A hot galvanizing coating thickness prediction method based on random forest algorithm

Abstract

The application discloses a hot galvanizing coating thickness prediction method based on a random forest algorithm, relates to the technical field of hot galvanizing steel coil production, and comprises the following steps: producing hot galvanizing steel coils through a hot galvanizing process, and acquiring historical data of the hot galvanizing process; pre-processing the historical data and eliminating the time lag of a thickness gauge; performing feature screening based on a mechanism and a correlation coefficient; establishing a random forest network coating thickness prediction model and training; verifying and judging the accuracy and applicability of the random forest network coating thickness prediction model; if the prediction accuracy of the model does not meet the requirement, optimizing the prediction model parameters; if the requirement is met, putting the prediction model into use; and outputting a zinc layer thickness prediction value. The application improves the time lag problem in detection in zinc layer thickness control of galvanized sheets, and optimizes process parameter adjustment and thickness target control.

Description

Technical Field

[0001] This invention relates to the field of hot-dip galvanized steel coil production technology, and in particular to a method for predicting the thickness of hot-dip galvanized coatings based on a random forest algorithm. Background Technology

[0002] Galvanized steel sheet refers to steel sheet with a layer of zinc coated on its surface. It is widely used in automobile manufacturing, refrigeration, construction and heating facilities. The important technical indicators for measuring the quality of galvanized steel sheet include the thickness and uniformity of the zinc coating.

[0003] This type of hot-dip galvanized steel coil mainly utilizes cold-rolled steel coil raw materials, which are uncoiled, heated in an annealing furnace, dipped into a zinc pot to absorb molten zinc, and have excess molten zinc scraped off the surface using an air knife to control the zinc layer thickness. After cooling, the final zinc layer thickness is obtained. In continuous hot-dip galvanizing production, the online measurement of the coating thickness is accomplished using a thickness gauge. However, due to the requirements of the galvanizing process, the thickness gauge is installed at a considerable distance from the air knife, resulting in significant measurement lag in the system. Furthermore, in actual galvanizing production, the production line speed changes in real time, causing the system measurement lag time to also change in real time. Directly using the thickness gauge information for coating thickness feedback closed-loop control results in significant overshoot and long settling times, further increasing the difficulty of thickness control in galvanizing production and leading to large fluctuations in coating thickness. The galvanized sheet production process is affected by multiple complex physical factors such as aerodynamics and boundary laminar flow. The final zinc layer thickness is the result of the combined effects of multiple variables, including air knife pressure, air knife distance, and production line speed. The influence of these variables on coating thickness is complex and involves many disturbances, exhibiting strong nonlinearity. Meanwhile, in the actual galvanizing industrial production process, there are a variety of external disturbances, such as the composition and temperature of the zinc bath, the specifications of the strip steel, the shape and temperature of the strip steel, etc. These external disturbances will affect the coating thickness and cause large fluctuations in coating quality.

[0004] In summary, galvanized steel sheet production is a typical time-varying, large-lag, nonlinear, and highly perturbed process. Therefore, establishing a predictive model for zinc coating thickness is crucial. However, traditional mechanistic modeling often fails to create an accurate system model, resulting in inaccurate prediction of coating thickness. This leads to a lack of timely and effective control over the coating thickness, causing product quality issues. Summary of the Invention

[0005] To address the problems in existing technologies, this invention provides a method for predicting the thickness of hot-dip galvanized coatings based on a random forest algorithm; it can effectively improve the problem of lag in thickness gauge detection after adjusting air knife process parameters in the hot-dip galvanizing process.

[0006] To achieve the objectives of this invention, the following solution is adopted:

[0007] A method for predicting the thickness of hot-dip galvanized coatings based on a random forest algorithm includes the following steps: Step 1: Producing hot-dip galvanized steel coils through a hot-dip galvanizing process and acquiring historical production data; Step 2: Preprocessing the historical data to eliminate the time lag of the thickness gauge; Step 3: Performing feature filtering on the preprocessed data using mechanistic analysis and correlation coefficients; Step 4: Establishing a random forest network coating thickness prediction model, inputting historical production data into the random forest network coating thickness prediction model, and training the random forest network coating thickness prediction model; Step 5: Verifying and evaluating the accuracy and applicability of the random forest network coating thickness prediction model. Step 6: If the prediction accuracy of the random forest network coating thickness prediction model does not meet the requirements, optimize the parameters of the random forest network coating thickness prediction model and further train the random forest network coating thickness prediction model; if the prediction accuracy of the random forest network coating thickness prediction model meets the requirements, put the random forest network coating thickness prediction model into use; Step 7: Input the feature parameters selected in step 3 into the random forest network coating thickness prediction model and output the zinc layer thickness prediction value; among which, the feature parameters selected in step 3 include air knife pressure, air knife distance, air knife angle, air knife height, strip speed, strip thickness and zinc liquid temperature.

[0008] Based on the above technical solution, further, in step 1, the historical data includes at least: production specifications and control target data: steel grade, coating type, strip thickness, strip width, target coating thickness; parameter data controlled by process personnel: air knife pressure, air knife distance, air knife height; various working conditions, environmental parameters, and relevant parameter data of the previous process flow before the steel plate reaches the air knife: zinc liquid temperature, strip speed, nitrogen pressure, nitrogen flow rate, actual value of straightening roller, strip tension, and air knife angle.

[0009] Based on the above technical solution, further, in step 2, the preprocessing includes handling outliers. The processing procedure is as follows: when the operator is calibrating, the data with a thickness gauge value of 0 is removed; the data with a coating thickness setting that does not match the actual thickness are deleted; when the coating specification is about to be changed, the operator removes the data with the same zinc layer thickness at the tail of the current steel coil as the next steel coil when adjusting the air knife parameters in advance.

[0010] Based on the above technical solution, further, in step 2, the preprocessing includes the response time lag processing after the air knife parameters are set. The processing process is as follows: During the process of setting and adjusting the air knife parameters, the data measured by the thickness gauge is difficult to accurately match the corresponding process parameters. For each record in the dataset of the steel coil, first, it is judged whether the air knife pressure and air knife distance parameters of the steel coil are adjusted, and the deviation between the set value and the actual value is calculated. 1. For the air knife pressure adjustment: For each record in the dataset, subtract the actual value of the air knife pressure from the set value of the air knife pressure to obtain ΔP. By statistically calculating the data of the stable process, the standard deviation of the actual air knife pressure is obtained as;

[0011] ΔP = P Set - P Act ; In the formula, P Set is the set value of the air knife pressure, and P Act is the actual value of the air knife pressure;

[0012] If |ΔP| < σ mbar, it is determined as a record of stable process parameters and the data is retained; if |ΔP| > σ mbar, it is determined as a record of non - stable process parameters and the data is excluded;

[0013] 2. When it is detected that the air knife distance parameter in the record entry is adjusted, no corresponding time lag processing is performed.

[0014] Based on the above technical solution, further, in step 2, the preprocessing includes the detection time lag processing of the thickness gauge. The processing process is as follows:

[0015] The thickness gauge measures the thickness of the zinc coating at a distance of d meters before the air knife; it is necessary to retrieve the thickness of the zinc coating controlled by the air knife process parameters through the corresponding space - time conversion algorithm; [[ID=२५]]

[0016] First, set the speed of the i - th record as v i , define the relative displacement distance as S, with the unit of m. Then, when it comes to the (i + j) - th record, the advancing distance of the steel coil position corresponding to the i - th record is: In the formula, v k is the speed of the k - th record, where k takes values in [i, i + j]; T is the time, with the unit of s;

[0017] Second, set the actual displacement of the thickness gauge lagging behind the air knife as d meters. Then, when the (i + j) - th record satisfies d < S < d + 5, the thickness gauge value of the (i + j) - th record is used as the thickness of the zinc coating controlled by the air knife process parameters corresponding to the i - th record. At this time, d + 5 meters is used as the maximum allowable relative displacement value; if no (i + j) - th record satisfying the condition of d < S < d + 5 can be retrieved through the loop, the i - th data record is excluded.

[0018] Based on the above technical solution, further, in step 3, the mechanism analysis process is as follows: Fluid analysis is performed on the air knife blowing process of the strip steel, and the formula for the coating thickness can be obtained according to the Stokes equation:

[0019]

[0020] In the formula, W is the coating thickness; ρ is the zinc liquid density; μ is the zinc liquid viscosity, which is related to the zinc liquid temperature; V is the strip speed; τ is the shear stress on the zinc liquid surface; p is the air knife pressure; and g is the acceleration due to gravity.

[0021] Based on the above technical solution, further, in step 3, the process of feature selection using correlation coefficients is as follows: The Spearman correlation coefficient ρ1 between the data is calculated as follows:

[0022] In the formula, x i This refers to the sample values ​​of a certain variable in the preprocessed dataset from step 2. This is the sample mean of the variable; similarly, Y i This refers to the sample values ​​of another variable in the preprocessed dataset from step 2. Let be the sample mean of another variable.

[0023] Based on the above technical solution, further, in step 4, the training process is as follows: Data preparation: the data is divided into training set and test set in a ratio of 7:3; Input variables: air knife pressure, air knife distance, air knife angle, air knife height, strip speed, strip thickness, zinc liquid temperature; Output variable: front thickness gauge value; Parameter settings: the number of trees is 278, the maximum depth is set to 38, the minimum sample segmentation is 2, the minimum sample leaf node number is 1, and the maximum number of features is 80%; A random forest network coating thickness prediction model is established based on the network model parameters, and the parameters preprocessed and feature-selected in step 3 are input to train the random forest network coating thickness prediction model.

[0024] Based on the above technical solution, further, in step 5, the verification and judgment process is as follows:

[0025]

[0026] In the formula, The predicted value of the front thickness gauge, y i This is the actual value from the front thickness gauge. The average value of the front thickness gauge; n is the total number of data samples; MSE is the mean square error, MAE is the mean absolute error, and R... 2 The coefficient of determination reflects the predictive ability of the random forest network coating thickness prediction model; among them, R 2 The larger the value, the better the capability; the smaller the MSE and MAE, the better the prediction performance.

[0027] Based on the above technical solution, further, in step 6, the optimization process is as follows: using a grid search algorithm, an exhaustive search is performed within a predefined parameter range to find the optimal parameter combination for the random forest network coating thickness prediction model; wherein, the search range is as follows: number of trees: 50-500; maximum depth: 5-50; minimum number of sample splits: 2-20; minimum number of sample nodes: 1-10; finally, the parameter combination with the best performance is selected.

[0028] Compared with the prior art, the beneficial effects of the present invention are specifically reflected in:

[0029] This invention effectively improves the detection time lag in zinc layer thickness control of galvanized steel sheets and optimizes process parameter adjustment and thickness target control by acquiring historical production data; preprocessing the data and eliminating thickness gauge lag; performing feature selection based on mechanism and algorithm; establishing a random forest network coating thickness prediction model and training the network by inputting historical production data; verifying the accuracy and applicability of the random forest network coating thickness prediction model; and inputting process parameters and other influencing characteristic parameters during the production process into the random forest network coating thickness prediction model to obtain the zinc layer thickness prediction value. Attached Figure Description

[0030] Figure 1 This is a simplified flowchart of the prediction method of the present invention;

[0031] Figure 2 This is a simplified diagram of the hot-dip galvanizing process of the present invention;

[0032] Figure 3 This describes the variation and fluctuation of the zinc layer thickness during the actual production process of this invention.

[0033] Figure 4 This is a diagram showing the processing effect of step 2 of the present invention;

[0034] Figure 5 This is a diagram of the random forest network coating thickness prediction model of the present invention;

[0035] Figure 6 The zinc layer thickness model of this invention shows the predicted and measured thicknesses.

[0036] Figure 7 This is the result of the Spearman correlation coefficient calculation in this invention;

[0037] Figure 8 This refers to the prediction accuracy of the model in this invention.

[0038] Figure label:

[0039] 1. Annealing furnace; 2. Zinc pot; 3. Air knife; 4. Thickness gauge. Detailed Implementation

[0040] To make the objectives and technical solutions of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with embodiments.

[0041] Example

[0042] Reference Figure 1 and Figure 2 This embodiment provides a method for predicting the thickness of hot-dip galvanized coating based on the random forest algorithm. Preferably, the hot-dip galvanized coating is the zinc layer of a hot-dip galvanized steel coil. The hot-dip galvanized steel coil is produced by a hot-dip galvanizing process, which is as follows: the cold-rolled steel coil raw material is first uncoiled by an uncoiler, then annealed in an annealing furnace 1 and immersed in a zinc pot 2. The surface of the steel coil is dipped into the zinc liquid in the zinc pot 2, and then scraped by an air knife 3 to control the thickness of the surface zinc layer. After cooling over a distance, it reaches a thickness gauge 4 to detect the thickness of the hot-dip galvanized layer, and finally is rolled into a hot-dip galvanized steel coil. The prediction method includes the following steps:

[0043] Step 1: Produce hot-dip galvanized steel coils using the hot-dip galvanizing process and obtain historical production data for the hot-dip galvanizing process. This historical data should include at least: production specifications and control target data: steel grade, coating type, strip thickness, strip width, target coating thickness, etc.; key parameter data controlled by process engineers: air knife pressure, air knife distance, air knife height, etc.; various operating conditions, environmental parameters, and relevant parameter data from the previous process flow before the steel plate reaches the air knife: zinc bath temperature, strip speed, nitrogen pressure, nitrogen flow rate, actual value of the straightening roller, strip tension, air knife angle, etc.

[0044] Step 2: Preprocess historical data to eliminate thickness gauge lag; data preprocessing includes handling outliers, handling response lag after air knife parameter setting, and handling thickness gauge detection lag.

[0045] In this embodiment, the outlier handling process is as follows: During calibration, the operator removes data where the thickness gauge reading is 0; data where the set coating thickness does not match the actual thickness are deleted. Specifically, this includes the following situations: For a very small number of steel coils, the set coating thickness does not match the actual thickness, resulting in abnormal measurement data; these coil data are deleted. At certain weld seams, the air knife distance may be adjusted to a relatively large value, causing the measured thickness to far exceed the set value; this data needs to be discarded. For example, a sudden difference between the measured and set values ​​may occur due to a thickness gauge malfunction, or due to significant parameter adjustments during manual coil changing; this data is detrimental to model training. When about to change coating specifications, the operator removes data where the zinc layer thickness at the tail of the current steel coil is the same as that of the next steel coil when adjusting the air knife parameters in advance.

[0046] In this embodiment, the process of dealing with the response time lag after setting the air knife parameters is as follows: During the process of adjusting the air knife parameters, it is difficult for the data measured by the thickness gauge to accurately match the corresponding process parameters. For each record in the dataset of the steel coil, first, it is judged whether the air knife pressure and air knife distance parameters of the steel coil are adjusted, and the deviation between the set value and the actual value is calculated. 1. For the adjustment of the air knife pressure: It takes 5 - 10 s response time for the air knife pressure to adjust to the corresponding set value, and the process data is unstable, which is not suitable as a model training sample. Therefore, for each record in the dataset, subtract the actual value of the air knife pressure from the set value of the air knife pressure to get ΔP. By statistically calculating the stable process data, the standard deviation of the actual air knife pressure is obtained as: ΔP = P Set - P Act ; In the formula, P Set is the set value of the air knife pressure, and P Act is the actual value of the air knife pressure; If |ΔP| < σ mbar, it is determined as a record of stable process parameters and the data is retained; If |ΔP| > σ mbar, it is determined as a record of non - stable process parameters and the data is excluded. 2. Since the air knife distance is adjusted by a high - speed motor for micro - displacement and the response speed is extremely fast, when it is detected that the air knife distance parameter in the record entry is adjusted, no corresponding time - lag processing is done.

[0047] In this embodiment, for the detection time - lag processing of the thickness gauge, the processing process is as follows: The thickness gauge measures the thickness of the zinc coating at a distance of d meters before the air knife. It is necessary to retrieve the thickness of the zinc coating controlled by the corresponding air knife process parameters through the corresponding space - time conversion algorithm. First, take the i - th data record as an example. Set the speed of the i - th record as v i , define the relative displacement distance as S, with the unit of m. Then, when it comes to the (i + j) - th record, the forward distance of the steel coil position corresponding to the i - th record is: In the formula, v k is the speed of the k - th record, where k takes values from [i, i + j]; T is the time, with the unit of s. Second, set the actual displacement of the thickness gauge lagging behind the air knife as d meters. Then, when the (i + j) - th record satisfies d < S < d + 5, take the thickness gauge value of the (i + j) - th record as the thickness of the zinc coating controlled by the air knife process parameters corresponding to the i - th record, and at this time, take d + 5 meters as the maximum allowable value of the relative displacement; If the (i + j) - th record satisfying the condition of d < S < d + 5 cannot be retrieved through the loop, the i - th data record is excluded. Thus, the outliers in the dataset, the response time lag after setting the air knife parameters, and the detection time - lag of the thickness gauge are eliminated, and the matching of the process parameters and the thickness gauge thickness value is completed; Through the combined processing of the above methods, the best dataset for model training is obtained.

[0048] Step 3: Feature selection is performed on the preprocessed data based on mechanism and correlation coefficient. Specifically, feature selection is performed on the preprocessed data from the previous step from two dimensions: algorithm and mechanism. The mechanism analysis process involves performing fluid analysis on the air knife blowing process of the strip steel, and deriving the formula for coating thickness based on the Stokes equations. In the formula, W is the coating thickness; ρ is the zinc liquid density; μ is the zinc liquid viscosity; V is the strip speed; τ is the zinc liquid surface shear stress; p is the air knife pressure; and g is the gravitational acceleration.

[0049] The process of feature selection using correlation coefficients is as follows: The Spearman correlation coefficient ρ1 between data points is calculated as follows:

[0050] In the formula, x i This refers to the sample values ​​of a certain variable in the preprocessed dataset from step 2. This represents the sample mean of the variable; similarly, Y... i This refers to the sample values ​​of another variable in the preprocessed dataset from step 2. Let be the sample mean of another variable.

[0051] Based on the analysis of the galvanizing process mechanism, air knife pressure, air knife distance, air knife height, zinc bath temperature, strip speed, and strip thickness are all parameters that have a significant impact on the coating thickness.

[0052] Step 4: Establish a random forest network coating thickness prediction model. Input historical production data into the random forest network coating thickness prediction model and train the random forest network coating thickness prediction model. In this embodiment, the network model parameters are first determined: the number of trees is 278, the maximum depth is set to 38, the minimum sample split is 2, the minimum sample leaf node number is 1, and the maximum number of features is 80%. Based on the network model parameters, establish a random forest network coating thickness prediction model and input the parameters that have been preprocessed and feature-selected in Step 3 to train the random forest network coating thickness prediction model.

[0053] Step 5: Verify and determine the accuracy and applicability of the random forest network coating thickness prediction model; in this embodiment, the verification and determination process is as follows:

[0054]

[0055] In the formula, The predicted value of the front thickness gauge, y i This is the actual value from the front thickness gauge. The average value of the front thickness gauge; n is the total number of data samples; MSE is the mean square error, MAE is the mean absolute error, and R... 2 The coefficient of determination reflects the predictive ability of the random forest network coating thickness prediction model; among them, R2 The larger the value, the better the ability.

[0056] Step 6: If the model's prediction accuracy does not meet the requirements, optimize the parameters of the random forest network coating thickness prediction model and further train the random forest network coating thickness prediction model; if the random forest network coating thickness prediction model's prediction accuracy meets the requirements, then put the random forest network coating thickness prediction model into use.

[0057] Step 7: Input the process parameters and other influencing characteristic parameters of the production process into the random forest network coating thickness prediction model, and output the predicted value of zinc layer thickness.

[0058] This invention completes the prediction of hot-dip galvanized coating thickness based on the random forest algorithm. It can effectively improve the problem of thickness gauge detection lag after air knife process parameter adjustment in the hot-dip galvanizing process. It can control the deviation between the zinc layer thickness of the galvanized sheet and the actual value of the air knife pressure within 3g / ㎡ in a short time, reduce the fluctuation in the thickness control process, optimize the effectiveness of process parameter adjustment settings, control the uniformity and consistency of coating thickness and the pass rate, reduce the production cost of enterprises, and reduce the economic losses caused by the problem of substandard coating thickness.

[0059] The method will be further explained below with reference to specific implementation methods:

[0060] A steel mill's hot-dip galvanizing production line produces a coil with a coating thickness of 80g / m. 2 The thickness gauge data for the steel coil is attached. Figure 3 When the steel coil passes the air knife at the head position, the thickness gauge still measures the thickness data of the previous steel coil.

[0061] Step 1: Obtain historical production data for the hot-dip galvanizing process;

[0062] Continuous coating thickness data and air knife parameter data were obtained from a thickness gauge on the hot-dip galvanizing production line at the steel plant. Relevant parameters such as coating thickness, production speed, temperature, and air knife pressure were recorded. The data format is shown in Table 1.

[0063] Table 1

[0064]

[0065] Step 2: Preprocess historical data to eliminate thickness gauge lag;

[0066] In this embodiment, outliers in the data are detected and removed, and missing data points are filled using interpolation to ensure data continuity and integrity.

[0067] In this embodiment, an integral method is used to eliminate the lag of the thickness gauge in the data: at time t, the speed of the production line is tracked and its change over time is integrated. The actual value of the air knife pressure for integration is set to 200. When the integrated value reaches this target, the thickness gauge value recorded at this time is considered to be the coating thickness matching that time. Iterating this step for all thickness gauge values ​​eliminates the lag of the thickness gauge in the data. The processing effect is as follows: Figure 4 As shown.

[0068] Step 3: Perform feature selection on the preprocessed data based on the mechanism and correlation coefficient.

[0069] In this embodiment, Spearman's correlation coefficient was used for correlation analysis. This method was chosen because of its low sensitivity to data distribution and its ability to effectively quantify monotonic relationships in non-normally distributed data and scenarios containing outliers. By converting variables in the dataset into rankings, the correlations between these rankings were calculated. The Spearman coefficient ranges from -1 to +1, indicating perfectly negative to perfectly positive correlations, respectively. This analysis revealed the variables most strongly correlated with zinc coating thickness, providing crucial information for further model development and optimization. The correlation contour maps of each influencing factor with zinc coating thickness are attached, along with the correlations between various parameters and coating thickness shown in Table 2. Figure 7 As shown.

[0070] Table 2

[0071]

[0072] Taking into account the actual production process and the detailed evaluation of the correlation analysis results, the final input characteristics determined are: air knife pressure, air knife distance, air knife height, air knife angle, strip speed, strip thickness, and zinc liquid temperature.

[0073] Step 4: Establish a random forest network model for predicting coating thickness, such as... Figure 5 As shown, historical production data is input into the random forest network coating thickness prediction model to train the model. The training process is as follows: 1. Data preparation: The data is divided into training and testing sets in a ratio of 7:3; 2. Input variables: air knife pressure, air knife distance, air knife angle, air knife height, strip speed, strip thickness, and zinc liquid temperature; 3. Output variable: front thickness gauge value; 4. Parameter settings: the number of trees is 278, the maximum depth is set to 38, the minimum sample split is 2, the minimum sample leaf node number is 1, and the maximum number of features is 80%. A random forest network coating thickness prediction model is established based on the network model parameters, and the preprocessed dataset from step 3 is input to train the model.

[0074] Step 5: Verify and determine the accuracy and applicability of the random forest network coating thickness prediction model;

[0075] (1) Calculate the predicted value of the front thickness gauge Compared with the actual value y i Mean square error (MSE):

[0076]

[0077] (2) Calculate the predicted value of the front thickness gauge Compared with the actual value y i Mean Absolute Error (MAE):

[0078]

[0079] (3) Calculate the coefficient of determination R 2 :

[0080] (4) Model evaluation results:

[0081]

[0082] Appendix Figure 6 This scatter plot shows the comparison between the predicted and actual values ​​of all data points in the test set of the random forest network zinc layer thickness prediction model trained in step 4. In this plot, black dots represent the model's predictions, and the proximity of these points to the central diagonal (y = x) reflects the accuracy of the predictions. MSE < 1, MAE < 1, R 2 The value is >0.95, and the error between the predicted and actual values ​​of the random forest network coating thickness prediction model is within ±2 g / m. 2 The proportion within ±5g / m³ reached 95.30%, with an error within ±5g / m³. 2 The proportion of coating thickness within the specified range reached 99.66%, achieving the actual production control target. Thus, the establishment of the random forest network coating thickness prediction model was completed.

[0083] Step 6: If the model's prediction accuracy does not meet the requirements, optimize the parameters of the random forest network coating thickness prediction model and further train the random forest network coating thickness prediction model; if the random forest network coating thickness prediction model's prediction accuracy meets the requirements, then put the random forest network coating thickness prediction model into use.

[0084] In this embodiment, the optimization process is as follows: using a grid search algorithm, an exhaustive search is performed within a predefined parameter range to find the optimal combination of hyperparameters for the model. The search range is as follows: number of trees: 50-500; maximum depth: 5-50; minimum number of sample splits: 2-20; minimum number of sample nodes: 1-10; finally, the optimal parameter combination is selected, thereby significantly improving the prediction performance and robustness of the random forest prediction model.

[0085] Step 7: Input the preprocessed and feature-filtered feature parameters from Step 3 into the random forest network coating thickness prediction model, and output the predicted zinc layer thickness value. Refer to... Figure 8 As shown.

[0086] The above are merely embodiments of the present invention, described in a relatively specific and detailed manner, but should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention.

Claims

1. A hot dip galvanizing coating thickness prediction method based on a random forest algorithm, characterized by, The method comprises the following steps: Step 1: producing a hot-dip galvanized steel coil through a hot-dip galvanizing process, and obtaining historical data of the hot-dip galvanizing process; Step 2: preprocessing the historical data to eliminate the thickness gauge time lag; the historical data at least comprises product specifications and control target data, process personnel regulated parameter data, various working conditions, environmental parameters, and relevant parameter data of a steel plate reaching a previous process before the air knife; Step 3: performing feature screening on the preprocessed data in combination with mechanism analysis and correlation coefficients; The mechanism analysis process is: performing fluid analysis on the process of the air knife blowing the steel strip, and obtaining the coating thickness according to the Stokes equation; Using the correlation coefficient to determine the variable with the strongest correlation with the zinc layer thickness; Step 4: establishing a random forest network coating thickness prediction model, inputting the historical production data into the random forest network coating thickness prediction model, and training the random forest network coating thickness prediction model; Step 5: verifying and determining the accuracy and applicability of the random forest network coating thickness prediction model; Step 6: if the prediction accuracy of the random forest network coating thickness prediction model does not meet the requirements, optimizing the parameters of the random forest network coating thickness prediction model and further training the random forest network coating thickness prediction model; if the prediction accuracy of the random forest network coating thickness prediction model meets the requirements, putting the random forest network coating thickness prediction model into use; Step 7: inputting the feature parameters screened in step 3 into the random forest network coating thickness prediction model, and outputting the zinc layer thickness prediction value. 2.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 1, The product specifications and control target data: steel grade, coating type, strip thickness, strip width, target coating thickness; The process personnel regulated parameter data: air knife pressure, air knife distance, air knife height; The various working conditions, environmental parameters, and relevant parameter data of a steel plate reaching a previous process before the air knife: zinc liquid temperature, strip speed, nitrogen pressure, nitrogen flow, actual value of a straightening roll, strip tension, air knife angle. 3.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 2, the preprocessing includes abnormal value processing, and the processing process is: The operator removes the data with a thickness gauge value of 0 when calibrating; The data with a mismatch between the set coating thickness and the actual thickness is deleted; When the coating specification is changed and the operator adjusts the air knife parameters in advance, the data with the same zinc layer thickness of the tail of the current coil and the next coil is removed. 4.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 2, the preprocessing includes response time lag processing after the air knife parameter setting, and the processing process is: In the process of setting and adjusting the air knife parameters, for each record of the data set of the coil, first determine whether the air knife pressure and air knife distance parameters of the coil are adjusted, and calculate the deviation between the set value and the actual value; I. For the air knife pressure adjustment: for each record of the data set, the air knife pressure set value is subtracted from the air knife pressure actual value to obtain By statistical calculation on the stable process data, the standard deviation of the actual air knife pressure is obtained ; wherein is the gas knife pressure setpoint, is the gas knife pressure actual value; If mbar, it is determined as stable process parameter record, data retention; if mbar, it is determined as unstable process parameter record, data rejection; II. When it is detected that the air knife distance parameter in the record entry is adjusted, no corresponding time lag processing is performed. 5.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 2, the preprocessing includes detection time lag processing of the thickness gauge, and the processing process is: The thickness gauge measures the thickness of the zinc layer at a distance of d meters before the air knife; the corresponding space-time conversion algorithm is needed to retrieve the zinc layer thickness controlled by the air knife process parameters; First, set the speed of the i-th record as v i , define the relative displacement distance as S, unit: m, then when the i+j-th record, the i-th record corresponding to the coil position forward distance is: ; In the formula, is the speed of the k-th record, where k takes the value of [i, i+j]; T is the time, unit: s; Second, set the actual thickness gauge hysteresis air knife displacement d meters, when the i + j records meet d < s < d + 5, then the i + j records of the thickness gauge value as the i record of the air knife process parameters corresponding to the control of the zinc layer thickness, at this time d + 5 meters as the relative displacement of the maximum value; If the loop does not retrieve the i + j records meet d < s < d + 5, then the i data record. 6.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 3, the formula for the coating thickness: ; where, is the coating thickness; is the zinc bath density; μ is the zinc bath viscosity; V is the strip speed, τ is the surface shear stress of the zinc bath; p is the air knife pressure; and g is the acceleration of gravity. 7.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 3, the process of feature selection using correlation coefficient is: Spearman's correlation coefficient between each data The calculation method is: ； wherein is the sample value of a variable of the pre-processed data set in step 2, is the sample mean of the variable; Y i is the sample value of another variable of the pre-processed data set in step 2, is the sample mean of the other variable. 8.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 4, the training process is: Data preparation: divide the data into training set and test set, the ratio is 7:3; Input variables: air knife pressure, air knife distance, air knife angle, air knife height, strip speed, strip thickness, zinc liquid temperature; Output variable: front thickness gauge value; Parameter setting: the number of trees is 278, the maximum depth is set to 38, the minimum sample split is 2, the minimum sample leaf node number is 1, and the maximum feature number is 80%; Based on the network model parameter, a random forest network coating thickness prediction model is established, and the parameters preprocessed in step 3 and screened are used to train the random forest network coating thickness prediction model. 9.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 5, the verification and judgment process is: ； ； ； In the formula, is the predicted value of the front thickness gauge, is the actual value of the front thickness gauge, is the average value of the front thickness gauge; n is the total amount of data samples; MSE is the mean square error, MAE is the mean absolute error, R 2 is the determination coefficient, which reflects the pros and cons of the prediction ability of the random forest network coating thickness prediction model; wherein, R 2 The larger the value is, the better the ability is, and the smaller the values of MSE and MAE are, the better the prediction effect is. 10.The hot dip galvanizing coating thickness prediction method based on random forest algorithm according to claim 1, characterized in that, In step 6, the optimization process is: Using grid search algorithm, exhaustive search is performed in the pre-defined parameter range to find the best parameter combination of the random forest network coating thickness prediction model; Among them, the search range is as follows: the number of trees: 50-500; Maximum depth: 5-50; Minimum sample split number: 2-20; Minimum sample node number: 1-10; Finally, the best parameter combination is selected.

Images