Method and system for controlling zinc layer thickness based on hot galvanizing air knife distance

Abstract

The invention discloses a method for controlling the thickness of a zinc layer based on a hot galvanizing air knife distance. The method comprises the following steps: S1, data acquisition; s2, establishing data; s3, deducing data; s4, error calculation; and S5, error correction. The invention further discloses a system for controlling the thickness of the zinc layer based on the hot galvanizing air knife distance. The technical problems that according to a method for manually correcting according to data fed back by the coating thickness gauge, distance correction is not stable and is greatly influenced by objective factors, and due to manual correction, the distance correction precision is low, and the product quality is influenced are solved.

Description

Technical Field

[0001] The invention relates to an automatic control technology for hot-dip galvanizing, and more specifically to a method and a system for controlling the thickness of a zinc layer based on a hot-dip galvanizing air knife distance. Background Art

[0002] The thickness of the zinc layer refers to the thickness of the zinc covering the surface of the material. Usually, the thickness of the zinc layer is formed on the surface of the material through processes such as electroplating or hot-dip plating. The thickness of the zinc layer can be adjusted according to different application requirements. The thickness of the zinc layer is very important for the corrosion resistance of the material. Generally speaking, the thicker the zinc layer, the stronger the corrosion resistance of the material. The thickness of the zinc layer can be determined by measurement technology, such as using equipment such as electron microscopes or thickness gauges. In different applications, the thickness of the zinc layer should be determined according to specific requirements. For example, in automobile manufacturing, the thickness of the zinc layer is generally between 5 and 25 microns; in the construction field, the thickness of the zinc layer may need to be thicker to provide better weather resistance and corrosion resistance. Hot-dip galvanizing air knife is a device used in hot-dip galvanizing process to remove oxides and dirt on the surface of the material to better form a zinc layer.

[0003] The traditional mechanical principle of controlling the zinc layer thickness is usually achieved by adjusting the air knife distance. The zinc layer thickness is controlled by adjusting the distance between the air knife and the surface of the object. According to the specific production process and requirements, it can be achieved by mechanical adjustment, electric adjustment or manual adjustment. Through continuous testing and adjustment, the desired zinc layer thickness control effect can be achieved. The method of manually adjusting the data fed back by the coating thickness gauge is used to control the zinc layer thickness. However, the method of manually correcting the distance based on the data fed back by the coating thickness gauge is unstable and greatly affected by objective factors. In addition, the distance correction accuracy is low due to manual correction, which affects the quality of the product. Summary of the invention

[0004] In view of the defects existing in the prior art, the purpose of the present invention is to provide a method and system for controlling the thickness of the zinc layer based on the hot-dip galvanizing air knife distance, so as to solve the technical problem that the distance correction is unstable and greatly affected by objective factors, and the distance correction is manually corrected, and the accuracy of the distance correction is low, which affects the product quality.

[0005] To achieve the above object, the present invention adopts the following technical solution:

[0006] A first aspect of the present invention provides a method for controlling the thickness of a zinc layer based on a hot-dip galvanizing air knife distance, comprising the following steps:

[0007] S1, data collection;

[0008] S2, data establishment;

[0009] S3, data derivation;

[0010] S4, error calculation;

[0011] S5. Error correction.

[0012] Preferably, the data collection in step S1 specifically includes:

[0013] A data capturer is used to capture multiple sets of distance data from the air knife to the strip, as well as multiple sets of zinc layer thickness data of the strip after processing at the distance.

[0014] Preferably, the step S2 data establishment specifically includes:

[0015] The multiple sets of data captured by the data capturer are transmitted to a processing center, and the processing center establishes a priori model of Gaussian process regression (GPR) based on the multiple sets of data, and then constructs a covariance matrix through the similarities between the multiple sets of data.

[0016] Preferably, the prior model of the Gaussian process regression GPR is optimized, and the hyperparameters of the prior model of the Gaussian process regression GPR are calculated by the Bayesian formula, and then the prior model of the Gaussian process regression GPR is trained by using multiple sets of data captured by the data capturer, and the hyperparameters are optimized to obtain the formula p(Y|X,x*)=N(m*,Σ*);

[0017] Where x is the zinc layer thickness of the strip, y is the distance between the air knife and the strip, m* is the mean vector of the unobserved output, and Σ* is the covariance matrix of the unobserved output.

[0018] Preferably, the data derivation in step S3 specifically includes:

[0019] The data capturer is used to detect the thickness of the zinc layer of the strip after processing and record it as special data. The processing center imports the special data into the prior model of the Gaussian process regression GPR, and further derives data for predicting the distance between the air knife and the strip according to the formula p(Y|X,x*)=N(m*,Σ*), which is recorded as predicted data.

[0020] Preferably, the error calculation in step S4 specifically includes:

[0021] The actual distance between the air knife and the strip is recorded as actual data, and the processing center calculates the derived predicted data and the actual data, calculates the prediction error between the predicted data and the actual data, and compares the predicted error with the error range.

[0022] Preferably, the error correction in step S5 specifically includes:

[0023] When the prediction error exceeds the error range, the processing center starts the correction controller of the air knife to correct the distance between the air knife and the strip;

[0024] After the correction is completed, the data capturer detects the zinc layer thickness of the strip again, inputs the actual data obtained into the formula p(Y|X,x*)=N(m*,Σ*), derives the predicted data, and calculates the derived predicted data with the actual data. When it is within the error value range, the correction of the spacing between the air knife and the strip is completed, otherwise the data derivation will continue.

[0025] The second aspect of the present invention provides a system for controlling the thickness of a zinc layer based on a hot-dip galvanizing air knife distance, comprising:

[0026] A data capturer is used to capture multiple sets of distance data between the air knife and the strip, and multiple sets of zinc layer thickness data after processing at the distance;

[0027] The processing center builds a priori models of Gaussian process regression (GPR) based on multiple sets of data, and compares, calculates and corrects the data;

[0028] The system for controlling the thickness of a zinc layer based on hot-dip galvanizing air knife distance realizes the method for controlling the thickness of a zinc layer based on hot-dip galvanizing air knife distance.

[0029] Preferably, the data capture device comprises:

[0030] A laser measuring instrument, used to detect the distance data between the air knife and the strip;

[0031] Thickness gauge, used to measure the thickness of the zinc layer of the strip after processing.

[0032] Preferably, the processing center includes:

[0033] An algorithm module for analyzing the data from the data capture device;

[0034] a processing module, for comparing the data derived by the algorithm module with the actual distance between the air knife and the strip;

[0035] A control module is used to control the correction controller.

[0036] The present invention provides a method and system for controlling the thickness of a zinc layer based on the distance of a hot-dip galvanizing air knife, which has the following beneficial effects:

[0037] 1) The present invention detects the zinc layer thickness in real time through a processing center, and calculates it through Gaussian process regression GPR to ensure that the distance between the air knife and the strip is within a reasonable error range. When it exceeds the reasonable error range, the processing center will start a correction controller to correct the distance between the air knife and the strip by the calculated error value. After the correction is completed, the data is calculated again to ensure that the data is within a reasonable error range. Gaussian process regression GPR has strong flexibility and can handle high-dimensional input and multi-output situations, and can achieve better prediction effects.

[0038] 2) The present invention uses Gaussian process regression GPR and uses methods such as cross-validation to train the Gaussian process regression GPR model for multiple groups of strip zinc layer thicknesses and the distance between the air knife and the strip, and can derive the optimal kernel function and parameters to improve the accuracy of the model.

[0039] 3) The present invention calculates the correction distance from the air knife to the strip through Gaussian process regression GPR, ensuring the optimal distance between the air knife and the strip under the algorithm calculation, making the air knife adjustment more accurate, and improving the correction efficiency of the air knife to a certain extent, ensuring that the product quality meets the index requirements and reducing the generation of unqualified products. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic flow chart of a method for controlling the thickness of a zinc layer based on hot-dip galvanizing air knife distance according to the present invention;

[0041] Figure 2 It is a schematic diagram of the framework structure of the system for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance of the present invention;

[0042] Figure 3 It is a schematic flow chart of data establishment in the method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance of the present invention. DETAILED DESCRIPTION

[0043] In order to better understand the above technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0044] Combination Figures 1 to 3 As shown, the present invention provides a method for controlling the thickness of a zinc layer based on a hot-dip galvanizing air knife distance, comprising the following steps:

[0045] S1. Data acquisition: using a data capturer 1 to capture multiple sets of distance data between the air knife 2 and the strip 3, and multiple sets of zinc layer thickness data of the strip 3 after processing at the distance.

[0046] S2, data establishment, transmitting the multiple sets of data captured by the data capturer 1 to the processing center 4, the processing center 4 establishes a priori model of Gaussian process regression GPR based on the multiple sets of data, and then constructs a covariance matrix through the similarity between the multiple sets of data.

[0047] The prior model of Gaussian process regression GPR is optimized, and the hyperparameters of the prior model of Gaussian process regression GPR are calculated by Bayesian formula. Then, the prior model of Gaussian process regression GPR is trained by multiple sets of data captured by data capturer 1, and the hyperparameters are optimized to obtain the formula p(Y|X,x*)=N(m*,Σ*), where x is the zinc layer thickness of the strip, y is the distance between the air knife and the strip, m* is the mean vector of the unobserved output, and Σ* is the covariance matrix of the unobserved output. This formula represents the conditional probability distribution of the unobserved output given the observed data and the new input x*.

[0048] S3. Data derivation: The zinc layer thickness of the processed strip 3 is detected by the data capturer 1 and recorded as special data. The processing center 4 imports the special data into the prior model of Gaussian process regression GPR, and further derives the data for predicting the distance between the air knife 2 and the strip 3 according to the formula p(Y|X,x*)=N(m*,Σ*), which is recorded as predicted data.

[0049] S4, error calculation, the actual distance between the air knife 2 and the strip 3 is recorded as actual data, the processing center 4 calculates the derived predicted data and the actual data, calculates the predicted error between the predicted data and the actual data, and compares the predicted error with the error range.

[0050] S5, error correction, when the predicted error exceeds the error range, the processing center 4 starts the correction controller of the air knife 2 to correct the distance between the air knife 2 and the strip 3;

[0051] After the correction is completed, the data capture device 1 detects the zinc layer thickness of the strip 3 again, inputs the actual data into the formula p(Y|X,x*)=N(m*,Σ*), derives the predicted data, and calculates the derived predicted data with the actual data. When it is within the error value range, the correction of the spacing between the air knife 2 and the strip 3 is completed, otherwise the data derivation will continue.

[0052] refer to Figure 2 As shown, the present invention also provides a system for controlling the thickness of a zinc layer based on a hot-dip galvanizing air knife distance, comprising:

[0053] A data capturer 1 is used to capture multiple sets of distance data between the air knife 2 and the strip 3, and multiple sets of zinc layer thickness data of the strip 3 after processing at the distance;

[0054] Processing center 4, establishes a priori model of Gaussian process regression GPR based on multiple sets of data, and compares, calculates and corrects the data;

[0055] The invention realizes a method for controlling the thickness of a zinc layer based on the distance of a hot-dip galvanizing air knife and a system for controlling the thickness of a zinc layer based on the distance of a hot-dip galvanizing air knife.

[0056] The data capture device 1 includes:

[0057] A laser measuring instrument 101 is used to detect the distance data between the air knife 2 and the strip 3;

[0058] The thickness gauge 102 is used to measure the thickness of the zinc layer of the strip 3 after processing.

[0059] The thickness gauge 102 measures the thickness of the zinc layer of the strip 3 using an electromagnetic induction method.

[0060] The processing center 4 includes:

[0061] Algorithm module 401, used for analyzing the data captured by the data capturer 1;

[0062] A processing module 402 is used to compare the data derived by the algorithm module 401 with the actual distance between the air knife 2 and the strip 3;

[0063] The control module 403 is used to control the correction controller.

[0064] The algorithm module 401 is used to establish a Gaussian process regression GPR prior model for the data measured by the laser measuring instrument 101 and the thickness gauge 102, and to analyze the predicted data and the actual data. By inputting the actual thickness data of the zinc layer of the strip 3 measured by the thickness gauge 102 into the Gaussian process regression GPR prior model, the predicted distance between the air knife 2 and the strip 3 is derived according to the formula p(Y|X,x*)=N(m*,Σ*).

[0065] The processing module 402 calculates the error value between the actual distance between the air knife 2 and the strip 3 and the predicted distance between the air knife 2 and the strip 3 by subtracting the actual distance between the air knife 2 and the strip 3 from the predicted distance between the air knife 2 and the strip 3 .

[0066] The control module 403 is used to control the correction controller to adjust the distance between the air knife 2 and the strip 3 according to the calculation result of the processing module 402 .

[0067] The present invention changes the transmission mechanical principle method of controlling the zinc layer thickness into a three-dimensional model, which is calculated and processed by a processing center, and the air knife is controlled by a computer to perform distance correction. The air knife adjustment is made more accurate through Gaussian process regression (GPR) algorithm calculation and correction, which improves the correction efficiency of the air knife to a certain extent, ensures that the product quality meets the index requirements, and reduces the generation of unqualified products.

[0068] Example

[0069] The method for controlling the thickness of the zinc layer based on the hot-dip galvanizing air knife distance in this embodiment specifically includes the following steps:

[0070] S1. Data acquisition. Multiple sets of distance data between the air knife 2 and the strip 3 and multiple sets of zinc layer thickness data of the processed strip 3 at the distance are captured by a data capturer 1. The data capturer 1 comprises a laser measuring instrument 101 and a thickness gauge 102. The laser measuring instrument 101 is used to detect the distance between the air knife 2 and the strip 3. The thickness gauge 102 is used to detect the zinc layer thickness of the processed strip 3. The thickness gauge 102 uses an electromagnetic induction method to measure the thickness of the zinc layer of the strip 3.

[0071] S2. Data establishment: multiple sets of data captured by the data capturer 1 are transmitted to the processing center 4. The processing center 4 establishes a prior model of Gaussian process regression GPR based on the multiple sets of data. Through the similarity between the multiple sets of data, a covariance matrix is ​​constructed to optimize the prior model of Gaussian process regression GPR. The hyperparameters of the prior model of Gaussian process regression GPR are calculated through the Bayesian formula. Then, the prior model of Gaussian process regression GPR is trained by the multiple sets of data captured by the data capturer 1, and the hyperparameters are optimized to obtain the formula p(Y|X,x*)=N(m*,Σ*), where x is the zinc layer thickness of the strip, y is the distance between the air knife and the strip, m* is the mean vector of the unobserved output, and Σ* is the covariance matrix of the unobserved output. This formula represents the conditional probability distribution of the unobserved output given the observed data and the new input x*. The processing center 4 includes an algorithm module 401, a processing module 402 and a control module 403. The algorithm module 401 is used to analyze the data captured by the data capturer 1, the processing module 402 is used to compare the data derived by the algorithm module 401 with the actual distance between the air knife 2 and the strip 3, and the control module 403 is used to control the correction controller.

[0072] S3. Data derivation: The thickness of the zinc layer of the processed strip 3 is detected by the data capturer 1 and recorded as special data. The processing center 4 imports the special data into the prior model of the Gaussian process regression GPR, and further derives the data for predicting the distance between the air knife 2 and the strip 3 according to the formula p(Y|X,x*)=N(m*,Σ*), which is recorded as predicted data. The algorithm module 401 is used to establish the prior model of the Gaussian process regression GPR for the data measured by the laser measuring instrument 101 and the thickness gauge 102, and to analyze the predicted data and the actual data. The actual thickness data of the zinc layer of the strip 3 measured by the thickness gauge 102 is input into the prior model of the Gaussian process regression GPR, and the predicted distance between the air knife 2 and the strip 3 is derived according to the formula p(Y|X,x*)=N(m*,Σ*).

[0073] S4, error calculation, the actual distance between the air knife 2 and the strip 3 is recorded as the actual data, the processing center 4 calculates the derived predicted data and the actual data, calculates the prediction error between the predicted data and the actual data, and compares the predicted error with a reasonable error range. The processing module 402 is used to compare the predicted distance between the air knife 2 and the strip 3 derived by the algorithm module 401 with the actual distance between the air knife 2 and the strip 3. The algorithm module 401 calculates the error value between the actual distance between the air knife 2 and the strip 3 and the predicted distance between the air knife 2 and the strip 3 by subtracting the actual distance between the air knife 2 and the strip 3 from the predicted distance between the air knife 2 and the strip 3.

[0074] S5. Error correction. When the predicted error exceeds a reasonable error range, the processing center 4 starts the correction controller of the air knife 2 to reasonably correct the distance between the air knife 2 and the strip 3. After the correction is completed, the data capturer 1 detects the zinc layer thickness of the strip 3 again, and inputs the actual data obtained into the formula p(Y|X, x*)=N(m*,Σ*) to derive the predicted data. The derived predicted data is calculated with the actual data. When the data is within a reasonable error value range, the correction of the distance between the air knife 2 and the strip 3 is completed, otherwise the data derivation will continue. The control module 403 is used to control the correction controller to adjust the distance between the air knife 2 and the strip 3 according to the calculation result of the processing module 402.

[0075] The zinc layer thickness of the strip 3 after processing detected by the thickness gauge 102, the actual distance between the air knife 2 and the strip 3, the distance between the air knife 2 and the strip 3 calculated by the control module 402, the zinc layer thickness of the strip 3 after correction, and the zinc layer thickness of the strip 3 after manual correction are compared. The specific comparison data are shown in the following table:

[0076]

[0077] The above data show that through the distance adjustment calculated by Gaussian process regression GPR, the zinc layer thickness of strip 3 after adjustment is closer to the processed zinc layer thickness required by the product, while the error value of the zinc layer thickness of strip 3 after manual correction is greater than the zinc layer thickness of strip 3 adjusted by Gaussian process regression GPR. The correction distance from air knife 2 to strip 3 is calculated by Gaussian process regression GPR to ensure the optimal distance between air knife 2 and strip 3 under algorithm calculation, making the adjustment of air knife 2 more accurate, and to a certain extent improving the correction efficiency of air knife 2, ensuring that product quality meets index requirements and reducing the production of unqualified products.

[0078] The error data calculation table of the zinc layer thickness when the control module 402 adjusts the distance between the air knife 2 and the strip 3 and the actual zinc layer thickness:

[0079]

[0080] Error calculation table of zinc layer thickness under manual adjustment of the distance between air knife 2 and strip 3 and the actual zinc layer thickness:

[0081]

[0082] Comparison table of the zinc layer thickness error value when the control module 402 adjusts the distance between the air knife 2 and the strip 3 and the zinc layer thickness error value of the strip after manual correction:

[0083]

[0084] The above data show that the error value of the zinc layer thickness when the control module 402 adjusts the distance from the air knife 2 to the strip 3 is smaller than the error value of the zinc layer thickness of the strip after manual correction, which can further reduce the error between the zinc layer thickness on the product surface and the target value, making the product more sophisticated.

[0085] Those skilled in the art should recognize that the above embodiments are only used to illustrate the present invention, and are not intended to limit the present invention. As long as they are within the spirit of the present invention, any changes or modifications to the above embodiments will fall within the scope of the claims of the present invention.

Claims

1. A method for controlling the thickness of a zinc layer based on hot-dip galvanizing air knife distance, characterized in that: The following steps are involved: S1, data collection; S2, data establishment; S3, data derivation; S4, error calculation; S5. Error correction.

2. The method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance according to claim 1 is characterized in that: The step S1 data collection specifically includes: A data capturer is used to capture multiple sets of distance data from the air knife to the strip, as well as multiple sets of zinc layer thickness data of the strip after processing at the distance.

3. The method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance according to claim 2 is characterized in that: The step S2 data establishment specifically includes: The multiple sets of data captured by the data capturer are transmitted to a processing center, and the processing center establishes a priori model of Gaussian process regression (GPR) based on the multiple sets of data, and then constructs a covariance matrix through the similarities between the multiple sets of data.

4. The method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance according to claim 3 is characterized in that: The prior model of the Gaussian process regression GPR is optimized, and the hyperparameters of the prior model of the Gaussian process regression GPR are calculated by using the Bayesian formula, and then the prior model of the Gaussian process regression GPR is trained by using multiple sets of data captured by the data capturer, and the hyperparameters are optimized to obtain the formula p(Y|X,x*)=N(m*,Σ*); Where x is the zinc layer thickness of the strip, y is the distance between the air knife and the strip, m* is the mean vector of the unobserved output, and Σ* is the covariance matrix of the unobserved output.

5. The method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance according to claim 4 is characterized in that: The data derivation in step S3 specifically includes: The data capturer is used to detect the thickness of the zinc layer of the strip after processing and record it as special data. The processing center imports the special data into the prior model of the Gaussian process regression GPR, and further derives data for predicting the distance between the air knife and the strip according to the formula p(Y|X,x*)=N(m*,Σ*), which is recorded as predicted data.

6. The method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance according to claim 5 is characterized in that: The error calculation in step S4 specifically includes: The actual distance between the air knife and the strip is recorded as actual data, and the processing center calculates the derived predicted data and the actual data, calculates the prediction error between the predicted data and the actual data, and compares the predicted error with the error range.

7. The method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance according to claim 6 is characterized in that: The error correction step S5 specifically includes: When the prediction error exceeds the error range, the processing center starts the correction controller of the air knife to correct the distance between the air knife and the strip; After the correction is completed, the data capturer detects the zinc layer thickness of the strip again, inputs the actual data obtained into the formula p(Y|X,x*)=N(m*,Σ*), derives the predicted data, and calculates the derived predicted data with the actual data. When it is within the error value range, the correction of the spacing between the air knife and the strip is completed, otherwise the data derivation will continue.

8. A system for controlling zinc layer thickness based on hot-dip galvanizing air knife distance, characterized in that: include: A data capturer is used to capture multiple sets of distance data between the air knife and the strip, and multiple sets of zinc layer thickness data after processing at the distance; The processing center builds a priori models of Gaussian process regression (GPR) based on multiple sets of data, and compares, calculates and corrects the data; The system for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance realizes the method for controlling the zinc layer thickness based on the hot-dip galvanizing air knife distance as described in any one of claims 1-7.

9. The system for controlling zinc layer thickness based on hot-dip galvanizing air knife distance according to claim 8 is characterized in that: The data capture device comprises: A laser measuring instrument, used to detect the distance data between the air knife and the strip; Thickness gauge, used to measure the thickness of the zinc layer of the strip after processing.

10. The system for controlling zinc layer thickness based on hot dip galvanizing air knife distance according to claim 8, characterized in that: The processing center includes: An algorithm module for analyzing the data from the data capture device; a processing module, for comparing the data derived by the algorithm module with the actual distance between the air knife and the strip; A control module is used to control the correction controller.

Images