A Method and System for Monitoring the Status of Cantilever Box Girder Formwork Based on Image Recognition

Abstract

This invention discloses a method and system for monitoring the condition of cantilever box girder formwork based on image recognition, relating to the field of cantilever box girder formwork condition monitoring. The method dynamically acquires multi-source data, including recognition images of the formwork, stress data, environmental data, and external loads. First deformation data is extracted from the recognition images, and the stress and environmental data are mapped to second and third deformation data, respectively. The deformation data are then fused to obtain fused deformation data. Deformation disturbances of the formwork are estimated based on the fused deformation data and environmental data to obtain disturbance estimates. A dynamic model is constructed relating the formwork deformation to the external loads and disturbance estimates. The dynamic model is fitted using the acquired multi-source data, and the deformation is corrected using observation equations and a Kalman filter algorithm. Risk warnings are then issued based on the corrected formwork deformation and a set deformation threshold.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of suspended box girder hanging basket state monitoring, particularly to a suspended box girder hanging basket state monitoring method and system based on image recognition. BACKGROUND

[0002] The hanging basket, as an important construction equipment, is widely used in the construction process of suspended box girder in bridge construction. The deformation monitoring of the hanging basket is crucial for ensuring construction safety and engineering quality. Traditional hanging basket deformation monitoring methods mainly rely on single sensors (such as strain sensors, displacement sensors) or simple image recognition technology.

[0003] Although the above methods can provide deformation information to some extent, they have many limitations in complex construction environments. Single sensors, such as strain sensors, can accurately measure local strain, but cannot fully reflect the overall deformation of the hanging basket; displacement sensors can measure the displacement of key points, but cannot provide detailed deformation information and have limited measurement range; traditional methods usually cannot effectively consider the influence of environmental factors (such as wind speed, temperature changes) on the deformation of the hanging basket, resulting in inaccurate monitoring results. With the development of intelligent monitoring technology, image recognition technology can be used to realize real-time monitoring of the deformation of the hanging basket, but it is easily affected by factors such as lighting conditions and background interference, and the fusion of image recognition data and other sensor data in existing technology is not perfect, which cannot fully utilize the advantages of multi-source data. Therefore, we propose a suspended box girder hanging basket state monitoring method and system based on image recognition. SUMMARY

[0004] The main purpose of the present application is to provide a suspended box girder hanging basket state monitoring method and system based on image recognition, which can more comprehensively and accurately monitor the deformation of the hanging basket by fusing multiple sensor data, including image data, strain data and environmental data, and combining deformation interference estimation, thereby realizing precise monitoring of the running state of the hanging basket and effectively solving the problems in the background technology.

[0005] To achieve the above purpose, the technical scheme adopted by the present application is,

[0006] The suspended box girder hanging basket state monitoring method based on image recognition comprises the following steps:

[0007] a) dynamically collecting the recognition image of the hanging basket, stress data μ(t), environmental data ν(t) and external load F ext (t);

[0008] b) extracting first deformation data δ image (t) based on the recognition image, mapping the stress data μ(t) and the environmental data ν(t) into second deformation data δsensor (t) and the third deformation data δ env (t);

[0009] c) The first deformation data δ image (t), the second deformation data δ sensor (t) and the third deformation data δ env (t) is fused to obtain the fused deformation data δ fused (t);

[0010] d) Based on the fused deformation data δ fused The deformation disturbance of the hanging basket is estimated using the environmental data ν(t) and the environmental data ν(t), and the disturbance estimate is obtained.

[0011] e) Construct the basket deformation and the external load F ext (t) and the interference estimate Dynamic model between;

[0012] f) Fit the dynamic model using the collected multi-source data, and calculate the deformation δ(t) of the hanging basket using the fitted dynamic model.

[0013] g) Correct the deformation variable δ(t) using the observation equation and Kalman filter algorithm to obtain the corrected hanging basket deformation variable δ. corrected (t).

[0014] h) Set the deformation threshold of the hanging basket to δ safe1 δ safe2 and δ safe3 , and δ safe1 <δ safe2 <δ safe3 According to the corrected basket deformation δ corrected (t) and the set deformation threshold are used for risk warning.

[0015] A method and system for monitoring the condition of cantilever box girder formwork based on image recognition, including:

[0016] Image acquisition module, used to dynamically acquire recognition images of the hanging basket;

[0017] The stress data acquisition module is used to dynamically acquire the stress data μ(t) of the hanging basket;

[0018] The environmental data acquisition module is used to dynamically collect environmental data ν(t) of the environment in which the hanging basket is located;

[0019] The load data acquisition module is used to dynamically acquire the external load F of the hanging basket. ext (t);

[0020] The first deformation data acquisition module is used to extract the first deformation data δ based on the recognized image. image (t);

[0021] The second deformation data acquisition module is used to map the stress data μ(t) to the second deformation data δ. sensor (t);

[0022] The third deformation data acquisition module is used to map the environmental data ν(t) into third deformation data δ. env (t);

[0023] The data fusion module is used to integrate the first deformation data δ image (t), the second deformation data δ sensor (t) and the third deformation data δ env (t) is fused to obtain the fused deformation data δ fused (t);

[0024] The interference estimation module is used to estimate the interference based on the fused deformation data δ. fused The deformation disturbance of the hanging basket is estimated using the environmental data ν(t) and the environmental data ν(t), and the disturbance estimate is obtained.

[0025] The model building module is used to construct the basket deformation and the external load F. ext (t) and the interference estimate Dynamic model between;

[0026] The data fitting module is used to fit the dynamic model with the collected multi-source data, and calculate the deformation δ(t) of the hanging basket through the fitted dynamic model.

[0027] The data correction module is used to correct the deformation variable δ(t) using the observation equation and the Kalman filter algorithm, and obtain the corrected basket deformation variable δ. corrected (t);

[0028] The risk warning module is used to set the deformation threshold of the hanging basket to δ. safe1 δ safe2 and δ safe3 , and δ safe1 <δ safe2 <δ safe3 According to the corrected basket deformation δ corrected (t) and the set deformation threshold are used for risk warning.

[0029] Furthermore, in step a), the environmental data includes one or more combinations of wind speed, temperature, and humidity; the external load includes one or more combinations of construction load, self-weight load, environmental load, and dynamic load.

[0030] Furthermore, in step b),

[0031] The second deformation data δ sensor The method for obtaining (t) is as follows:

[0032] The third deformation data δ env The method for obtaining (t) is: δ env (t)=k·ν(t)

[0033] In the formula, L is the structural length of the stress data acquisition point; E is the elastic modulus of the material at the stress data acquisition point; and k is a proportionality coefficient, representing the sensitivity of environmental data to deformation.

[0034] Furthermore, in step c), the fused deformation data δ fused The calculation method for (t) is as follows:

[0035] δ fused (t)=α·δ image (t)+β·δ sensor (t)+γ·δ env (t)

[0036] In the formula, α, β and γ are constant coefficients between 0 and 1, and α+β+γ=1.

[0037] Furthermore, in step d), the expression for the dynamic model is:

[0038]

[0039] In the formula, M is the mass matrix of the hanging basket; N is the damping matrix of the hanging basket; K is the stiffness matrix of the hanging basket; δ(t) is the second derivative of the deformation δ(t); and δ'(t) is the first derivative of the deformation δ(t).

[0040] The physical meaning of the dynamic model is as follows:

[0041] M·δ”(t): represents the inertial force of the hanging basket system, that is, the inertial effect caused by the mass M of the hanging basket. When the hanging basket deforms, its mass will resist this deformation and generate an inertial force.

[0042] N·δ'(t): Represents the damping force, which is the force generated due to the damping characteristics of the hanging basket system (such as the internal damping of the material, air resistance, etc.). The damping force is proportional to the deformation rate and plays a role in slowing down the deformation change.

[0043] K·δ(t): Represents the elastic restoring force, which is the force generated due to the elastic properties of the basket (such as the elastic modulus of the material). The elastic restoring force is proportional to the deformation and attempts to restore the basket to its original state.

[0044] F ext (t): represents external forces, such as construction loads, wind loads, etc., which will cause the hanging basket to deform.

[0045] This indicates multi-source interference, including both known and unknown interference, which can also affect the deformation of the hanging basket.

[0046] δ(t): is the deformation variable, which describes the deformation state of the hanging basket.

[0047] δ'(t): is the time first derivative of the deformation, describing the rate of deformation.

[0048] δ(t): is the second time derivative of the deformation, describing the acceleration of the deformation.

[0049] Furthermore, the expression for the observation equation is: δ fused (t)=δ(t)+ε(t); where ε(t) is the observation noise, representing the fused deformation data δ fused The difference between (t) and the actual deformation.

[0050] Furthermore, the specific steps for correcting the deformation variable δ(t) using the observation equation and the Kalman filter algorithm are as follows:

[0051] δ corrected (t)=δ(t)+K(t)×[δ fused (t)-δ(t)]

[0052] In the formula, K(t) is the Kalman gain.

[0053] Furthermore, the risk warning principle is as follows:

[0054] When δ corrected (t)<δ safe1 If the hanging basket is deemed safe, no warning will be issued.

[0055] When δ safe1 ≤δ corrected (t)<δ safe2 If the hanging basket status is deemed to pose a low risk, a Level 1 warning is issued.

[0056] When δ safe2 ≤δ corrected (t)<δ safe3 At that time, the hanging basket status was determined to be of medium risk, and a level two warning was issued;

[0057] When δ corrected (t)≥δ safe3 If the hanging basket status is deemed to pose a high risk, a level three warning will be issued.

[0058] The present invention has the following beneficial effects:

[0059] Compared with existing technologies, this solution fully leverages the advantages of different data sources by integrating data from multiple sensors, effectively reducing the errors and limitations of single sensor data, thereby significantly improving the accuracy of basket deformation monitoring.

[0060] Compared with existing technologies, this solution can effectively improve the anti-interference capability of the monitoring system in complex construction environments by estimating multi-source interference, thus ensuring the accuracy and reliability of monitoring results.

[0061] Compared with existing technologies, this solution, based on dynamic models and numerical solutions, can calculate the deformation of the hanging basket in real time and perform real-time correction through a Kalman filter. This enables real-time dynamic monitoring of the hanging basket's deformation, promptly reflecting its deformation state and providing strong protection for construction safety.

[0062] Compared with existing technologies, this solution comprehensively considers the impact of environmental factors on the deformation of the hanging basket and incorporates them into the monitoring model, enabling the monitoring system to maintain good performance under different environmental conditions and enhancing the system's environmental adaptability.

[0063] Compared with existing technologies, this solution can detect potential safety hazards in advance by monitoring the deformation of the hanging basket in real time and issuing early warning signals. This provides construction personnel with sufficient time to take measures, thereby effectively reducing safety risks during construction and ensuring the safety of construction personnel and the quality of the project.

[0064] Compared with existing technologies, this solution provides construction managers with real-time and accurate information on the deformation of the hanging basket, which helps to better grasp the construction progress and quality, optimize construction plans and resource allocation, and improve construction efficiency and management level. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of the overall structure of the image recognition-based cantilever box girder formwork status monitoring method and system of the present invention.

[0066] Figure 2 This is a schematic diagram of the image recognition-based cantilever box girder formwork status monitoring system of the present invention. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0068] The specific implementation process of the technical solution of this invention includes the following steps:

[0069] a) Dynamically acquire the identification image of the hanging basket, stress data μ(t), environmental data ν(t), and external load F. ext (t);

[0070] Environmental data includes one or more of wind speed, temperature, and humidity;

[0071] External loads include:

[0072] Construction loads include concrete pouring loads, the weight of construction personnel and equipment, etc.

[0073] Self-weight load: The weight of the hanging basket itself and the weight of other structural components on the hanging basket.

[0074] Environmental loads: such as wind loads, thermal expansion and contraction caused by temperature changes, etc.

[0075] Dynamic loads: such as vibrations and impacts during construction.

[0076] External load determination method:

[0077] Construction loads: including concrete pouring load F concrete and the weight load F of construction personnel and equipment personnel ;

[0078] Calculation method: The load is calculated based on the volume and density of the concrete poured. For example, the density of concrete is approximately 2400 kg / m³. 3 The pouring volume V can be obtained from the construction design drawings.

[0079] Formula: F concrete =ρ concrete ·V·g; where ρ concrete V is the density of the concrete, V is the volume of the concrete poured, and g is the acceleration due to gravity.

[0080] Measurement method: During the concrete pouring process, the weight of the concrete can be measured in real time using a weighing sensor.

[0081] The weight load F of construction personnel and equipment personnel :

[0082] Calculation method: Estimated based on the weight of construction personnel and equipment.

[0083] formula: Where N is the number of construction workers and equipment, m i It is the weight of the i-th construction worker or piece of equipment.

[0084] Self-weight load F self

[0085] Calculation method: Calculate the weight of the hanging basket and its auxiliary structures based on the structural design drawings of the hanging basket.

[0086] Formula: F self =m self ·g; where m self It is the total weight of the hanging basket and its auxiliary structures.

[0087] Environmental loads (with wind load F) wind (For example)

[0088] Wind load F wind :

[0089] Calculation method: Calculate the wind load based on the wind speed and the windward area of ​​the hanging basket.

[0090] The formula for wind load is: F wind =0.5ρ air ·A·v 2 ·C d ; where ρ air It is the density of air (approximately 1.225 kg / m³). 3 A is the windward area of ​​the hanging basket, v is the wind speed, and C is the windward area of ​​the hanging basket. d It is the drag coefficient (usually around 1.2).

[0091] Measurement method: Wind speed is measured in real time using a wind speed sensor.

[0092] Thermal expansion and contraction caused by temperature changes F thermal :

[0093] Calculation method: Calculate the force caused by deformation based on the material's coefficient of thermal expansion and temperature changes.

[0094] The formula is: F thermal =E·A·a·ΔT

[0095] Where E is the elastic modulus of the material, A is the cross-sectional area, α is the coefficient of thermal expansion, and ΔT is the temperature change.

[0096] Measurement method: Temperature changes are measured in real time using a temperature sensor.

[0097] Dynamic load F dynamic

[0098] Calculation method: Dynamic loads are usually difficult to calculate accurately, but can be estimated through vibration analysis or empirical formulas.

[0099] Measurement method: Dynamic load is measured in real time using an accelerometer or vibration sensor.

[0100] By summing up the various loads mentioned above, the external load F is obtained. ext (t); the calculation formula is:

[0101] F ext (t)=F concrete +F personnel +F self +F wind +F thermal +F dynamic .

[0102] b) Extracting the first deformation data δ based on the recognized image image (t), mapping the stress data μ(t) and environmental data ν(t) to the second deformation data δ respectively. sensor (t) and the third deformation data δ env (t);

[0103] Image recognition technology is used to extract the geometric features of the hanging basket from the image, such as the degree of bending of the main truss and the displacement of the nodes, and these geometric features are converted into deformation data. For example, by calculating the bending angle of the main truss or the displacement of the nodes, the first deformation data δ is obtained. image (t);

[0104] There is a physical relationship between strain and deformation. For example, strain can be converted into displacement or deformation using the elastic modulus and cross-sectional geometric parameters of a material; hence, the second deformation data δ sensor The method to obtain (t) can be:

[0105] Environmental data (such as wind speed and temperature) can indirectly affect the deformation of the hanging basket. For example, wind speed may cause vibration of the hanging basket, and temperature changes may cause thermal expansion and contraction of materials. A third deformation data δ can be obtained by establishing a model of the relationship between environmental factors and deformation. env (t), specifically: δ env (t)=k·ν(t);

[0106] In the formula, L is the structural length of the stress data acquisition point; E is the elastic modulus of the material at the stress data acquisition point; and k is a proportionality coefficient, representing the sensitivity of environmental data to deformation.

[0107] c) The first deformation data δ image (t), second deformation data δ sensor (t) and the third deformation data δenv (t) is fused to obtain the fused deformation data δ fused (t);

[0108] Fusion deformation data δ fused The calculation method for (t) is as follows:

[0109] δ fused (t)=α·δ image (t)+β·δ sensor (t)+γ·δ env (t)

[0110] In the formula, α, β and γ are constant coefficients between 0 and 1, and α+β+γ=1.

[0111] d) Based on fused deformation data δ fused The deformation disturbance of the hanging basket is estimated using the environmental data ν(t) and ν(t), and the disturbance estimate is obtained.

[0112] Specifically, interference estimation It can be divided into known interference. and unknown interference in,

[0113] Known interference For example, changes in ambient temperature and wind loads can be measured by sensors.

[0114] Specifically, in this embodiment, several known disturbances under different environmental factors are presented. Calculation method:

[0115] When only wind speed disturbance exists, the known disturbance Where ρ is the air density, C d This refers to the drag coefficient (calibrated through wind tunnel testing).

[0116] When only temperature disturbance exists, the known disturbance Where E is the elastic modulus, r is the coefficient of linear expansion (materials handbook), and ΔT(t) is the temperature change.

[0117] When only solar interference exists, the known interference Where I is the moment of inertia of the cross section and h is the elevation;

[0118] When wind speed disturbance and temperature disturbance coexist, the known disturbances

[0119] When wind speed interference and solar radiation interference coexist, the known interference

[0120] When temperature disturbance and solar radiation disturbance coexist, the known disturbances

[0121] When wind speed disturbance, temperature disturbance, and solar radiation disturbance coexist, the known disturbances

[0122] Unknown interference For example, random vibrations during construction and equipment operating noise are often difficult to measure precisely, but we can assume they are bounded, i.e., satisfy the following conditions: Where D is the upper bound of the interference, which can be determined based on experience in the specific implementation process.

[0123] e) Construct the basket deformation and external load F ext (t) and disturbance estimation Dynamic model between;

[0124] The expression for the dynamic model is:

[0125]

[0126] In the formula, M is the mass matrix of the hanging basket; N is the damping matrix of the hanging basket; K is the stiffness matrix of the hanging basket; δ(t) is the second derivative of the deformation δ(t); and δ'(t) is the first derivative of the deformation δ(t).

[0127] It should be noted that the physical meaning of the dynamic model is as follows:

[0128] M·δ”(t): represents the inertial force of the hanging basket system, that is, the inertial effect caused by the mass M of the hanging basket. When the hanging basket deforms, its mass will resist this deformation and generate an inertial force.

[0129] N·δ'(t): Represents the damping force, which is the force generated due to the damping characteristics of the hanging basket system (such as the internal damping of the material, air resistance, etc.). The damping force is proportional to the deformation rate and plays a role in slowing down the deformation change.

[0130] K·δ(t): Represents the elastic restoring force, which is the force generated due to the elastic properties of the basket (such as the elastic modulus of the material). The elastic restoring force is proportional to the deformation and attempts to restore the basket to its original state.

[0131] F ext (t): represents external forces, such as construction loads, wind loads, etc., which will cause the hanging basket to deform.

[0132] This indicates multi-source interference, including both known and unknown interference, which can also affect the deformation of the hanging basket.

[0133] δ(t): is the deformation variable, which describes the deformation state of the hanging basket.

[0134] δ'(t): is the time first derivative of the deformation, describing the rate of deformation.

[0135] δ(t): is the second time derivative of the deformation, describing the acceleration of the deformation.

[0136] These variables together describe the dynamic behavior of the hanging basket under external forces and disturbances.

[0137] The process for determining the mass matrix M, damping matrix N, and stiffness matrix K of the hanging basket system can be obtained through the following steps:

[0138] Finite element modeling → Parameter acquisition → Matrix assembly → Experimental calibration

[0139] The specific method is as follows:

[0140] Determining the mass matrix M: First calculate the "unit mass matrix", then assemble it into the "overall mass matrix".

[0141] Element mass matrix

[0142] The main truss, bottom formwork, side formwork, and slings of the hanging basket are discretized into beam, slab, and rod elements using the uniform mass formula;

[0143] Assembly overall quality matrix

[0144] The mass matrix M can be obtained by superimposing the mass matrices of all elements according to the correspondence of the degrees of freedom of the nodes.

[0145] The method for determining the stiffness matrix K is the same as above.

[0146] Determination of the damping matrix N: Damping is difficult to "directly assemble" like mass and stiffness. It is usually obtained by three engineering approximations, namely Rayleigh damping, composite modal damping, and experimental identification method (using the frequency response function measured by impact or environmental vibration test and inverting it with optimization algorithm).

[0147] Boundary conditions and load corrections

[0148] Support constraints: Delete or modify the corresponding degree of freedom in K;

[0149] Additional mass: The wet weight of concrete, construction equipment, etc., are added to M by lumped mass;

[0150] Additional stiffness: The elastic effect of the formwork and concrete contact surface and the sling is converted into an equivalent spring connected in parallel to K.

[0151] Calibration and Verification

[0152] After completing the finite element model, the first 3–5 natural frequencies and damping ratios are measured through field modal tests (hammer impact or environmental excitation). The results are compared with the calculation results, and parameters are finely adjusted in reverse to ensure that the error is less than 5%. Then the mass matrix M, damping matrix N, and stiffness matrix K can be finally determined.

[0153] f) Fit the dynamic model using the collected multi-source data, and calculate the deformation δ(t) of the hanging basket using the fitted dynamic model.

[0154] g) Correct the deformation variable δ(t) using the observation equation and Kalman filter algorithm to obtain the corrected basket deformation variable δ. corrected (t).

[0155] The expression for the observation equation is: δ fused (t)=δ(t)+ε(t); where ε(t) is the observation noise.

[0156] The specific steps for correcting the deformation δ(t) using the Kalman filter algorithm are as follows:

[0157] δ corrected (t)=δ(t)+K(t)×[δ fused (t)-δ(t)]

[0158] In the formula, K(t) is the Kalman gain.

[0159] It should be noted that the observation noise ε(t) represents the fused deformation data δ fused The difference between (t) and the true deformation can be determined by the following method:

[0160] Method 1:

[0161] Statistical analysis: Perform statistical analysis on the measurement data to calculate its mean and variance. Assume the observation noise is zero-mean Gaussian noise, i.e.: ∈(t)~N(0,σ) 2 ); where σ 2 It is the variance of the observation noise, which can be estimated using experimental or historical data.

[0162] Method 2:

[0163] Model error analysis: This involves analyzing the errors in the data transformation model and the environmental impact model, as these errors directly affect the magnitude of observation noise. For example, if the image recognition algorithm has low accuracy, or the environmental impact model is not accurate enough, the observation noise will increase accordingly.

[0164] h) Set the deformation threshold of the hanging basket to δ safe1 δ safe2 and δ safe3 , and δ safe1 <δ safe2<δ safe3 According to the corrected basket deformation δ corrected (t) and the set deformation threshold are used for risk warning. The risk warning principle is as follows:

[0165] When δ corrected (t)<δ safe1 If the hanging basket is deemed safe, no warning will be issued.

[0166] When δ safe1 ≤δ corrected (t)<δ safe2 If the hanging basket status is deemed to pose a low risk, a Level 1 warning is issued.

[0167] When δ safe2 ≤δ corrected (t)<δ safe3 At that time, the hanging basket status was determined to be of medium risk, and a level two warning was issued;

[0168] When δ corrected (t)≥δ safe3 If the hanging basket status is deemed to pose a high risk, a level three warning will be issued.

[0169] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring the state of a hanging basket for a cast-in-place box girder based on image recognition, characterized in that, Comprising the following steps: a) dynamically collecting an identification image, stress data , environmental data and external loads of the hanging basket; b) extracting first morphing data based on the identification image mapping the stress data and the environmental data to second morphing data and third morphing data respectively; c) fusing the first deformation data , the second deformation data and the third deformation data to obtain fused deformation data ; d) estimating a deformation disturbance of the hanging basket based on the fusion deformation data and the environmental data ; and ; e) constructing a dynamic model of the hanging basket shape variation with respect to the external load and the disturbance estimate ​ f) fitting the dynamic model with the collected multi-source data, calculating the deformation of the hanging basket through the fitted dynamic model , In step b), the environment data includes one or more combinations of wind speed, temperature and humidity; the external load includes one or more combinations of construction load, self-weight load and environmental load. The second deformation data The acquisition method is: = ; The third deformation data The acquisition method is: = ; in which, Ls is the structural length of the stress data acquisition point; Es is the modulus of elasticity of the material of the stress data acquisition point; is a proportionality coefficient, representing the sensitivity of the environmental data to the deformation variable, the expression of the dynamic model being, in step e), + + = + ； wherein is the mass matrix of the hanging basket; is the damping matrix of the hanging basket; is the stiffness matrix of the hanging basket; is the second derivative of the deformation variable ; and is the first derivative of the deformation variable .

2. The image recognition-based state monitoring method and system for the hanging basket of the cast-in-place box girder, according to claim 1, characterized in that, In step a), the environment data includes one or more combinations of wind speed, temperature and humidity; the external load includes one or more combinations of construction load, self-weight load and environmental load.

3. The image recognition-based state monitoring method of the hanging basket of the cast-in-place box girder according to claim 2, characterized in that, In step c), the fusion deformation data The calculation method is: = + + ； wherein , and are constant coefficients between 0 and 1, and + + = 1.

4. The image recognition-based state monitoring method of the hanging basket of the cast-in-place box girder according to claim 3, characterized in that, Further comprising: g) correcting the deformation variables using an observation equation and a Kalman filter algorithm, obtaining corrected deformation variables of the hanging basket .​ 5. The image recognition-based state monitoring method of the hanging pouring box girder basket according to claim 4, characterized in that, The expression of the observation equation is: = + ; wherein, is the observation noise, representing the difference between the fused deformation data and the true deformation variable.

6. The image recognition-based state monitoring method of the hanging pouring box girder basket according to claim 5, characterized in that, The deformation is analyzed using the observation equation and the Kalman filter algorithm. The specific steps for calibration are as follows: = + ×[ - ] In the formula, is the Kalman gain.

7. The image recognition-based state monitoring method of the hanging box girder basket according to claim 6, characterized in that, Further comprising: h) setting the deformation threshold value of the hanging basket as , and , and < < , according to the corrected hanging basket deformation and the set deformation threshold value, a risk warning is carried out, and the risk warning principle is: When < the hanging basket state is safe, and no pre-warning is performed. When ≤ < , it is determined that the hanging basket state has low risk, and a first-level early warning is performed. When ≤ < , it is determined that the hanging basket state has a medium risk, and a second-level early warning is performed. When ≥ When the hanging basket state is determined to have high risk, a three-level early warning is performed.

8. A system for monitoring the state of a hanging basket for casting a box girder based on image recognition according to claim 7, characterized in that, Comprising: An image acquisition module for dynamically acquiring an identification image of the hanging basket; A stress data acquisition module is used to dynamically acquire stress data of the hanging basket ; An environmental data collection module is configured to dynamically collect environmental data of an environment in which the hanging basket is located. ; The load data acquisition module is used for dynamically collecting the external load of the hanging basket ; a first deformation data acquisition module configured to extract first deformation data based on the identified image ; a second deformation data acquisition module, configured to map the stress data into second deformation data ;​ a third deformation data acquisition module, configured to map the environment data into third deformation data ; a data fusion module configured to fuse the first deformation data , the second deformation data , and the third deformation data to obtain fused deformation data ; an interference estimation module configured to estimate the deformation interference of the hanging basket based on the fused deformation data and the environmental data an interference estimation module configured to estimate the deformation interference of the hanging basket based on the fused deformation data ; a model construction module, configured to construct a dynamic model between a hanging basket deformation variable and the external load and the interference estimate ​ a data fitting module configured to fit the dynamic model with the collected multi-source data, and to calculate the deformation of the hanging basket through the fitted dynamic model ; A data correction module is configured to correct the deformation variable by using an observation equation and a Kalman filtering algorithm and obtain a corrected deformation variable of the hanging basket ​ The risk warning module is configured to set a deformation threshold of the hanging basket as 、 and , and < < , and perform risk warning according to the corrected deformation of the hanging basket and the set deformation threshold.

Images