associate the BIM model with the measured data in real time, and provides data integration for further AI decision-making; 3、The present application proposes a specific design scheme of the construction health index SHI and its calculation method, unifies the risk judgment standard, integrates all information into a comprehensive indicator, and dynamically adjusts the comprehensive indicator result through its own weight factor, so that the comprehensive indicator result is more consistent with the actual working condition, and solves the problems of decision lag and coordination efficiency of all parties; 4、The present application proposes an AE-LSTM hybrid model as a prediction model of instability probability, AE realizes data denoising and information fusion through dimensionality reduction and feature extraction of multi-source sensor raw data, and provides key low-dimensional features for the LSTM model to analyze the time sequence rule of the mobile support frame construction process; the LSTM model predicts the instability probability, and solves the problems of inability to monitor the instability trend in real time and frequent quality defects; 5、The present application proposes a scheme for jointly evaluating whether to trigger a mobile support frame adjustment instruction according to the construction health index SHI and the predicted instability probability, and forms a bidirectional driving intelligent decision-making system through bidirectional enhancement of SHI and the prediction model, realizes early and accurate warning of structural instability, reduces the false alarm rate, and improves the response speed of risk emergency. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart of embodiment 1 of the present application.

[0024] Figure 2 is a schematic diagram of the intelligent monitoring architecture of embodiment 1 of the present application.

[0025] Figure 3 is a schematic diagram of the association and cooperation of SHI and prediction model of embodiment 1 of the present application. DETAILED DESCRIPTION

[0026] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0027] The main design idea of the present application is to realize intelligent all-around monitoring of mobile formwork cast-in-place box girder in complex working conditions (such as small radius curve hole) or extreme environment (such as typhoon environment) construction, based on multi-modal perception fusion AI decision, to solve the problems of quality defect prevention and real-time safety warning.

[0028] The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0029] Embodiment 1: This embodiment proposes a mobile formwork intelligent monitoring method based on digital twinning and multi-modal perception, as shown in the steps of Figure 1 , an architecture as shown in Figure 2 is established, which specifically includes: S1, a multi-modal sensor network is established on the mobile formwork entity.

[0030] In order to realize all-around monitoring of mobile formwork cast-in-place box girder, a multi-modal sensor network needs to be established in the mobile formwork.

[0031] The types and related parameters of the multi-modal sensor network established in this embodiment are shown in the following table:

[0032] Based on the above multi-modal sensor network, multi-source sensor data of the mobile formwork can be collected, which can help to improve construction accuracy and real-time monitoring, enhance safety and risk warning, and promote digital transformation of bridge construction, and help to realize intelligent construction.

[0033] S2, a digital twinning platform is established: a BIM model mapped with the mobile formwork entity is constructed, virtual monitoring points are pre-buried in the BIM model, matched and calibrated with the sensors of the multi-modal sensor network, and real-time fusion of multi-source sensor data collected by the multi-modal sensor network.

[0034] The purpose of this step is to establish a digital twin platform, specifically including: 1. Construct a mobile formwork BIM model, which is mapped 1:1 with the physical entity of the mobile formwork, and establish a parameterized mobile formwork BIM family library, including small-radius curve adaptive components.

[0035] 2. Pre-bury virtual monitoring points in the BIM model.

[0036] 2.1. Establish the association between the GIS coordinates of each sensor in the multi-modal sensor network and the position of the mobile formwork component in the BIM model, and realize the spatial binding of monitoring data and the BIM model; Establish a coordinate system conversion matrix, construct a 4x4 homogeneous transformation matrix (including a rotation matrix R and a translation vector T), and realize the mathematical conversion of the sensor local coordinate system to the BIM global coordinate system; the least squares method can be used to optimize the calibration parameters to solve the coordinate offset problem caused by sensor installation deviation, ensuring that the coordinate transfer error is ≤±2mm; Kalman filter algorithm is used to process the coordinate drift caused by construction vibration, and the transformation parameters are updated in real time.

[0037] 2.2. Create monitoring point families for different parameters, mark different sensor types, and label key sensor information; Assign a unique code (such as an RFID tag) to the sensor and establish a bidirectional index relationship with the BIM component ID; extend the SensorEntity entity class in the IFC standard to represent sensor devices, and add dynamic attribute fields such as coordinate offset and installation angle; a graph database (Neo4j) can also be used to store the sensor network topology, realizing the visualization of the data link between "sensor - collection terminal - digital twin platform".

[0038] 2.3. Use the MQTT protocol to associate sensor data streams with BIM model mobile formwork components in real time; Based on the MQTT protocol, transmit sensor data packets, including timestamp, coordinate system identifier, and measurement value triplets; Delaunay triangulation can be used for data completion on sparse monitoring points to improve the mapping coverage of special-shaped structures; then Shader programming in Unity3D / Unreal engine can be used to realize dynamic coloring of BIM models driven by monitoring data.

[0039] S3. Based on the multi-source sensor data, calculate the construction health index SHI in real time.

[0040] The construction health index SHI essentially quantifies the health status of the mobile formwork cast-in-place box girder construction with a number. In traditional engineering, safety officers need to check stress tables, deformation data, and weather forecasts, which may easily overlook critical information. SHI integrates all information into a comprehensive indicator.

[0041] Specifically, the construction health index SHI is established to solve three key problems: The first is the problem of non-uniform risk judgment criteria. For example, in the same extreme environment such as typhoon weather, the same structural deformation may have completely different risk levels. The construction health index SHI can be dynamically adjusted through its own weight factor, making the comprehensive indication result more in line with the actual working condition.

[0042] The second is the problem of decision-making lag. When the on-site personnel find problems such as stress exceeding the standard and report layer by layer, the best disposal time may have been missed. However, the SHI is calculated in real time, and the instability probability predicted by the prediction model can provide early warning for a period of time (for example, 30 minutes), which is enough to start the emergency plan.

[0043] The third is the problem of coordination efficiency. The supervision unit focuses on the safety factor, the construction unit is concerned about the progress, and the design unit values the linear control. SHI uses a unified indicator to make all parties reach a consensus on the risk level.

[0044] The calculation method of the construction health index SHI includes: ; Wherein, SHI represents the construction health index; represents the stress safety factor, represents the weight factor of the stress safety factor; represents the deformation deviation, represents the weight factor of the deformation deviation; represents the environmental risk value, represents the weight factor of the environmental risk value; 、 、 calculated in real time according to multi-source sensor data; updated in real time through the AE-LSTM hybrid model prediction.

[0045] Wherein, 、 、 In this embodiment, the following method can be used for calculation: ; represents the real-time maximum stress; represents the allowable stress value, which can be a fixed threshold or dynamically adjusted according to historical fatigue data through the prediction model (AE-LSTM hybrid model) described below; ; represents the deformation amount; represents the allowable deformation value; ; T represents the concrete temperature; Indicates the mold temperature; Indicates the critical temperature rise value, which is a nonlinear parameter; when the temperature gradient inside the concrete exceeds 15℃ The number of items will grow exponentially and can accurately capture the risk of cracking.

[0046] The proposed Construction Health Index (SHI) and its calculation method transform the previously vague risk perception into a quantifiable index, enabling unified and precise risk control. However, this embodiment does not rely solely on the SHI for risk perception in mobile formwork cast-in-place box girder construction. Instead, the SHI is linked and coordinated with the instability prediction model described below to form a bidirectionally driven AI intelligent decision-making system.

[0047] S4. Based on the multi-source sensor data, the AE-LSTM hybrid model is used to predict the instability probability in real time; at the same time, the weight factors of each parameter of the construction health index SHI are predicted for the next calculation of the construction health index SHI.

[0048] The instability prediction model in this embodiment uses an AE-LSTM hybrid model. The AE-LSTM hybrid model uses a cascaded structure of an autoencoder (AE) and a long short-term memory (LSTM) network. Its core functions are: feature extraction from multi-source sensor data, temporal dependency modeling, and instability trend prediction. The specific architecture is as follows: Autoencoder (AE) module: Contains an encoder and a decoder. The encoder consists of 2-3 fully connected layers, which inputs the raw data x (dimension d) from multiple source sensors and outputs a low-dimensional feature vector h (dimension d′, d′ <d);解码器通过对称的全连接层将h重构得到的重构数据 , used for self-supervised learning of feature extraction. For example, the original data is image data (pixel matrix) of template seam misalignment, where d is the number of pixels. AE extracts the geometric features of the misalignment (such as width and position) as h, reducing the dimensionality while preserving the misalignment information.

[0049] LSTM module: Input is the low-dimensional feature sequence output by the encoder , The low-dimensional feature vector h at time t is used to capture the temporal dependency through a two-layer LSTM layer, and finally the fully connected layer outputs the predicted probability of instability in the next k time steps.

[0050] The input data of the AE-LSTM hybrid model is the multi-source sensor data collected by the multimodal sensor network in real time monitoring, specifically including: main beam stress data monitored by fiber grating sensors, formwork lateral inclination data monitored by inclination sensors, concrete vibration density data monitored by acoustic wave sensor arrays, support leg settlement data monitored by laser displacement meters, environmental parameter data monitored by temperature, humidity and anemometers, and template joint misalignment data monitored by industrial cameras.

[0051] The AE-LSTM hybrid model, serving as an instability prediction model, is linked and coordinated with the Construction Health Index (SHI). Therefore, input data undergoes feature engineering based on the parameters involved in the SHI calculation, converting them into corresponding eigenvectors. Feature engineering transforms fragmented input data into physically meaningful eigenvectors, which can be used to predict the SHI weight factor and improve the prediction accuracy of the AE-LSTM hybrid model. For the SHI stress safety factor, the time-series rate of change of stress values ​​(such as main beam stress data) in multi-source sensor data is calculated. For the SHI deformation deviation, a spatial gradient distribution is generated based on deformation data (such as lateral inclination and outrigger settlement data) from multi-source sensor data. For the SHI environmental risk value, the coupled oscillation frequency of environmental parameters (wind speed, temperature, and humidity) in the multi-source sensor data is calculated. This coupled oscillation frequency is obtained by calculating the resonant frequencies of wind speed, temperature, and humidity and is a directly correlated quantity of the eigenvalue.

[0052] The AE-LSTM hybrid model also introduces an attention layer to dynamically adjust the weights of each transformed feature vector.

[0053] The AE-LSTM hybrid model outputs multiple targets through different fully connected layers. First, the output is the probability of instability of the mobile formwork in the next k time steps, with a value range of [0,1] (0 indicates stability, 1 indicates instability), which is used to provide early warning of structural instability trends. Second, the output is the weighting factors of each parameter of the construction health index (SHI). .

[0054] The AE-LSTM hybrid model uses a joint loss function that combines the reconstruction loss of the AE module and the prediction loss of the LSTM module. The formula is as follows: ; Where L represents the loss and MSE is the mean square error; is a weight coefficient used to balance the importance of feature extraction and prediction tasks. In this embodiment, the value can be 0.3; x is the original data, is the reconstructed data; y is the true instability value, is the predicted probability of instability.

[0055] The training process of the AE-LSTM hybrid model is as follows: 1. Pre-training phase: First, fix the LSTM module and only train the AE module. Use historical stable state data (no instability event) to minimize the reconstruction loss Optimize the AE parameters to ensure that the encoder can effectively extract the key features of the multi-source sensor data.

[0056] 2. Joint training phase: Load the pre-trained AE module, access the LSTM module, and use historical data containing instability events (labeled future k time step instability value y) for end-to-end training. In this embodiment, the optimizer is Adam (learning rate 1e-4), the batch size is set to 32, the training rounds are 200, the validation set is verified every round, and the early stopping strategy (stop if there is no decrease for 10 consecutive rounds).

[0057] 3. Online update: After deployment, regularly collect new monitoring data to fine-tune the model parameters with a small learning rate (1e-5) to adapt to changes in mobile formwork working conditions.

[0058] The trained AE-LSTM hybrid model can predict instability probability and SHI weight factor by inputting real-time multi-source sensor data.

[0059] S5, according to the construction health index SHI and the predicted instability probability, jointly evaluate whether to generate mobile formwork adjustment instructions and issue a warning.

[0060] In this embodiment, the AE-LSTM hybrid model generates a prediction every 30 seconds, and the SHI is calculated every second. The two can be decoupled through an asynchronous message queue. Each calculation of SHI within 30 seconds after the AE-LSTM hybrid model predicts once uses the weight factor predicted by the AE-LSTM hybrid model this time, until the next AE-LSTM hybrid model prediction obtains a new weight factor after 30 seconds, and the weight factor is updated.

[0061] Set the instability probability threshold and the SHI threshold. When the instability probability predicted by the AE-LSTM hybrid model is greater than the instability probability threshold, and the SHI calculated according to the weight factor predicted by the AE-LSTM hybrid model is greater than the SHI threshold, generate mobile formwork adjustment instructions and issue a warning. In this embodiment, the instability probability threshold is set to 85%, and the SHI threshold is set to 0.8.

[0062] The generation of mobile formwork adjustment instructions and the issuance of warnings include three aspects of synchronous triggering: (1) Calibrate the risk area on the digital twin platform; (2) Based on real-time multi-source sensor data, formulate a hydraulic correction scheme for the mobile mold frame actuator as an emergency plan; (3) Push emergency plan alerts.

[0063] like Figure 3 As shown, in the method of this embodiment, the AE-LSTM hybrid model dynamically adjusts the weighting factors for SHI, increasing the weights of corresponding parameters under complex working conditions. SHI provides feature engineering support for the AE-LSTM hybrid model. Weakly correlated features that are difficult to detect in the data, such as short-term fluctuations in leg settlement rate, are quantified through the deformation deviation component of SHI and become high-value input features for the AE-LSTM hybrid model. When hydraulic correction is triggered, the execution effect data is fed back to the AE-LSTM hybrid model for online learning.

[0064] The method of this embodiment is specifically applied to the construction of cast-in-place box girders using mobile formwork, and the false alarm rate is reduced from 18% to 5%, and the response speed is increased by 36 times.

[0065] Under the condition of a sudden typhoon of level 14, it would normally cause an emergency shutdown for a period of time, but through the prediction of the AE-LSTM hybrid model, the environmental risk value of SHI The weight factor It was adjusted from 0.3 to 0.6, strengthening the environmental risk weight. The probability of instability and SHI both exceeded the corresponding thresholds, triggering the formulation of an emergency plan, automatically calculating the counterweight demand, and dynamically adjusting the counterweight without any shutdown.

[0066] The complex working conditions of small-radius curve through-holes usually rely on manual judgment and adjustment, with an error of ±10mm. After applying this method, the prediction of instability probability and SHI calculation results trigger the formulation of emergency plans, and the error of automatic correction is ±1.5mm.

[0067] Example 2: This embodiment proposes a mobile formwork intelligent monitoring system based on digital twins and multimodal perception, including: Multimodal sensor network: Establish a multimodal sensor network on the mobile mold frame entity; Digital twin platform: Build a BIM model that maps to the mobile formwork entity, embed virtual monitoring points in the BIM model, match and calibrate them with sensors in the multimodal sensor network, and integrate multi-source sensor data collected by the multimodal sensor network in real time; SHI calculation module: calculates the construction health index SHI in real time based on the multi-source sensor data; An AE-LSTM hybrid model prediction module: based on the multi-source sensor data, an AE-LSTM hybrid model is used to predict the instability probability in real time; meanwhile, the weight factor of each parameter of the construction health index SHI is predicted for the next calculation of the construction health index SHI. A warning evaluation module: according to the construction health index SHI and the predicted instability probability, it is jointly evaluated whether to generate a mobile formwork adjustment instruction and issue a warning.

[0068] The multi-modal sensor network includes: Fiber Bragg grating sensors are arranged in the middle of the main girder of the mobile formwork and the fulcrum to monitor the stress of the main girder; Inclination sensors are arranged at the four corners of the top plate of the mobile formwork to monitor the lateral inclination of the formwork; An array of acoustic wave sensors is arranged inside the formwork of the mobile formwork to monitor the compaction degree of the concrete; A laser displacement meter is arranged at the bottom of the support leg of the mobile formwork to monitor the settlement of the support leg; A temperature and humidity anemometer is arranged at the top of the formwork of the mobile formwork to monitor the environmental parameters; Industrial cameras are arranged at the key joints of the mobile formwork to monitor the template joint misalignment.

[0069] The digital twin platform includes: A space binding unit: establishes the association between the GIS coordinates of each sensor in the multi-modal sensor network and the position of the mobile formwork component in the BIM model, and realizes the spatial binding of the monitoring data and the BIM model; A monitoring marker unit: creates monitoring point families for different parameters, marks different sensor types, and annotates key sensor information; A real-time association unit: uses the MQTT protocol to associate the sensor data stream with the mobile formwork component of the BIM model in real time.

[0070] The SHI calculation module includes: ; Wherein, SHI represents the construction health index; represents the stress safety factor, represents the weight factor of the stress safety factor; represents the deformation deviation, represents the weight factor of the deformation deviation; represents the environmental risk value, represents the weight factor of the environmental risk value; 、 、 Real-time calculation according to multi-source sensor data; Real-time update through AE-LSTM hybrid model prediction.

[0071] The AE-LSTM hybrid model prediction module comprises: The first training unit: a fixed LSTM module, training the AE module, so that the AE module effectively extracts key features of historical multi-source sensor data; The second training unit: loading the AE module, accessing the LSTM module, training through historical data containing instability events, and outputting instability probability, the instability events being instability values corresponding to future k time steps of historical multi-source sensor data; The prediction unit predicts instability probability through the AE module and the LSTM module based on real-time received multi-source sensor data.

[0072] The AE-LSTM hybrid model prediction module further comprises: The historical multi-source sensor data is used as input data, and feature engineering of input data corresponding to each parameter of the construction health index SHI is performed; in the historical multi-source sensor data, the data corresponding to the stress safety factor parameter of the SHI is used to calculate the time series change rate of the stress value; the data corresponding to the deformation deviation parameter of the SHI is used to form a spatial gradient distribution according to the deformation data; the data corresponding to the environmental risk value parameter of the SHI is used to calculate the coupling oscillation frequency of the environmental parameters; the time series change rate of the stress value, the spatial gradient distribution, and the coupling oscillation frequency of the environmental parameters are used as feature vectors for training; The AE-LSTM hybrid model introduces an attention layer for dynamically adjusting the weights of the feature vectors; The output is a weight factor of each parameter of the construction health index SHI.

[0073] The early warning evaluation module comprises: The instability probability threshold and the SHI threshold are set, when the instability probability predicted by the AE-LSTM hybrid model is greater than the instability probability threshold, and the SHI calculated according to the weight factor predicted by the AE-LSTM hybrid model is greater than the SHI threshold, a mobile formwork adjustment instruction is generated and a warning is issued.

[0074] The early warning evaluation module comprises: The risk area is calibrated on the digital twin platform, a hydraulic correction scheme of the mobile formwork actuator is formulated, and an emergency plan warning is pushed.

[0075] The early warning evaluation module further comprises: the AE-LSTM hybrid model predicts every 30 seconds, the calculation of the SHI is performed every 1 second, and the weight factor predicted by the current AE-LSTM hybrid model is used for each calculation of the SHI until the next AE-LSTM hybrid model prediction weight factor is updated.

[0076] The mobile jig intelligent monitoring system based on digital twinning and multi-modal perception provided in the embodiment can implement the mobile jig intelligent monitoring method based on digital twinning and multi-modal perception described in embodiment 1 and has the same technical effects as embodiment 1.

[0077] The above embodiments are only preferred embodiments of the present application and are not used to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A mobile formwork intelligent monitoring method based on digital twin and multimodal perception, characterized in that: include: S1, establishing a multimodal sensor network on the mobile mold frame entity; S2. Establish a digital twin platform: Build a BIM model that maps to the mobile formwork entity, embed virtual monitoring points in the BIM model, match and calibrate them with the sensors of the multimodal sensor network, and integrate multi-source sensor data collected by the multimodal sensor network in real time; S3. Calculating the construction health index (SHI) in real time based on the multi-source sensor data; S4. Based on the multi-source sensor data, predict the instability probability in real time through the AE-LSTM hybrid model; and simultaneously predict the weight factors of each parameter of the construction health index (SHI) for the next calculation of the construction health index (SHI); S5. Based on the construction health index (SHI) and the predicted probability of instability, a joint assessment is conducted to determine whether to generate a mobile formwork adjustment instruction and issue an early warning.

2. The mobile formwork intelligent monitoring method based on digital twin and multimodal perception according to claim 1 is characterized in that: In step S1, the establishment of a multimodal sensor network includes: S101. Install fiber grating sensors at the mid-span and support points of the main beam of the mobile formwork to monitor the stress of the main beam; S102, installing inclination sensors at the four corners of the top plate of the movable mold frame to monitor the lateral inclination of the mold frame; S103, setting an acoustic wave sensor array on the inner side of the formwork of the movable formwork to monitor the density of the concrete vibration; S104. Install a laser displacement meter at the bottom of the supporting legs of the mobile formwork to monitor the settlement of the supporting legs; S105. A temperature, humidity and anemometer are installed on the top of the mobile mold frame to monitor environmental parameters. S106. Install industrial cameras at the key joints of the mobile formwork to monitor the misalignment of the formwork joints.

3. The mobile formwork intelligent monitoring method based on digital twin and multimodal perception according to claim 1 is characterized in that: The method for pre-embedding virtual monitoring points in the BIM model in step S2 includes: S201, establishing an association relationship between the GIS coordinates of each sensor in the multimodal sensor network and the position of the mobile module components of the BIM model, thereby achieving spatial binding of the monitoring data and the BIM model; S202. Create monitoring point families for different parameters, mark and distinguish different sensor types, and annotate key sensor information; S203, using the MQTT protocol to associate the sensor data stream with the BIM model mobile module components in real time.

4. The method for intelligent monitoring of mobile formwork based on digital twin and multimodal perception according to claim 1 is characterized in that: The calculation method of the construction health index SHI in step S3 includes: ； Among them, SHI stands for Construction Health Index; represents the stress safety factor, Weight factor representing stress safety factor; represents the deformation deviation, The weight factor representing the deformation deviation; represents the environmental risk value, A weighting factor representing the environmental risk value; 、 、 Real-time calculation based on multi-source sensor data; Real-time updates through AE-LSTM hybrid model prediction.

5. The mobile formwork intelligent monitoring method based on digital twin and multimodal perception according to claim 1 is characterized in that: In step S4, the real-time prediction of the instability probability using the AE-LSTM hybrid model includes: S401, fix the LSTM module and train the AE module through historical multi-source sensor data, so that the AE module can effectively extract key features of the multi-source sensor data; S402, loading the AE module, connecting to the LSTM module, training with historical data containing instability events, and outputting an instability probability, where the instability event is the instability value corresponding to the future k time steps of historical multi-source sensor data; S403: Based on the multi-source sensor data received in real time, predict the instability probability through the AE module and the LSTM module.

6. The method for intelligent monitoring of mobile formwork based on digital twin and multimodal perception according to claim 5 is characterized in that: The training of the AE-LSTM hybrid model also includes: Using historical multi-source sensor data as input, feature engineering is performed on the input data corresponding to each parameter of the Construction Health Index (SHI). The temporal change rate of stress values ​​corresponding to the SHI stress safety factor parameter in the historical multi-source sensor data is calculated. A spatial gradient distribution is formed based on the deformation data of the SHI deformation deviation parameter. The coupled oscillation frequency of the environmental parameter is calculated based on the data of the SHI environmental risk value parameter. Training is performed using the temporal change rate of stress values, spatial gradient distribution, and coupled oscillation frequency of the environmental parameter as feature vectors. The AE-LSTM hybrid model introduces an attention layer to dynamically adjust the weight of each feature vector; The output is the weight factor of each parameter of the construction health index SHI.

7. The method for intelligent monitoring of mobile formwork based on digital twin and multimodal perception according to claim 1 is characterized in that: Step S5 includes: Set the instability probability threshold and SHI threshold. When the instability probability predicted by the AE-LSTM hybrid model is greater than the instability probability threshold, and the SHI calculated based on the weight factor predicted by the AE-LSTM hybrid model is greater than the SHI threshold, a mobile mold frame adjustment instruction is generated and an early warning is issued.

8. The method for intelligent monitoring of mobile formwork based on digital twin and multimodal perception according to claim 7 is characterized in that: Generate mobile mold adjustment instructions and issue early warnings including: Risk areas are calibrated on the digital twin platform, a hydraulic correction plan for the mobile mold frame actuator is developed, and emergency plan alerts are pushed.

9. The method for intelligent monitoring of mobile formwork based on digital twin and multimodal perception according to claim 7 is characterized in that: Also includes: The AE-LSTM hybrid model makes predictions every 30 seconds, and the SHI calculation is performed every 1 second. Each SHI calculation uses the weight factor predicted by the current AE-LSTM hybrid model until the next AE-LSTM hybrid model predicts the weight factor, at which point the weight factor is updated.

10. A mobile formwork intelligent monitoring system based on digital twin and multimodal perception, characterized in that: include: Multimodal sensor network: Establish a multimodal sensor network on the mobile mold frame entity; Digital twin platform: Build a BIM model that maps to the mobile formwork entity, embed virtual monitoring points in the BIM model, match and calibrate them with sensors in the multimodal sensor network, and integrate multi-source sensor data collected by the multimodal sensor network in real time; SHI calculation module: calculates the construction health index SHI in real time based on the multi-source sensor data; AE-LSTM hybrid model prediction module: Based on the multi-source sensor data, the AE-LSTM hybrid model is used to predict the probability of instability in real time. At the same time, the weight factors of each parameter of the construction health index (SHI) are predicted for the next calculation of the construction health index (SHI); Early warning assessment module: Based on the construction health index (SHI) and the predicted probability of instability, it jointly assesses whether to generate a mobile formwork adjustment instruction and issue an early warning.

Images