Bucket foundation structure corrosion protection coating thickness control method, device and medium considering multi-factor coupling

Abstract

The application relates to a barrel foundation structure anticorrosion coating thickness control method, device and medium considering multi-factor coupling, wherein the method comprises the following steps: S1, a coating dry film thickness prediction model is constructed based on simulation data sets and historical data sets; S2, real-time environmental wind speed and spraying distance are acquired, and reference spraying pressure and reference nozzle walking speed are obtained based on a preconfigured response surface database query; S3, based on the coating dry film thickness prediction model, spraying pressure correction amount sequences and nozzle walking speed correction amount sequences are obtained by minimizing film thickness control deviation in a future period of time and fluctuation targets of the spraying pressure and the nozzle walking speed; and S4, the spraying pressure correction amount sequences and the nozzle walking speed correction amount sequences are respectively superimposed on corresponding reference spraying pressure and reference nozzle walking speed to obtain target spraying pressure and target nozzle walking speed. Compared with the prior art, the application improves film thickness prediction accuracy under variable working conditions and realizes accurate closed-loop control of coating thickness.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a method, device and medium for controlling the thickness of anti-corrosion coatings on barrel-type foundation structures that takes into account the coupling effect of multiple factors. Background Technology

[0002] The application of barrel-type foundation structures in marine engineering construction in my country is increasing. These structures are exposed to harsh marine corrosive environments and rely on high-performance anti-corrosion coatings for protection. The thickness and uniformity of the coating are key indicators determining its protective lifespan. The highest point of the anti-corrosion coating on barrel-type foundation structures reaches approximately 30 meters above the ground. Current anti-corrosion operations are mainly carried out manually, which is a high-risk, high-altitude operation. This results in low spraying efficiency, significant influence on coating quality from human intervention, and potential material waste. Therefore, it is essential to develop intelligent anti-corrosion spraying equipment to replace manual labor. This requires research into intelligent equipment development, process parameters, and intelligent equipment construction techniques to achieve unmanned spraying operations.

[0003] While some existing technologies, such as Chinese patent CN116628875A, disclose methods and systems for predicting coating thickness and spraying parameters for arc-shaped spraying trajectories, these methods are not applicable to intelligent spraying operations on barrel-type foundation structures. Intelligent spraying equipment for barrel-type foundation structures is a complex system under the coupling of multiple factors, and currently suffers from several problems and shortcomings. First, process parameters such as spraying pressure and nozzle movement speed are set manually before spraying, making it difficult to ensure uniform film thickness when wind speed changes significantly, and coating quality is greatly influenced by human experience. Second, traditional film thickness prediction models are mostly simple linear formulas that do not consider the complex nonlinear coupling effects between various factors such as wind speed, spraying distance, and paint viscosity, resulting in low prediction accuracy. Third, there is a lack of effective real-time control methods; when encountering changes in wind speed, it is impossible to adjust spraying process parameters in a timely manner, leading to either excessively thin or excessively thick coatings in some areas, resulting in orange peel or sagging phenomena and material waste.

[0004] Therefore, there is an urgent need for a method to control the thickness of the anti-corrosion coating for barrel-type foundation structures that takes into account the coupling effect of multiple factors, in order to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of the prior art by providing a method, device, and medium for controlling the thickness of anti-corrosion coatings on barrel-type foundation structures that considers the coupling effect of multiple factors.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for controlling the thickness of anti-corrosion coatings for barrel-type foundation structures considering the coupling effects of multiple factors, comprising:

[0008] Step S1: Construct a coating dry film thickness prediction model based on simulation dataset and historical dataset. The input of the coating dry film thickness prediction model includes flow coefficient, nozzle orifice area, spraying pressure, coating density, nozzle travel speed, spraying distance, ambient wind speed and coating viscosity. The output is the coating dry film thickness.

[0009] Step S2: Obtain real-time ambient wind speed and spraying distance, and combine them with pre-configured flow coefficient, nozzle orifice area, paint density and paint viscosity, and query the pre-configured response surface database to obtain the benchmark spraying pressure and benchmark nozzle travel speed.

[0010] Step S3: Based on the coating dry film thickness prediction model, in order to minimize the film thickness control deviation and the fluctuation targets of spraying pressure and nozzle travel speed in the future, the spraying pressure correction sequence and the nozzle travel speed correction sequence are obtained.

[0011] Step S4: Superimpose the first element of each of the spray pressure correction sequence and the nozzle travel speed correction sequence onto the corresponding reference spray pressure and reference nozzle travel speed to obtain the target spray pressure and target nozzle travel speed.

[0012] The mathematical expression for the coating dry film thickness prediction model is:

[0013] ;

[0014] in: DFT For the dry film thickness of the coating, K The volume solids content and density constant of the coating are given. The total mass of paint sprayed from the nozzle. For transmission efficiency, v The nozzle travel speed, For error compensation, P For spraying pressure, D For spraying distance, U w For ambient wind speed, μ This refers to the viscosity of the coating.

[0015] The transmission efficiency is specifically:

[0016] ;

[0017] in, a This is the coefficient for the spraying distance term. b This is the coefficient for the environmental wind speed term. c This is the coefficient for the viscosity term of the coating. d The coefficient of the zero-degree term.

[0018] Step S1 includes:

[0019] Step S1-1: Eliminate errors through simulation experiments to obtain a simulation dataset, wherein each sample of the simulation dataset includes at least the coating dry film thickness, the total mass of paint sprayed from the nozzle, the spraying distance, the ambient wind speed, the paint viscosity, and the nozzle travel speed;

[0020] Step S1-2: Based on the coating dry film thickness, total mass of paint sprayed from the nozzle and nozzle travel speed in each sample of the simulation dataset, the transmission efficiency of each sample is calculated by combining the paint volume solids content and density constant.

[0021] Step S1-3: Based on the spraying distance, ambient wind speed and coating viscosity in each sample of the simulation dataset, the coefficients of the spraying distance term, ambient wind speed term, coating viscosity term and zero-order term are obtained by fitting the transmission efficiency, and a coating dry film thickness prediction model without error compensation is obtained.

[0022] Step S1-4: Obtain historical datasets, wherein each sample in the historical datasets includes at least the actual dry film thickness of the coating, spraying pressure, total mass of paint sprayed from the nozzle, spraying distance, ambient wind speed, paint viscosity, and nozzle travel speed;

[0023] Step S1-5: Substitute the spraying distance, total mass of paint sprayed from the nozzle, ambient wind speed, paint viscosity and nozzle travel speed from each sample in the historical dataset into the coating dry film thickness prediction model without error compensation to obtain the predicted coating dry film thickness, and calculate the difference between the actual coating dry film thickness and the predicted coating dry film thickness as the error compensation term corresponding to each sample in the historical dataset.

[0024] Steps S1-6: Construct the first sample, wherein the error compensation term of each sample in the historical dataset is used as the label value of the first sample, and the spraying pressure, nozzle travel speed, spraying distance, ambient wind speed and paint viscosity of each sample in the historical dataset are used as the input values ​​of the first sample.

[0025] Steps S1-7: Train the error compensation model using the first sample, wherein the inputs to the error compensation are spraying pressure, nozzle travel speed, spraying distance, ambient wind speed and paint viscosity, and the output is the error compensation term;

[0026] Steps S1-8: Combine the coating dry film thickness prediction model without error compensation with the error compensation term to obtain the coating dry film thickness prediction model.

[0027] The total mass of the paint sprayed from the nozzle is:

[0028] ;

[0029] in: For flow coefficient, This represents the nozzle orifice area. This refers to the density of the coating.

[0030] Step S2 includes:

[0031] Step S2-1: Obtain real-time ambient wind speed and spraying distance;

[0032] Step S2-2: Discretize the future time period into multiple time periods, and obtain the environmental wind speed for the future multiple time periods based on the real-time environmental wind speed sequence of the current and past multiple time periods;

[0033] Step S2-3: Combine the current and future ambient wind speeds with the spraying distance, as well as the pre-configured flow coefficient, nozzle orifice area, paint density, and paint viscosity, and query the pre-configured response surface database to obtain the reference spraying pressure and reference nozzle travel speed corresponding to each future time period.

[0034] Step S3 includes:

[0035] Step S3-1: Initialize the correction amount sequence, wherein each element in the correction amount sequence corresponds to the spraying pressure correction amount and nozzle travel speed correction amount for each future time period;

[0036] Step S3-2: Construct an objective function by minimizing the sum of film thickness control deviations over all time periods and the sum of fluctuations in spraying pressure and nozzle travel speed. Iteratively update the correction sequence to obtain the optimal correction sequence, and then analyze the optimal correction sequence into a spraying pressure correction sequence and a nozzle travel speed correction sequence.

[0037] The process of obtaining the film thickness control deviation for a single time period in step S3-2 is as follows:

[0038] The correction sequence is analyzed to obtain the spray pressure correction and nozzle travel speed correction for a single time period. The obtained spray pressure correction and nozzle travel speed correction are then superimposed on the corresponding reference spray pressure and reference nozzle travel speed, respectively.

[0039] The superimposed spraying pressure and nozzle travel speed, combined with ambient wind speed, spraying distance, and pre-configured flow coefficient, nozzle orifice area, coating density, and coating viscosity, are substituted into the coating dry film thickness prediction model to obtain the first predicted film thickness. The difference between the target dry film thickness and the first predicted film thickness is calculated as the film thickness control deviation.

[0040] A device for controlling the thickness of anti-corrosion coating on a barrel-type foundation structure that considers the coupling effect of multiple factors includes a memory, a processor, and a program stored in the memory, wherein the processor executes the program to implement the method described above.

[0041] A storage medium having a program stored thereon, which, when executed, implements the method described above.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. This application can achieve high-precision prediction. By using a hybrid modeling method that combines physical mechanisms and data-driven approaches, it fully considers the complex coupling effects between multiple factors and significantly improves the accuracy of film thickness prediction under varying wind speed conditions.

[0044] 2. This application can realize intelligent real-time control, achieve feedforward control through response surface database, and achieve feedback through correction amount, thereby realizing feedforward and feedback composite control. It can not only respond quickly to measurable wind speed disturbances, but also overcome model mismatch and unmeasurable disturbances through continuous optimization of correction amount, thus realizing accurate closed-loop control of film thickness and having strong robustness.

[0045] 3. This application is fully automated and adaptive. The system can automatically adapt to changes in distance and environmental wind speed, reducing reliance on human experience and improving the stability and consistency of construction quality.

[0046] 4. This application can promote material conservation and efficiency improvement. Through precise control, it avoids overspraying and material waste, while reducing rework caused by unqualified film thickness, thus comprehensively improving construction efficiency and economy.

[0047] 5. In the construction of the coating dry film thickness prediction model, a coating dry film thickness prediction model without error compensation is first built based on the simulation dataset. Then, an error-compensated model is trained based on historical datasets. Finally, the outputs of the coating dry film thickness prediction model without error compensation and the trained error-compensated model are superimposed to obtain the coating dry film thickness prediction model. This approach clarifies that the data sources for model construction are divided into simulation datasets and historical datasets, and specifies in detail the minimum sample parameters that each dataset should contain. Obtaining simulation datasets by eliminating errors through simulation experiments can efficiently and cost-effectively generate clean data covering multiple working conditions and multiple parameter combinations, providing a high-quality data foundation for establishing the core physical relationship model. Using historical datasets can reflect the complex factors in actual operations, providing a real basis for training the error-compensated model. This phased and data-source-based model construction method ensures that the model has both solid theoretical support and fits the actual application scenario. Furthermore, it clearly divides the model construction into two main parts. First, a coating dry film thickness prediction model without error compensation is obtained based on simulation data. This part is mainly based on physical laws such as fluid mechanics, and the model has clear physical meaning and interpretability. Then, an error compensation model is trained using historical data. This part employs a data-driven approach to learn and compensate for dynamic characteristics and non-ideal factors that are not fully represented by the physical model. This hybrid modeling architecture, combining a physical mechanism model with data-driven error compensation, possesses both the structural integrity and stability of a physical model and the flexibility and high accuracy of a data-driven model, significantly improving the overall performance of the model. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the main steps of the method of the present invention. Detailed Implementation

[0049] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0050] A method for controlling the thickness of anti-corrosion coatings for barrel-type foundation structures considering the coupling effect of multiple factors, such as... Figure 1 As shown, it includes:

[0051] Step S1: Construct a coating dry film thickness prediction model based on the simulation dataset and historical dataset. The inputs of the coating dry film thickness prediction model include the flow coefficient, nozzle orifice area, spraying pressure, coating density, nozzle travel speed, spraying distance, ambient wind speed and coating viscosity. The output is the coating dry film thickness.

[0052] In this embodiment, the basic relationship between the coating dry film thickness and the volume of the coating dry film deposited per unit area is defined as being directly proportional to the coating dry film volume and inversely proportional to the nozzle travel speed. Specifically, the mathematical expression for the coating dry film thickness prediction model is as follows:

[0053] ;

[0054] in: DFT For the dry film thickness of the coating, K The volume solids content and density constant of the coating are given. The total mass of paint sprayed from the nozzle. For transmission efficiency, v The nozzle travel speed, For error compensation, P For spraying pressure, D For spraying distance, U w For ambient wind speed, μ This refers to the viscosity of the coating.

[0055] Step S1 includes:

[0056] Step S1-1: Eliminate errors through simulation experiments to obtain a simulation dataset. Each sample in the simulation dataset includes at least the coating dry film thickness, the total mass of paint sprayed from the nozzle, the spraying distance, the ambient wind speed, the paint viscosity, and the nozzle travel speed.

[0057] Step S1-2: Based on the coating dry film thickness, total mass of paint sprayed from the nozzle and nozzle travel speed in each sample of the simulation dataset, the transmission efficiency of each sample is calculated by combining the paint volume solids content and density constant.

[0058] Step S1-3: Based on the spraying distance, ambient wind speed and coating viscosity in each sample of the simulation dataset, the coefficients of the spraying distance term, ambient wind speed term, coating viscosity term and zero-order term are obtained by fitting the transmission efficiency, and a coating dry film thickness prediction model without error compensation is obtained.

[0059] The transmission efficiency is defined as a fusion function of spraying distance, ambient wind speed, and paint viscosity. This fusion function characterizes the ratio of the actual mass of paint reaching the surface of the barrel-shaped foundation structure to the total mass of paint sprayed from the nozzle under the coupled effects of multiple factors. The parameters of this fusion function are obtained through joint calibration using computational fluid dynamics simulations and experimental data. Specifically, the transmission efficiency is as follows:

[0060] ;

[0061] in, a This is the coefficient for the spraying distance term. b This is the coefficient for the environmental wind speed term. cThis is the coefficient for the viscosity term of the coating. d Step S1-4 for zero-order term coefficients: Obtain historical datasets, wherein each sample in the historical dataset includes at least the actual dry film thickness of the coating, spraying pressure, total mass of paint sprayed from the nozzle, spraying distance, ambient wind speed, paint viscosity, and nozzle travel speed;

[0062] Step S1-5: Substitute the spraying distance, total mass of paint sprayed from the nozzle, ambient wind speed, paint viscosity and nozzle travel speed from each sample in the historical dataset into the coating dry film thickness prediction model without error compensation to obtain the predicted coating dry film thickness, and calculate the difference between the actual coating dry film thickness and the predicted coating dry film thickness as the error compensation term corresponding to each sample in the historical dataset.

[0063] Steps S1-6: Construct the first sample, wherein the error compensation term of each sample in the historical dataset is used as the label value of the first sample, and the spraying pressure, nozzle travel speed, spraying distance, ambient wind speed and paint viscosity of each sample in the historical dataset are used as the input values ​​of the first sample.

[0064] Steps S1-7: Train the error compensation model using the first sample. The inputs to the error compensation are spraying pressure, nozzle travel speed, spraying distance, ambient wind speed, and paint viscosity. The output is the error compensation term.

[0065] Steps S1-8: Combine the coating dry film thickness prediction model without error compensation with the error compensation term to obtain the coating dry film thickness prediction model.

[0066] Based on the correlation model, an error compensation term is introduced, which is obtained by training the machine learning model based on historical data. The error compensation term is used to learn and compensate for dynamic characteristics and non-ideal factors that are not fully represented in the model. The core physical relationship, nozzle volume flow rate sub-model, transmission efficiency coupling function and error compensation term are integrated to form a complete coating dry film thickness prediction model.

[0067] Furthermore, in this embodiment, the total mass of paint sprayed from the nozzle is:

[0068] ;

[0069] in: For flow coefficient, This represents the nozzle orifice area. This refers to the density of the coating.

[0070] As described above, by clarifying the input and output variables of the coating dry film thickness prediction model, defining the core physical relationships, calculating the nozzle volume flow rate, calibrating the transmission efficiency function of multi-factor coupling, and introducing error compensation terms, a coating dry film thickness prediction model that can accurately reflect the coupling effect of multiple factors such as spraying pressure, nozzle travel speed, spraying distance, ambient wind speed, and coating viscosity is constructed.

[0071] The formation of the dry film thickness is the result of the entire process of paint being sprayed from the nozzle, transported to the surface of the barrel-type base structure, and deposited. Significant nonlinear coupling effects exist among various parameters in this process. Spraying pressure directly determines the volumetric flow rate of paint sprayed from the nozzle; higher pressure generally results in higher flow rates. However, the effect of increased flow rates is affected by paint density. The nozzle travel speed determines the time it takes for a unit area of ​​the structure surface to receive paint; faster speeds result in less paint deposited per unit area. Excessive spraying distance leads to increased paint loss during transport. Excessive ambient wind speed can blow away some atomized paint. Paint viscosity that is too high or too low will affect atomization and transport stability. These factors collectively determine the final dry film thickness. If the intrinsic relationship and coupling effects between each parameter and the dry film thickness cannot be clearly defined, the predictive model will not accurately reflect the actual spraying process, leading to subsequent control failures.

[0072] The input variables for the correlation model are defined as spraying pressure, nozzle travel speed, spraying distance, ambient wind speed, and paint viscosity, while the output variable is the dry film thickness. This is because these five input parameters directly govern the entire process of paint spraying, transport, and deposition. Specifically, spraying pressure controls the amount of paint sprayed, travel speed controls the deposition time, distance and wind speed control transport losses, and viscosity controls atomization and flow characteristics. The absence of any one of these parameters will prevent the model from fully reflecting the thickness formation mechanism. Therefore, clearly defining these five input variables is fundamental to ensuring the integrity of the model.

[0073] The basic relationship between coating dry film thickness and nozzle travel speed is defined: the dry film thickness is directly proportional to the volume of paint deposited per unit area and inversely proportional to the dry film density (when the density is constant). Therefore, a larger volume results in a larger thickness. Nozzle travel speed determines the spraying time per unit area; a faster speed results in a smaller volume of paint received per unit area at the same flow rate, leading to a smaller thickness. This fundamental relationship provides a physically sound framework for the model, ensuring that the model does not deviate from the actual spraying process.

[0074] The nozzle volumetric flow rate is calculated based on the nozzle's flow coefficient, orifice area, spraying pressure, and paint density. The volumetric flow rate is directly proportional to the square root of the flow coefficient, orifice area, and spraying pressure, and inversely proportional to the square root of the paint density. Nozzle volumetric flow rate is a quantitative indicator of the amount of paint sprayed. The flow coefficient reflects the degree to which the nozzle's internal structure hinders paint flow; for example, a nozzle with a flow coefficient of 0.9 has a higher paint throughput than a nozzle with a flow coefficient of 0.8. The orifice area directly determines the cross-sectional size through which the paint passes, for example, 1.2 mm. 2 Nozzle ratio 1.0mm 2 A larger nozzle flow rate and spray pressure provide the driving force for paint flow; for example, a pressure of 0.3 MPa provides a stronger driving force than 0.2 MPa. Paint density affects flow resistance; for example, a density of 1.3 g / cm³... 3 The coating ratio is 1.1 g / cm³. 3 The flow resistance is slightly high. Using this calculation method, the volumetric flow rate can accurately quantify the amount of paint sprayed, providing accurate basic data for subsequent thickness calculations.

[0075] The transmission efficiency is defined as a fusion function of spraying distance, ambient wind speed, and paint viscosity, and its parameters are calibrated jointly through computational fluid dynamics simulation and experimental data. Transmission efficiency is the effective utilization rate of paint from the nozzle to the structural surface, characterizing the ratio of the actual paint mass reaching the surface to the sprayed mass. This parameter directly reflects the coupling effect of multiple factors. For example, a spraying distance of 0.25m has a lower transmission efficiency than 0.15m, meaning more paint is lost; an ambient wind speed of 4m / s has a lower transmission efficiency than 2m / s, meaning more atomized paint is blown away; and a paint viscosity of 0.5Pa·s has a lower transmission efficiency than 0.8Pa·s, indicating that the lower viscosity makes it easier for the paint to be dispersed by the wind. In practice, a CFD simulation experiment matrix covering wind speeds of 0-6 m / s, spraying distances of 0.15-0.30 m, and coating viscosities of 0.5-1.0 Pa·s is first constructed using ANSYS Fluent to simulate the transmission process under different working conditions and obtain the simulated transmission efficiency. Then, key working conditions, such as wind speed of 3 m / s, distance of 0.20 m, and viscosity of 0.8 Pa·s, are selected to conduct actual spraying experiments and measure the transmission efficiency. Finally, the simulation model is corrected using experimental data, and the parameters of the fusion function are calibrated to ensure that the transmission efficiency can accurately reflect the coupling effect of multiple factors and avoid errors caused by traditional models ignoring transmission efficiency or subjectively estimating transmission efficiency.

[0076] An error compensation term, trained by a machine learning model based on historical data, is introduced on the basis of the correlation model, and the various parts are integrated to form a complete prediction model. The error compensation term is used to compensate for dynamic characteristics and non-ideal factors not covered by the core physical model, such as slight viscosity fluctuations caused by minor sedimentation of the paint during the spraying process, and subtle changes in the flow coefficient caused by nozzle wear. These factors are difficult to accurately represent through physical formulas. The historical data comes from the datasets of previous CFD simulations and actual spraying experiments. The residual between the measured dry film thickness and the thickness predicted by the physical model is used as the learning target to train the machine learning model to obtain the error compensation term. By integrating the core physical relationship (thickness is directly proportional to volumetric flow rate and transmission efficiency, and inversely proportional to velocity), the nozzle volumetric flow rate sub-model, the transmission efficiency coupling function, and the error compensation term, when a set of parameters is input, the model can first calculate the volumetric flow rate and transmission efficiency, then obtain the basic thickness through the core relationship, and finally superimpose the error compensation term to obtain the final predicted value. This design can significantly improve the prediction accuracy of the model under complex coupled conditions and provide reliable thickness prediction results for subsequent real-time control.

[0077] In this embodiment, CFD simulation experiments in the full parameter space and physical spraying experiments under key working conditions are carried out to obtain transmission efficiency and coating thickness data under multiple working conditions. After data fusion, the data is used to fit the parameters of the transmission efficiency coupling function and train the error compensation model. Finally, a high-precision film thickness prediction model and response surface database are constructed to provide data support and model foundation for real-time prediction and control of anti-corrosion coating thickness of barrel foundation structures, ensuring that the model can cover complex working conditions and the prediction accuracy meets actual needs.

[0078] The accuracy of coating dry film thickness prediction models depends on two aspects: first, a transmission efficiency function that accurately describes the coupling effects of multiple factors; and second, an error compensation model that can compensate for dynamic characteristics not covered by the physical model. Both require a large amount of data covering a wide range of operating conditions. In barrel-type base spraying operations, there are numerous combinations of parameters such as spraying pressure, distance, and wind speed. Obtaining data for all operating conditions solely through physical experiments is not only costly and time-consuming but also difficult to cover extreme conditions, such as a wind speed of 6 m / s and a spraying distance of 0.30 m. While CFD simulation data can efficiently cover all operating conditions, the simulation model may deviate from reality due to simplified assumptions, such as ignoring minor agglomeration of the coating, resulting in insufficient data accuracy. Therefore, a data fusion approach combining simulation and experimental data is needed to obtain data for all operating conditions while ensuring data accuracy. This provides a reliable basis for model parameter fitting and training, thereby ensuring the accuracy of the prediction model.

[0079] Computational fluid dynamics software was used to construct an experimental matrix covering the full range of parameters, including wind speed, spraying distance, and spraying pressure, and simulations were conducted. Specifically, the parameter ranges were set as follows: wind speed 0-6 m / s, spraying distance 0.15-0.30 m, and spraying pressure 0.2-0.4 MPa. The matrix was constructed according to the principle of full factorial experimental design to ensure that each parameter combination was covered. During the simulation, the software's built-in paint atomization model and fluid flow model were used to simulate the atomization process of paint after it was sprayed from the nozzle, its transmission process in the air (paint loss due to wind speed), and its deposition process on the surface of the barrel-type foundation structure under different working conditions. The simulation transmission efficiency data (e.g., 88% transmission efficiency at wind speed 2 m / s, distance 0.20 m, and pressure 0.3 MPa) and the simulation coating thickness data (e.g., corresponding to a thickness of 98 μm) were then output, forming a simulation data pair of input parameters and output results. It can efficiently cover all working conditions, solving the problem that physical experiments are difficult to cover extreme or niche working conditions, and avoiding the failure of model prediction under specific conditions due to missing working conditions.

[0080] Physical spraying experiments were conducted at key operating points in the laboratory or on-site. Key operating points included boundary conditions of parameter ranges (e.g., wind speed 6 m / s, distance 0.30 m, pressure 0.2 MPa), commonly used operating conditions (e.g., wind speed 2-3 m / s, distance 0.20-0.25 m, pressure 0.3 MPa), and operating conditions where simulation and reality may have large deviations (e.g., wind speed 4 m / s, distance 0.15 m, pressure 0.4 MPa). During the experiment, an ultrasonic thickness gauge was used to directly measure the dry film thickness of the coating. At the same time, the corresponding spraying pressure (e.g., 0.3 MPa), nozzle travel speed (e.g., 0.5 m / s), spraying distance (e.g., 0.20 m), and ambient wind speed (e.g., 2 m / s) parameters were recorded synchronously through sensors to form experimental data pairs (e.g., when the input parameters are pressure 0.3 MPa, speed 0.5 m / s, distance 0.20 m, wind speed 2 m / s, and viscosity 0.8 Pa·s, the measured thickness is 95 μm). By acquiring data under real-world operating conditions, we can provide a basis for verifying and correcting CFD simulation models, avoid systematic errors caused by simplification assumptions in the simulation models, and ensure the authenticity of subsequent data.

[0081] In some embodiments, CFD simulation models are corrected using physical experimental data. For example, under a certain critical operating condition, the simulated coating thickness is 100 μm, while the experimentally measured thickness is 95 μm. In this case, parameters such as the coating atomization coefficient and air resistance coefficient in the CFD model need to be adjusted, and the simulation is repeated until the deviation between the simulation results and the experimental results is less than 5%. After correction, all corrected simulation data (covering all operating conditions) is merged with physical experimental data (covering the critical operating conditions) to form a comprehensive dataset. Through data fusion, the advantage of full operating condition coverage of simulation data is preserved, while the authenticity of experimental data is incorporated. This solves the limitations of a single data source. Using only simulation data can lead to systematic errors, while using only experimental data is insufficient. The comprehensive dataset can provide effective data support for subsequent model training and parameter fitting, thereby improving model accuracy.

[0082] Model parameter fitting and error compensation model training were conducted based on a comprehensive dataset. For the transmission efficiency coupling function, least squares and other parameter fitting methods were employed. The spraying distance D and ambient wind speed U from the comprehensive dataset were used as parameters. w Substituting the viscosity μ and the corresponding transport efficiency into the function, specific parameter values ​​such as a, b, c, and d are calculated to minimize the deviation between the function calculation result and the TE value in the dataset. For the error compensation model, machine learning algorithms such as random forest or neural networks are used, with spraying pressure, nozzle travel speed, spraying distance, ambient wind speed, and paint viscosity in the comprehensive dataset as input features, and the residuals of the measured dry film thickness and the predicted thickness based on the physical model (including the fitted parameters of the transport efficiency function) as output labels. The model is trained using the training set of the dataset, and the model accuracy is verified using the test set until the residual prediction error meets the requirements. This can transform the data into the core of a usable model. Fitting the transport efficiency parameters makes the coupling function have practical application value, and training the error compensation model can compensate for non-ideal factors not covered by the physical model (such as local viscosity changes caused by slight paint sedimentation). The combination of the two forms a high-precision film thickness prediction model, ensuring that the model can accurately predict the thickness under different working conditions. The calculation formula of the high-precision film thickness prediction model is as follows:

[0083] For the error compensation term, a machine learning model is trained based on historical data to learn unmodeled dynamic characteristics. A response surface database is constructed using a high-precision film thickness prediction model. Within a predefined parameter range, intensive calculations are performed at fixed step sizes to obtain the predicted dry film thickness of the coating for each parameter combination. For parameter combinations between step sizes, linear or cubic interpolation methods are used to complete the predicted thickness data, ultimately forming a response surface database with spraying pressure, nozzle travel speed, spraying distance, and ambient wind speed as input indices and predicted dry film thickness as output. The core function of the database is to ensure the timeliness of real-time control. During real-time control, the feedforward controller needs to output the baseline values ​​of spraying parameters immediately. Directly querying the database can obtain results within milliseconds, which is much faster than real-time model calculations. This avoids parameter adjustment delays caused by computation time and ensures timely compensation when parameters such as wind speed and distance change, maintaining stable coating thickness.

[0084] Step S2: Obtain real-time ambient wind speed and spraying distance. Combined with pre-configured flow coefficient, nozzle orifice area, paint density, and paint viscosity, query the pre-configured response surface database to obtain the baseline spraying pressure and baseline nozzle travel speed, including:

[0085] Step S2-1: Obtain real-time ambient wind speed and spraying distance;

[0086] Step S2-2: Discretize the future time period into multiple time periods, and obtain the environmental wind speed for the future multiple time periods based on the real-time environmental wind speed sequence of the current and past multiple time periods;

[0087] Step S2-3: Combine the current and future ambient wind speeds with the spraying distance, as well as the pre-configured flow coefficient, nozzle orifice area, paint density, and paint viscosity, and query the pre-configured response surface database to obtain the reference spraying pressure and reference nozzle travel speed corresponding to each future time period.

[0088] Step S3: Based on the coating dry film thickness prediction model, to minimize the film thickness control deviation and the fluctuation targets of spraying pressure and nozzle travel speed over a future period, obtain the spraying pressure correction sequence and the nozzle travel speed correction sequence, including:

[0089] Step S3-1: Initialize the correction sequence, where each element in the correction sequence corresponds to the spraying pressure correction and nozzle travel speed correction for each future time period.

[0090] Step S3-2: Construct an objective function by minimizing the sum of film thickness control deviations over all time periods and the sum of fluctuations in spraying pressure and nozzle travel speed. Iteratively update the correction sequence to obtain the optimal correction sequence, and then analyze the optimal correction sequence into a spraying pressure correction sequence and a nozzle travel speed correction sequence.

[0091] The process for obtaining the film thickness control deviation for a single time period in step S3-2 is as follows:

[0092] The correction sequence is analyzed to obtain the spray pressure correction and nozzle travel speed correction for a single time period. The obtained spray pressure correction and nozzle travel speed correction are then superimposed on the corresponding reference spray pressure and reference nozzle travel speed, respectively.

[0093] The superimposed spraying pressure and nozzle travel speed, combined with ambient wind speed, spraying distance, and pre-configured flow coefficient, nozzle orifice area, coating density, and coating viscosity, are substituted into the coating dry film thickness prediction model to obtain the first predicted film thickness. The difference between the target dry film thickness and the first predicted film thickness is calculated as the film thickness control deviation.

[0094] Step S4: Superimpose the first element of each of the spray pressure correction sequence and the nozzle travel speed correction sequence onto the corresponding reference spray pressure and reference nozzle travel speed to obtain the target spray pressure and target nozzle travel speed.

[0095] By deploying multiple types of sensors on intelligent spraying equipment to collect environmental and equipment status data in real time, a feedforward-feedback composite control architecture is constructed. Combined with response surface database and model predictive control algorithm, the spraying pressure and nozzle travel speed are dynamically adjusted to achieve precise control of the anti-corrosion coating thickness of the barrel-type foundation structure, ensuring that the coating thickness is uniform and consistently meets the standards in the complex and variable marine environment.

[0096] Corrosion protection spraying of barrel foundation structures is carried out at high altitudes and in marine environments with frequent wind speed fluctuations. The spraying distance changes in real time due to the geometry of the barrel foundation (such as the curvature of the barrel wall and the difference in diameter between the top and bottom). These disturbances directly affect the transport and deposition process of the coating. When the ambient wind speed suddenly increases, the amount of coating lost during transport increases. If the spraying pressure is not adjusted in time, the amount of coating deposited per unit area will decrease, resulting in a thinner coating thickness. When the spraying distance increases, the effective deposition amount of coating reaching the surface of the structure decreases. If the nozzle travel speed is not adjusted synchronously, it will also result in insufficient thickness. At the same time, real-time data on the nozzle position can help confirm the work area and avoid repeated spraying or missed spraying.

[0097] An ultrasonic anemometer, laser rangefinder, and displacement encoder are integrated into the intelligent spraying equipment, and a programmable logic controller (PLC) is used to achieve synchronous data acquisition and preprocessing. The ultrasonic anemometer is specifically used to measure the real-time ambient wind speed at the spraying point, accurately capturing instantaneous changes in wind speed. The laser rangefinder is used to measure the instantaneous spraying distance between the nozzle and the surface of the barrel-type base structure, such as the distance increasing from 0.20m to 0.24m due to changes in the barrel wall curvature. The displacement encoder provides real-time feedback on the current position of the nozzle along the barrel-type base structure. The PLC synchronously acquires signals from the three types of sensors according to a preset acquisition cycle (e.g., once every 100 milliseconds) and preprocesses the data, such as filtering high-frequency noise in the wind speed signal and calibrating the measurement deviation of the laser rangefinder. Without this step of real-time data acquisition, the feedforward control will not be able to obtain disturbance information, and the feedback control will lack current operating parameters. Subsequent parameter adjustments will lack a basis and will be unable to cope with real-time changes in the environment and equipment status.

[0098] A system architecture combining feedforward and feedback control is established. Feedforward control queries the response surface database based on real-time sensor data to output baseline parameters. The feedforward control channel directly calls the real-time collected environmental wind speed and spraying distance data, using these two parameters as input indices to quickly query the constructed response surface database. For example, when the real-time wind speed is 3.5 m / s and the spraying distance is 0.24 m, the response surface database calculates the corresponding baseline spraying pressure of 0.35 MPa and the baseline nozzle travel speed of 0.42 m / s through internal interpolation. The core function of these baseline values ​​is to pre-compensate for the effects of measurable disturbances. For example, if increased wind speed leads to increased paint loss, feedforward control increases the spraying pressure baseline value to increase the paint output and offset the thickness loss caused by loss. If increased spraying distance leads to reduced deposition, feedforward control increases the travel speed baseline value to extend the spraying time per unit area and compensate for insufficient deposition. This pre-compensation method avoids the lag problem of traditional single feedback control, which corrects deviations after they occur, and significantly reduces the generation of defective coatings.

[0099] The feedback control channel employs a model predictive control algorithm, using a high-precision film thickness prediction model as the internal model to calculate film thickness deviation. Within each control cycle, the model predictive controller receives current spraying pressure, nozzle travel speed, spraying distance, and ambient wind speed data, inputting these parameters into the constructed high-precision film thickness prediction model to calculate the predicted dry film thickness under the current operating conditions in real time. Assuming a preset target film thickness of 100 μm and a predicted film thickness of 96 μm, the calculated film thickness deviation is 4 μm. The advantage of the model predictive control algorithm lies in its ability to predict film thickness trends over multiple future control cycles (e.g., the next 5 cycles, totaling 500 milliseconds). For example, if the wind speed is predicted to remain at 3.5 m / s in the next cycle, maintaining the current parameters would result in a predicted thickness of 96 μm, thus providing a forward-looking basis for subsequent correction calculations and avoiding subsequent deviation expansion caused by short-sighted adjustments.

[0100] The reference value output from the feedforward is algebraically superimposed with the correction amount calculated from the feedback to generate the final control command, which is then applied to the spraying system. For example, based on a film thickness deviation of 4μm, the model predictive controller calculates a spraying pressure correction of +0.02MPa and a nozzle travel speed correction of -0.01m / s. The feedforward pressure reference value of 0.35MPa is superimposed with the correction amount +0.02MPa to obtain the final target spraying pressure of 0.37MPa. Similarly, the feedforward speed reference value of 0.42m / s is superimposed with the correction amount -0.01m / s to obtain the final target nozzle travel speed of 0.41m / s. These control commands are applied to the spraying system through actuators (such as spraying pressure regulating valves and nozzle travel motors) to adjust process parameters in real time. It achieves synergy between feedforward and feedback. Feedforward solves the impact of most measurable disturbances, while feedback corrects the residual deviation after feedforward compensation (such as prediction deviation caused by small changes in coating viscosity). The combination of the two forms a closed-loop dynamic adjustment, ensuring that the dry film thickness of the coating is always stable near the target value, avoiding the thickness fluctuation problem caused by the single control method and untimely correction in traditional control methods.

[0101] In barrel-type base spraying operations, ambient wind speed and spraying distance are the core measurable disturbances affecting coating thickness. Increased wind speed leads to increased paint loss during transport. If the original spraying pressure is maintained, the amount of paint deposited per unit area will decrease, resulting in a thinner coating. Increasing the spraying distance reduces the effective deposition density of paint on the structural surface. Without adjusting the nozzle travel speed, insufficient thickness will also occur. Feedforward control can output adapted reference parameters in advance based on real-time disturbance data to offset most of the disturbance effects. However, in actual operations, there are still non-ideal factors, such as fluctuations in paint viscosity due to slight changes in ambient temperature and flow deviations caused by slight nozzle wear. These factors are difficult to compensate for in advance through feedforward, resulting in residual deviations between the actual thickness after feedforward and the target. Therefore, feedback control is needed to calculate and correct these deviations in real time to achieve accurate thickness control under full disturbance.

[0102] The feedforward controller continuously receives real-time ambient wind speed and spraying distance signals. Using these two sets of data as input indexes, it queries the response surface database and outputs the baseline parameters required to maintain the target film thickness. The response surface database is constructed based on a high-precision film thickness prediction model through intensive calculations across the entire parameter range. It covers parameter combinations and their corresponding target thickness requirements for spraying pressure (0.2-0.4 MPa), nozzle travel speed (0.3-0.8 m / s), spraying distance (0.15-0.30 m), and ambient wind speed (0-6 m / s). For example, when the ultrasonic anemometer measures the ambient wind speed to be 3.2 m / s and the laser rangefinder measures the spraying distance to be 0.23 m, the feedforward controller queries the database using the wind speed of 3.2 m / s and the distance of 0.23 m as indexes. The database then performs internal interpolation calculations (such as interpolation based on reference parameters corresponding to adjacent wind speeds of 3 m / s and 3.4 m / s and distances of 0.22 m and 0.24 m) to output a reference value for spraying pressure of 0.34 MPa and a reference value for nozzle travel speed of 0.46 m / s. This allows for the rapid output of reference parameters adapted to the current disturbance, avoiding compensation lag caused by the time spent on parameter calculations.

[0103] In the model predictive control stage, each control cycle (preset to 100 milliseconds) invokes the constructed high-precision film thickness prediction model, inputting the actual parameters at the current moment, including the spraying pressure (0.34 MPa) and nozzle travel speed (0.46 m / s) output by the feedforward, as well as the spraying distance (0.23 m), ambient wind speed (3.2 m / s), and coating viscosity (0.8 Pa·s, collected by a preset viscosity sensor) acquired in real time by the sensor. The model calculates the predicted dry film thickness of the coating under the current operating conditions based on these parameters. It can capture the residual deviation after feedforward compensation in real time, providing a basis for subsequent corrections. The actual parameters are all from real-time data synchronously collected by the sensor, ensuring that the prediction deviation accurately reflects the current thickness status.

[0104] The model predictive controller aims to minimize film thickness control deviation and control quantity fluctuations over a future period. It constructs an optimization problem and solves for the correction values. The future time window is set to 5 control cycles (500 milliseconds in total). Based on current sensor data, it predicts the operating condition trends over the next 500 milliseconds. For example, based on the wind speed change rate, it predicts that the wind speed will remain between 3.1 and 3.3 m / s over the next 5 cycles, and the spraying distance will remain between 0.22 and 0.24 m due to changes in the barrel wall curvature. The decision variables of the optimization problem are the spraying pressure correction sequence and the nozzle travel speed correction sequence over the next 5 cycles. The objective function is the sum of squares of the film thickness deviation over the next 5 cycles plus the sum of squares of the control quantity correction amplitude (weighted coefficients are set to 0.8 and 0.2 respectively, prioritizing thickness accuracy while avoiding abrupt changes in control values). The constraints are a spraying pressure correction range of ±0.05 MPa and a travel speed correction range of ±0.03 m / s (to avoid exceeding the equipment's adjustment capacity). By solving this quadratic programming problem online, the optimal instantaneous correction for the spraying pressure at the current moment (first cycle) is obtained as +0.03 MPa, and the instantaneous correction for the nozzle travel speed is -0.02 m / s. By predicting future operating conditions and optimizing the correction amounts, the problem of adjustment lag caused by traditional feedback that only considers the present and ignores the future is avoided. For example, by considering the operating conditions of stable wind speed in the future, the current correction amount will not be over-adjusted, ensuring that the thickness remains stable in subsequent cycles.

[0105] The reference value output from the feedforward is algebraically superimposed with the real-time correction value obtained from the MPC solution to obtain the final control command, which is then issued and executed: the target value of the spraying pressure = 0.34MPa + 0.03MPa = 0.37MPa, and the target value of the nozzle travel speed = 0.46m / s - 0.02m / s = 0.44m / s. These control commands act on the spraying system through the actuators (spraying pressure regulating valve and nozzle travel motor) to adjust the process parameters in real time. In actual operation, the adjusted parameters increase the amount of paint sprayed (increased pressure) and extend the spraying time per unit area (decreased speed), jointly compensating for the 3.5μm film thickness deviation. The final dry film thickness of the coating is stabilized at around 99.8μm, with the error controlled within 0.2μm. This achieves synergistic optimization of feedforward and feedback, with feedforward quickly compensating for major disturbances and feedback accurately correcting residual deviations, forming a complete closed-loop control. This ensures that the coating thickness always meets the target requirements when the environment and equipment conditions change dynamically, avoiding the problems of large thickness fluctuations and many unqualified coatings under traditional control methods.

[0106] By setting both periodic and event-based calibration trigger conditions, offline sampling measurements are carried out during the spraying operation. The error between the measured film thickness and the model prediction value is compared. When the error exceeds the limit, the parameters of the high-precision film thickness prediction model are corrected and the response surface database is updated. This ensures that the model maintains high prediction accuracy throughout the long-term operation, providing continuous and reliable model support for precise control of coating thickness.

[0107] Corrosion protection spraying operations on barrel-type foundation structures are lengthy, and the marine environment is dynamically changing. During long-term operation, various factors can lead to a decrease in model accuracy. The viscosity of the coating may fluctuate slightly due to changes in ambient temperature (such as daytime and nighttime temperature differences) during storage and use. The nozzles of the intelligent spraying equipment may experience slight wear over time, causing deviations between the actual flow coefficient and the initial calibration value. Sensors may experience zero-point drift during long-term operation, resulting in errors in the collected environmental wind speed data. These factors can cause the initially constructed high-precision film thickness prediction model to gradually deviate from the actual spraying process. If not calibrated in time, subsequent control based on this model will lead to excessive coating thickness deviations. Therefore, regular or event-triggered calibration is necessary to correct the model parameters to match the actual operating conditions.

[0108] The calibration process is initiated under specific conditions, including periodic and event-triggered conditions. Periodic triggering is based on the typical accuracy decay pattern of spraying operations, with fixed time intervals, such as every 2 hours, to ensure that model accuracy does not accumulate deviations over long-term operation. Event-triggered conditions address sudden events that may cause a sharp drop in model accuracy. Preset trigger thresholds are used: a sudden change in ambient wind speed of 2 m / s (i.e., a wind speed change exceeding 2 m / s within a short period, such as a sudden increase from 3 m / s to 5 m / s), and a significant change in spraying distance of 0.05 m (i.e., a sudden change in spraying distance from 0.20 m to 0.25 m due to a sudden change in the geometry of the barrel-type foundation structure). When the data collected by the sensors meets either event trigger condition, calibration is initiated immediately. This approach achieves both proactive prevention and reactive response, avoiding the lag of fixed-period calibration and the omissions caused by the lack of event triggers, ensuring that any decrease in model accuracy is detected and addressed promptly.

[0109] When the calibration trigger conditions are met, measurements are taken on the surface of the coated barrel-shaped foundation according to a predetermined sampling plan. The sampling plan needs to cover different work areas. For example, two measurement points are selected at the top, middle, and bottom of the barrel-shaped foundation, for a total of six sampling points, to ensure that the sampling data can reflect the overall coating quality. During the measurement, an ultrasonic thickness gauge is used to directly measure the dry film thickness of the coating at each sampling point. At the same time, the location information of each measurement point and the corresponding measured film thickness value are recorded to avoid calibration deviations caused by relying solely on model prediction data.

[0110] For each sampling measurement point, the model predicts the film thickness and error. The actual process and environmental parameters during spraying at each sampling point are stored in real-time in the control system's database. For example, the parameters for a certain sampling point are: spraying pressure 0.32 MPa, nozzle travel speed 0.48 m / s, spraying distance 0.22 m, ambient wind speed 2.8 m / s, and coating viscosity 0.82 Pa·s. These parameters are input into the current high-precision film thickness prediction model to calculate the model's predicted film thickness value for that point. Then, the absolute error (the absolute value of the difference between the measured and predicted values) for each sampling point is calculated, and the average and maximum absolute errors for all sampling points are statistically analyzed. By quantifying the deviation between the model's predicted and actual values, a clear numerical basis is provided for determining whether calibration is needed, avoiding over- or under-calibration due to subjective judgment.

[0111] Determine if the error exceeds the limit and perform model calibration. The preset first threshold (mean absolute error threshold) is 3 μm, and the second threshold (maximum absolute error threshold) is 5 μm. When the mean absolute error exceeds the first threshold (3.5 μm), or the maximum absolute error reaches the second threshold (5 μm), model calibration is required. During calibration, the error compensation model is corrected first. The actual parameters of this sampling measurement minus the measured film thickness data pairs (6 sets in total) are used as incremental learning data and input into the machine learning-based error compensation model (such as a random forest model). The decision tree weights of the model are fine-tuned through online incremental learning, enabling the model to learn the error patterns caused by recent changes in coating viscosity and nozzle wear. For example, the error compensation term that originally output +2 μm is adjusted to +3 μm. If the error still does not fall within the threshold after incremental learning (e.g., the mean absolute error remains at 3.2 μm), the parameters of the transmission efficiency coupling function are further recalibrated. For example, the coefficient related to wind speed in the transmission efficiency function is adjusted from 0.8 to 0.85 to ensure that the coupling function accurately reflects the impact of the current wind speed on coating transmission. Layered calibration ensures both calibration efficiency and calibration depth, avoiding model instability caused by directly adjusting the core coupling function.

[0112] The real-time model and response surface database are updated synchronously. The new model parameters obtained after calibration are updated into the real-time high-precision film thickness prediction model to ensure that subsequent feedback control uses the latest model. Simultaneously, based on the updated high-precision film thickness prediction model, local intensive calculations are re-performed within predefined parameter ranges. For example, for parameter ranges with large errors identified in this calibration, such as wind speed 2.5-3.5 m / s and spraying distance 0.20-0.25 m, response surface data for these ranges is regenerated and replaced with the corresponding parts in the original database. This ensures that the response surface database queried by the feedforward control is consistent with the updated model. This step achieves synchronous updates between the model and the database, avoiding contradictions between feedforward and feedback control due to inconsistencies, and ensuring that the calibrated control effect is fully realized.

[0113] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for controlling the thickness of anti-corrosion coatings on barrel-type foundation structures considering the coupling effect of multiple factors, characterized in that, include: Step S1: Construct a coating dry film thickness prediction model based on simulation dataset and historical dataset. The input of the coating dry film thickness prediction model includes flow coefficient, nozzle orifice area, spraying pressure, coating density, nozzle travel speed, spraying distance, ambient wind speed and coating viscosity. The output is the coating dry film thickness. Step S2: Obtain real-time ambient wind speed and spraying distance, and combine them with pre-configured flow coefficient, nozzle orifice area, paint density and paint viscosity, and query the pre-configured response surface database to obtain the benchmark spraying pressure and benchmark nozzle travel speed. Step S3: Based on the coating dry film thickness prediction model, in order to minimize the film thickness control deviation and the fluctuation targets of spraying pressure and nozzle travel speed in the future, the spraying pressure correction sequence and the nozzle travel speed correction sequence are obtained. Step S4: Superimpose the first element of each of the spray pressure correction sequence and the nozzle travel speed correction sequence onto the corresponding reference spray pressure and reference nozzle travel speed to obtain the target spray pressure and target nozzle travel speed; The mathematical expression for the coating dry film thickness prediction model is: ; in: DFT For the dry film thickness of the coating, K The volume solids content and density constant of the coating are given. The total mass of paint sprayed from the nozzle. For transmission efficiency, v The nozzle travel speed, For error compensation, P For spraying pressure, D For spraying distance, U w For ambient wind speed, μ The viscosity of the coating; Step S2 includes: Step S2-1: Obtain real-time ambient wind speed and spraying distance; Step S2-2: Discretize the future time period into multiple time periods, and obtain the environmental wind speed for the future multiple time periods based on the real-time environmental wind speed sequence of the current and past multiple time periods; Step S2-3: Combine the current and future ambient wind speeds with the spraying distance, as well as the pre-configured flow coefficient, nozzle orifice area, paint density, and paint viscosity, and query the pre-configured response surface database to obtain the reference spraying pressure and reference nozzle travel speed corresponding to each future time period. Step S3 includes: Step S3-1: Initialize the correction amount sequence, wherein each element in the correction amount sequence corresponds to the spraying pressure correction amount and nozzle travel speed correction amount for each future time period; Step S3-2: Construct an objective function by minimizing the sum of film thickness control deviations and the sum of fluctuations in spraying pressure and nozzle travel speed over all time periods, iteratively update the correction sequence to obtain the optimal correction sequence, and then parse the optimal correction sequence into a spraying pressure correction sequence and a nozzle travel speed correction sequence. The process of obtaining the film thickness control deviation for a single time period in step S3-2 is as follows: The correction sequence is analyzed to obtain the spray pressure correction and nozzle travel speed correction for a single time period. The obtained spray pressure correction and nozzle travel speed correction are then superimposed on the corresponding reference spray pressure and reference nozzle travel speed, respectively. The superimposed spraying pressure and nozzle travel speed, combined with ambient wind speed, spraying distance, and pre-configured flow coefficient, nozzle orifice area, coating density, and coating viscosity, are substituted into the coating dry film thickness prediction model to obtain the first predicted film thickness. The difference between the target dry film thickness and the first predicted film thickness is calculated as the film thickness control deviation.

2. The method for controlling the thickness of the anti-corrosion coating of a barrel-type foundation structure considering the coupling effect of multiple factors, as described in claim 1, is characterized in that... The transmission efficiency is specifically: ; in, a This is the coefficient for the spraying distance term. b This is the coefficient for the environmental wind speed term. c The coefficient for the viscosity term of the coating. d The coefficient of the zero-degree term.

3. The method for controlling the thickness of the anti-corrosion coating of a barrel-type foundation structure considering the coupling effect of multiple factors, as described in claim 2, is characterized in that... Step S1 includes: Step S1-1: Eliminate errors through simulation experiments to obtain a simulation dataset, wherein each sample of the simulation dataset includes at least the coating dry film thickness, the total mass of paint sprayed from the nozzle, the spraying distance, the ambient wind speed, the paint viscosity, and the nozzle travel speed; Step S1-2: Based on the coating dry film thickness, total mass of paint sprayed from the nozzle and nozzle travel speed in each sample of the simulation dataset, the transmission efficiency of each sample is calculated by combining the paint volume solids content and density constant. Step S1-3: Based on the spraying distance, ambient wind speed and coating viscosity in each sample of the simulation dataset, the coefficients of the spraying distance term, ambient wind speed term, coating viscosity term and zero-order term are obtained by fitting the transmission efficiency, and a coating dry film thickness prediction model without error compensation is obtained. Step S1-4: Obtain historical datasets, wherein each sample in the historical datasets includes at least the actual dry film thickness of the coating, spraying pressure, total mass of paint sprayed from the nozzle, spraying distance, ambient wind speed, paint viscosity, and nozzle travel speed; Step S1-5: Substitute the spraying distance, total mass of paint sprayed from the nozzle, ambient wind speed, paint viscosity and nozzle travel speed from each sample in the historical dataset into the coating dry film thickness prediction model without error compensation to obtain the predicted coating dry film thickness, and calculate the difference between the actual coating dry film thickness and the predicted coating dry film thickness as the error compensation term corresponding to each sample in the historical dataset. Steps S1-6: Construct the first sample, wherein the error compensation term of each sample in the historical dataset is used as the label value of the first sample, and the spraying pressure, nozzle travel speed, spraying distance, ambient wind speed and paint viscosity of each sample in the historical dataset are used as the input values ​​of the first sample. Steps S1-7: Train the error compensation model using the first sample, wherein the inputs to the error compensation are spraying pressure, nozzle travel speed, spraying distance, ambient wind speed and paint viscosity, and the output is the error compensation term; Steps S1-8: Combine the coating dry film thickness prediction model without error compensation with the error compensation term to obtain the coating dry film thickness prediction model.

4. The method for controlling the thickness of the anti-corrosion coating of a barrel-type foundation structure considering the coupling effect of multiple factors, as described in claim 1, is characterized in that... The total mass of the paint sprayed from the nozzle is: ; in: For flow coefficient, This represents the nozzle orifice area. This refers to the density of the coating.

5. A device for controlling the thickness of an anti-corrosion coating on a barrel-type foundation structure considering the coupling effect of multiple factors, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-4.

6. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the method as described in any one of claims 1-4.

Images