Crane hoisting overload state monitoring method and system based on digital twinning

Abstract

The invention relates to the technical field of crane safety monitoring, and discloses a crane hoisting overload state monitoring method and system based on digital twinning, and the system comprises a twinning modeling module, a data acquisition module, a state simulation module, an overload decision module, and a model optimization and visualization module. Deep interaction between a physical world and a virtual space is realized through digital twin bodies, limitation of a traditional monitoring method is broken through, multi-source parameters such as wind speed, load and structural stress are fused for the first time, a dynamic coupling model is constructed, nonlinear correlation of wind load, inertia force and structural deformation is captured, the problem of missed judgment of hidden risks in the traditional method is solved, and the method has a wide application prospect. An accident early warning coverage range is expanded, a deflection calculation model is established based on an improved mechanical theory, high-sensitivity detection of fine deformation of the main beam is realized, compared with a static model, the magnitude order is improved, structural damage accumulation is effectively prevented, deep correlation mining is performed on historical and real-time data, and long-time-history advanced early warning of overload risks is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crane safety monitoring, in particular to a crane hoisting overload state monitoring method and system based on digital twinning. BACKGROUND

[0002] The crane monitoring system is an independent and completely computer-controlled safety operation system, which can automatically detect the mass of the load hoisted by the crane and the angle of the hoisting arm, and can display the rated load and actual load, working radius, and angle of the hoisting arm. The crane monitoring system monitors and detects the working condition of the crane in real time, has a diagnosis function, rapid danger condition alarm, and safety control.

[0003] However, the prior art has the following problems:

[0004] 1. Lack of dynamic coupling;

[0005] 2. Blank prediction mechanism;

[0006] 3. Insufficient model adaptation.

[0007] Therefore, the present application provides a crane hoisting overload state monitoring method and system based on digital twinning, which realizes full-factor perception, full-process prediction, and full-life cycle optimization of the overload state by constructing a digital twin body that integrates multi-body dynamics equations and finite element analysis, developing a spatio-temporal graph convolution risk prediction architecture, and designing a gradient boosting tree-driven model online optimization mechanism. SUMMARY

[0008] (I) Technical problems solved

[0009] In view of the deficiencies of the prior art, the present application provides a crane hoisting overload state monitoring method and system based on digital twinning, which solves the problems raised in the background art.

[0010] (II) Technical solutions

[0011] To achieve the above purpose, the present application provides the following technical solutions: a crane hoisting overload state monitoring method and system based on digital twinning, the method comprising the following steps:

[0012] S1, constructing a digital twin model of the crane, including a structure model, a dynamics model, and a working state model;

[0013] S2, collecting real-time working state data of the crane, including hoisting weight, amplitude, wind speed, inclination angle, and environmental parameters;

[0014] S3, inputting the working state data into the digital twin model for real-time simulation to generate real-time state parameters of the crane, including stress distribution, structural deformation, and stability coefficient;

[0015] S4, overload state judgment processing is performed based on the real-time state parameters to generate overload state judgment data; when it is judged that the overload state exists, S5 is performed;

[0016] S5, generating a warning information and executing a safety control strategy, including automatic load reduction, shutdown instruction and alarm signal;

[0017] S6, combining historical overload data and machine learning algorithm to optimize digital twin model parameters to generate twin model correction data;

[0018] S7, based on multi-source data fusion technology, the working state data and real-time state parameters are fused to generate overload risk prediction data;

[0019] S8, constructing overload monitoring summary data and visualizing output, including stress thermal map, deformation animation and risk level indication.

[0020] Preferably, S1 comprises:

[0021] S11, obtaining crane physical structure point cloud data by three-dimensional laser scanning to construct a high-precision structure model;

[0022] S12, establishing a dynamics model based on multi-body dynamics theory, and the motion equation is expressed as:

[0023]

[0024] Wherein M is the mass matrix, q is the generalized coordinate, C is the Coriolis force term, G is the gravity term, and τ is the generalized force;

[0025] S13, integrating sensor configuration scheme into working state model, including virtual mapping nodes of strain gauges, inclination sensors and anemometers.

[0026] Preferably, S2 comprises:

[0027] S21, collecting crane weight data W t , amplitude data R t , wind speed data V t and inclination data θ t by Internet of Things terminal, and the sampling frequency is 50Hz;

[0028] S22, using Kalman filter to process noise elimination of collected data to generate filtered working state data

[0029]

[0030] Wherein K k is the Kalman gain matrix, is the original data, is the predicted value of the previous moment.

[0031] Preferably, the S3 comprises:

[0032] S31, inputting the digital twin model into the finite element real-time simulation to calculate the stress distribution σ(x,y,z,t) of the main beam;

[0033] S32, calculating the structural deformation δ based on the Euler-Bernoulli beam theory max :

[0034]

[0035] wherein δ max is the maximum deflection of the main beam, W t is the real-time lifting capacity, is the real-time working amplitude, E is the elastic modulus, I is the sectional moment of inertia, k is the load coefficient, α is the wind resistance factor, is the real-time wind speed;

[0036] S33, generating a real-time stability coefficient wherein τ crit is the critical overturning moment, τ actual is the actual overturning moment.

[0037] Preferably, the S4 comprises:

[0038] S41, establishing an overload judgment rule: when σ max >σ yield , η t <1.5 and δ max >L / 400, it is determined to be overloaded, wherein L is the span;

[0039] S42, using a fuzzy logic system to process boundary conditions, the input variables being σ max , η t , δ max , and the output being the overload probability P overload ;

[0040] S43, when P overload >0.85, generating overload state judgment data A alert .

[0041] Preferably, the S5 comprises:

[0042] S51, constructing a hierarchical response strategy:

[0043] First-level overload: triggering an audible and light alarm;

[0044] ​Level 2 overload: Automatically reduces lifting speed to 30%;

[0045] Level 3 overload: Execute emergency stop command;

[0046] S52. Sends encrypted early warning messages to the monitoring center via the 5G module. The data structure is as follows:

[0047] Γ={ID crane ,t,W t ,R t ,η t action code}

[0048] Among them, ID crane This is a unique identifier for the crane, where 't' is a timestamp and 'action' is the action. code Coded for response actions.

[0049] Preferably, S6 includes:

[0050] S61. Extract historical overload event dataset Where X i Let y be the eigenvector. i This represents the actual deformation.

[0051] S62. The gradient boosting tree algorithm is used to optimize the stiffness parameter E and drag factor α of the digital twin model. The objective function is:

[0052]

[0053] in Let y be the deformation predicted for the i-th time. i Let be the measured deformation amount of the i-th time, β be the parameter vector to be optimized, λ be the regularization coefficient, and ||·||2 be the L2 norm penalty term;

[0054] S63. When the model error is less than 3%, generate the twin model correction data M. update .

[0055] Preferably, S7 includes:

[0056] S71, Integrating Real-Time Data Historical data Ω and environmental data E t Construct feature tensors

[0057] S72. Predicting the overload risk value R in the next 300 seconds using a spatiotemporal graph convolutional network. future :

[0058]

[0059] in is the spatio-temporal feature tensor, GCN is the graph convolution network, extracting the sensor topology relationship, is the graph convolution weight, is the Hadamard product, LSTM is the long short-term memory network, extracting the time-dependent feature;

[0060] S73, when R future >0.7, a pre-maintenance instruction is generated.

[0061] Preferably, the S8 comprises:

[0062] S81, constructing an overload monitoring summary data Pi = <sigma max , eta t , P overload , R future >;

[0063] S82, visualizing output on the digital twin platform:

[0064] The main beam stress thermodynamic map is mapped to the three-dimensional model;

[0065] The structure deformation animation is enlarged;

[0066] The real-time risk level indicator, the color mapping rule of which is: green, yellow, red.

[0067] Preferably, it comprises:

[0068] Twin modeling module: containing three-dimensional reconstruction unit, dynamics modeling unit and sensor configuration unit;

[0069] Data acquisition module: containing multi-source sensing unit, filtering and noise reduction unit and real-time transmission unit;

[0070] State simulation module: containing finite element analysis unit, deformation calculation unit and stability evaluation unit;

[0071] Overload decision module: containing fuzzy logic unit, hierarchical response unit and early warning execution unit;

[0072] Model optimization module: containing historical database, machine learning training unit and parameter correction unit;

[0073] Visualization module: containing AR rendering engine, risk map generation unit and interactive control interface.

[0074] (Three) beneficial effects

[0075] Compared with the prior art, the present application provides a crane overload state monitoring method and system based on digital twinning, which has the following beneficial effects:

[0076] 1. Full element dynamic coupling monitoring

[0077] Deep interaction between physical world and virtual space is realized through digital twin, breaking the limitations of traditional monitoring methods, and creating a dynamic coupling model by integrating wind speed, load, and structural stress, etc. Nonlinear correlation between wind load, inertia force and structural deformation is captured, solving the problem of missed judgment of hidden risks in traditional methods, improving the coverage of accident warning, and establishing a deflection calculation model based on improved mechanics theory to achieve high sensitivity detection of the main beam deformation, which improves the precision by several orders of magnitude compared with static model, effectively preventing structural damage accumulation.

[0078] 2. Advanced prediction and intelligent decision-making

[0079] An active defense system is built through artificial intelligence to overcome the problem of industry response lag, and time and space feature analysis algorithm is integrated to deeply correlate and mine historical and real-time data, realize long-term advanced warning of overload risk, and greatly shorten the response delay compared with traditional threshold mechanism. Combined with fuzzy logic and multi-level response strategy, the safety threshold is dynamically adapted to complex working conditions, reducing the probability of false triggering and ensuring the stability of continuous operation.

[0080] 3. Self-evolving model and immersive interaction

[0081] By building a continuously optimized digital twin ecosystem, the safety control mode is remodeled, and the model parameters are continuously optimized based on machine learning to give the digital twin self-evolution ability, maintain high precision state, extend the technology life cycle, and use augmented reality engine to realize the visualization linkage of stress distribution, structural deformation and risk level, improve human-computer interaction efficiency, and shorten the fault diagnosis decision chain.

[0082] 4. Safety and efficiency breakthrough

[0083] Through the dynamic coupling model, the overturning risk under extreme working conditions is accurately identified, the monitoring reliability is improved, the major accident rate is reduced, the predictive maintenance mechanism reduces unplanned downtime, and the whole life cycle management cost is optimized, and the self-adaptive ability to international safety standards is enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0084] Figure 1 The framework diagram of a crane hoisting overload state monitoring system based on digital twin is provided. Figure 2 The flowchart of a crane hoisting overload state monitoring method based on digital twin is provided. DETAILED DESCRIPTION

[0085] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0086] Please refer to Figure 1 The crane overload state monitoring method and system based on digital twinning, S1, a digital twinning model of the crane is constructed, including a structure model, a dynamics model and a working state model;

[0087] S2, real-time acquisition of crane working state data, including lifting capacity, amplitude, wind speed, inclination angle and environmental parameters;

[0088] S3, inputting the working state data into the digital twinning model for real-time simulation to generate crane real-time state parameters, including stress distribution, structure deformation and stability coefficient;

[0089] S4, overload state judgment processing based on the real-time state parameters to generate overload state judgment data; when judging overload, S5 is executed;

[0090] S5, generating early warning information and executing safety control strategy, including automatic load reduction, shutdown instruction and alarm signal;

[0091] S6, combining historical overload data and machine learning algorithm to optimize digital twinning model parameters to generate twinning model correction data;

[0092] S7, fusion processing of working state data and real-time state parameters based on multi-source data fusion technology to generate overload risk prediction data;

[0093] S8, constructing overload monitoring summary data and visualizing output, including stress heat map, deformation animation and risk level indication;

[0094] S11, obtaining crane physical structure point cloud data by three-dimensional laser scanning to construct a high-precision structure model;

[0095] S12, establishing a dynamics model based on multi-body dynamics theory, and the motion equation is expressed as:

[0096]

[0097] Wherein M is the mass matrix, q is the generalized coordinate, C is the Coriolis force term, G is the gravity term, and τ is the generalized force;

[0098] S13, configure the integrated sensor configuration scheme to the working state model, including the virtual mapping nodes of strain gauges, inclination sensors and anemometers;

[0099] S21, collect weight data W by the Internet of Things terminal t , amplitude data R t , wind speed data V t and inclination data θ t , with a sampling frequency of 50 Hz;

[0100] S22, adopt Kalman filtering to process the collected data to eliminate noise and generate filtered working state data

[0101]

[0102] wherein K k is the Kalman gain matrix, is the original data, is the predicted value at the previous time;

[0103] S31, input to the digital twin model for finite element real-time simulation to calculate the stress distribution σ(x,y,z,t) of the main beam;

[0104] S32, calculate the structural deformation δ max based on the Euler-Bernoulli beam theory:

[0105]

[0106] wherein δ max is the maximum deflection of the main beam, W t is the real-time weight, is the real-time working amplitude, E is the elastic modulus, I is the sectional moment of inertia, k is the load coefficient, α is the wind resistance factor, is the real-time wind speed;

[0107] S33, generate the real-time stability coefficient wherein τ crit is the critical overturning moment, τ actual is the actual overturning moment;

[0108] S41, establish an overload judgment rule: when σ max >σ yield , η t <1.5 and δ max >L / 400, it is determined to be overloaded, wherein L is the span;

[0109] S42, adopt a fuzzy logic system to process the boundary conditions, with σ max , η t and δmax , output overload probability P overload ;

[0110] S43, generate overload state judgment data A overload when P alert >0.85;

[0111] S51, build hierarchical response strategy:

[0112] Primary overload: trigger audible and visual alarm;

[0113] Secondary overload: automatically reduce lifting speed to 30%;

[0114] Tertiary overload: execute emergency stop instruction;

[0115] S52, send encrypted early warning message to monitoring center through 5G module, and its data structure is:

[0116] Γ={ID crane ,t,W t ,R t ,η t ,action code}

[0117] Where ID crane is the unique identification code of the crane, t is the timestamp, and action code is the response action code;

[0118] S61, extract historical overload event data set Where X i is the feature vector, and y i is the actual deformation;

[0119] S62, use gradient boosting tree algorithm to optimize stiffness parameter E and wind resistance factor α of digital twin model, and the objective function is:

[0120]

[0121] Where is the i-th predicted deformation, y i is the i-th actual deformation, β is the parameter vector to be optimized, λ is the regularization coefficient, and ||·||2 is the L2 norm penalty term;

[0122] S63, generate twin model correction data M update when the model error is less than 3%;

[0123] S71, fuse real-time data historical data Ω and environmental data E t , construct feature tensor

[0124] S72, predicting a future 300-second overload risk value R using a spatio-temporal graph convolution network future :

[0125]

[0126] wherein is a spatio-temporal feature tensor, GCN is a graph convolution network extracting sensor topological relations, Θ is a graph convolution weight, ⊙ represents a Hadamard product, LSTM is a long short-term memory network extracting time-dependent features;

[0127] S73, generating a pre-maintenance instruction when R future > 0.7;

[0128] S81, constructing an overload monitoring summary data Π = <σ max , η t , P overload , R future >;

[0129] S82, visualizing output on a digital twin platform:

[0130] a stress thermal map of the main girder is mapped to a three-dimensional model;

[0131] a structure deformation magnification animation;

[0132] a real-time risk level indicator, with color mapping rules being green, yellow, and red;

[0133] a twin modeling module including a three-dimensional reconstruction unit, a dynamics modeling unit, and a sensor configuration unit;

[0134] a data acquisition module including a multi-source sensing unit, a filtering and noise reduction unit, and a real-time transmission unit;

[0135] a state simulation module including a finite element analysis unit, a deformation calculation unit, and a stability evaluation unit;

[0136] an overload decision module including a fuzzy logic unit, a hierarchical response unit, and a pre-warning execution unit;

[0137] a model optimization module including a historical database, a machine learning training unit, and a parameter correction unit;

[0138] a visualization module including an AR rendering engine, a risk atlas generation unit, and an interactive control interface.

[0139] Embodiment 1: Implementation process of a crane overload state monitoring method

[0140] 1. Digital twin construction

[0141] three-dimensional reconstruction:

[0142] The main girder structure of the portal crane is scanned with high precision using a laser scanner, the scanning point distance is controlled at the millimeter level, and the point cloud data is converted into a boundary representation method solid model through a non-uniform rational B-spline surface reconstruction algorithm, so as to ensure that the model geometric error is less than one thousandth.

[0143] Dynamic modeling:

[0144] A three-degree-of-freedom dynamic equation set is established:

[0145] The first equation describes the balance of the slewing joint torque: including the rotational inertia term, the damping term and the elastic coupling term;

[0146] The second equation represents the dynamics of the variable amplitude mechanism: considering the gravity torque compensation and the interaction force between joints;

[0147] The third equation defines the lifting system motion: integrating mass inertia, viscous resistance and elastic restoring force.

[0148] 2. Real-time data acquisition and processing

[0149] Sensor deployment:

[0150] Load monitoring: high-precision tension sensors are installed in the wedge-shaped sleeve of the steel wire rope;

[0151] Wind speed monitoring: ultrasonic anemometer is configured at the top of the boom;

[0152] Attitude monitoring: micro-electro-mechanical system inclinometers are arranged at the four corners of the main girder;

[0153] Noise suppression:

[0154] The Kalman filter algorithm is used to process the original data, the process noise matrix is set as a diagonal dominant matrix, and the observation noise matrix is dynamically adjusted according to the sensor accuracy.

[0155] 3. State simulation and overload judgment

[0156] Finite element analysis:

[0157] The real-time load and wind speed data are input into the digital twin model, transient dynamics simulation is performed, stress distribution in the main girder mid-span region is calculated, and the position and value of the maximum stress point are identified.

[0158] Fuzzy decision:

[0159] Normalized input parameters are constructed: stress normalization value is calibrated based on material yield strength, and stability coefficient normalization value is set according to safety specifications;

[0160] Fuzzy rule base is applied for reasoning:

[0161] When high stress and low stability coefficient coexist, the output overload probability is high-risk level.

[0162] When medium stress and medium stability coefficient coexist, the output is medium risk level.

[0163] 4. Hierarchical control execution

[0164] When the third level overload state is monitored:

[0165] Send emergency stop command to programmable logic controller through controller area network bus, the command contains device identification code and risk level code;

[0166] Activate the hydraulic braking system to reduce the speed of the hook to zero at a preset slope;

[0167] Synchronously generate encrypted alarm message and transmit it to the monitoring center, the message uses key-value pair structure to record timestamp, load value, working amplitude and risk level.

[0168] Example 2: System implementation architecture

[0169] 1. Hardware configuration

[0170] The edge computing unit uses a high-performance artificial intelligence computing platform, equipped with large-capacity memory and parallel processing cores;

[0171] The 5G communication module supports the latest mobile communication standard, ensuring that the end-to-end transmission delay is less than the critical threshold;

[0172] The augmented reality display terminal is equipped with a wide-angle optical system and a holographic display unit.

[0173] 2. Software workflow

[0174] Data flow:

[0175] After the sensor data is preprocessed by the edge node, it is uploaded to the cloud digital twin platform through the 5G network;

[0176] The platform processing result is pushed to the augmented reality terminal in real time;

[0177] Risk prediction process:

[0178] The spatio-temporal graph convolution network sets a fixed length time window and multi-dimensional feature input;

[0179] The graph convolution adjacency matrix is defined according to the physical position relationship of the sensors: adjacent sensors are given high connection weight, different components in the same mechanism are given medium weight, and unrelated components are given zero weight.

[0180] Example 3: Model optimization implementation

[0181] 1. Construction of training data set:

[0182] A historical data set containing overload events and normal operating conditions is collected;

[0183] The feature vector contains the hoist weight, work amplitude, wind speed, inclination angle, and multi-position stress values.

[0184] Gradient boosting tree optimization:

[0185] Set the decision tree number, learning rate, and maximum depth hyperparameters;

[0186] Optimize the elastic modulus and wind resistance factor parameters with the measured deformation as the target value;

[0187] After optimization, the parameters are adjusted in proportion: the elastic modulus is corrected upwards, and the wind resistance factor is corrected downwards.

[0188] It should be noted that, in the present text, relational terms such as first and second and the like can only be used to distinguish one entity or action from another entity or action, without necessarily requiring or implying any such actual relationship or order between or among the entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover non-exclusive inclusions, so that a process, method, article, or apparatus that comprises a list of elements does not only include those elements, but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without more limitations, an element defined by the phrase "comprising a" does not exclude the existence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0189] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements, and variations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring the overload state of a crane based on digital twinning, characterized by: The method comprises the following steps: S1, constructing a digital twin model of the crane, including a structure model, a dynamics model and a working state model; S2, collecting crane working state data in real time, including lifting capacity, amplitude, wind speed, inclination angle and environmental parameters; S3, inputting the working state data into the digital twin model for real-time simulation to generate crane real-time state parameters, including stress distribution, structural deformation and stability coefficient; S4, performing overload state judgment processing based on the real-time state parameters to generate overload state judgment data; when it is judged that the crane is overloaded, performing S5; S5, generating a warning information and executing a safety control strategy, including automatic load reduction, shutdown instruction and alarm signal; S6, optimizing the digital twin model parameters in combination with historical overload data and machine learning algorithm to generate twin model correction data; S7, performing fusion processing on the working state data and real-time state parameters based on multi-source data fusion technology to generate overload risk prediction data; S8, constructing overload monitoring summary data and visualizing output, including stress heat map, deformation animation and risk level indication.

2. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S1 comprises: S11, obtaining crane physical structure point cloud data through three-dimensional laser scanning to construct a high-precision structure model; S12, establishing a dynamics model based on multi-body dynamics theory, and the motion equation is expressed as: wherein M is a mass matrix, q is a generalized coordinate, C is a Coriolis force term, G is a gravity term, and τ is a generalized force; S13, integrating a sensor configuration scheme into the working state model, including virtual mapping nodes of strain gauges, inclination angle sensors and anemometers.

3. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S2 comprises: S21, collecting weight data W through the Internet of Things terminal t , amplitude data R t , wind speed data V t and inclination data θ t , and the sampling frequency is 50 Hz; S22, using Kalman filtering to process the collected data to eliminate noise and generate filtered working state data where K k is the Kalman gain matrix, is the original data, is the previous predicted value.

4. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S3 comprises: S31, will Input the digital twin model for real-time finite element simulation and calculate the stress distribution σ(x,y,z,t) of the main beam; S32, calculate the structural deformation δ based on the Euler-Bernoulli beam theory max : where δ max is the maximum deflection of the main girder, W t is the real-time lifting weight, is the real-time working amplitude, E is the elastic modulus, I is the cross-sectional moment of inertia, k is the load coefficient, α is the wind resistance factor, is the real-time wind speed; S33, generating a real-time stability coefficient where τ crit is the critical overturning moment, τ actual is the actual overturning moment.

5. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S4 comprises: S41, establish overload judgment rule: when σ max >σ yield , η t <1.5 and δ max >L / 400, determine overload, wherein L is span; S42, using a fuzzy logic system to process the boundary conditions, input variable is σ max , η t , δ max , output overload probability P overload ; S43, when P overload > 0.85 generates overload state determination data A alert .

6. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S5 comprises: S51, constructing a hierarchical response strategy: First-level overload: triggering an audible and light alarm; Second-level overload: automatically reducing the lifting speed to 30%; Third-level overload: executing an emergency shutdown instruction; S52, sending an encrypted warning message to the monitoring center through a 5G module, and the data structure is: Γ = {ID crane , t, W t , R t , η t , action code} Wherein, ID crane is a unique crane identification code, t is a time stamp, action code is a response action code.

7. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S6 comprises: S61, extract historical overload event dataset where X i is the feature vector, y i is the actual deformation; S62, using a gradient boosting tree algorithm to optimize the stiffness parameter E and the wind resistance factor a of the digital twin model, and the objective function is: wherein is the i-th predicted deformation, y i is the i-th measured deformation, β is the vector of parameters to be optimized, λ is the regularization coefficient, and || · ||2is the L2 norm penalty term. S63, generating twin model correction data M when the model error is less than 3% update .

8. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S7 comprises: S71, fuse real-time data historical data Ω and environmental data E t , construct feature tensor S72, predicting a future 300 seconds overload risk value R using a spatio-temporal graph convolution network future : wherein is the spatio-temporal feature tensor, GCN is a graph convolutional network that extracts sensor topological relations, Θ is a graph convolution weight, ⊙ represents a Hadamard product, and LSTM is a long short-term memory network that extracts time-dependent features. S73, when R future >0.7 a pre-maintenance instruction is generated.

9. The crane overload state monitoring method based on digital twinning according to claim 1, characterized in that: The S8 comprises: S81, construct overload monitoring summary data Π = <σ max ,η t , P overload , R future >; S82, visualizing output on the digital twin platform: Main beam stress heat map mapping to the three-dimensional model; Structure deformation magnification animation; Real-time risk level indicator, with color mapping rules: green, yellow and red.

10. The crane overload state monitoring system based on digital twinning of claim 1, for implementing the method of any one of claims 1-9, characterized in that: Comprise: Twin modeling module: including three-dimensional reconstruction unit, dynamics modeling unit and sensor configuration unit; Data acquisition module: including multi-source sensing unit, filtering and noise reduction unit and real-time transmission unit; State simulation module: including finite element analysis unit, deformation calculation unit and stability evaluation unit; Overload decision module: including fuzzy logic unit, hierarchical response unit and warning execution unit; Model optimization module: including historical database, machine learning training unit and parameter correction unit; Visualization module: including AR rendering engine, risk map generation unit and interactive control interface.

Images