A steel quality analysis and feedback system

Abstract

The present invention discloses a steel quality analysis and feedback system, which specifically relates to the field of material science, and includes a steel data acquisition module, a steel data preprocessing module, a physical feature extraction module, a chemical composition quantitative analysis module, an image analysis module, a steel quality comprehensive analysis module, a steel quality judgment module, and a report and output module. A steel quality analysis and feedback system analyzes the physical property data of the target steel through the physical feature extraction module, constructs a physical feature matrix, calculates the mechanical properties of the target steel, and ensures the fit to the known data; quantifies the correlation between the spectral data and the element score through the chemical composition quantitative analysis module, reflects the characteristic spectral intensity of the elements in the target steel, and realizes the effective conversion from spectral data to chemical composition concentration; and quantifies the texture features of the image through the image analysis module to evaluate the surface quality of the steel.

Description

Technical Field

[0001] The present invention relates to the technical field of material science, and more specifically, to a steel quality analysis and feedback system. Background Art

[0002] With the development of intelligent manufacturing, high-quality steel is the key to ensuring the safety and performance of industries such as construction, automobiles, ships and machinery manufacturing. Ensuring the quality of steel has become a key factor in the development of the industry. The strength, toughness and corrosion resistance of steel are directly related to the safety and reliability of the product. Therefore, the steel quality analysis and feedback system is particularly important, which can not only enhance the market competitiveness of products, but also reduce the risks caused by quality problems.

[0003] Traditional steel quality analysis techniques, such as optical microscopy and chemical analysis, have already guaranteed the basic quality of products to a certain extent in steel quality analysis. Through regular physical testing and chemical analysis, detailed evaluation of the microstructure and chemical composition of steel can help researchers and engineers identify and classify various inclusions and defects, effectively screen out products that do not meet standards, and reduce the possibility of unqualified products entering the market. This not only improves product quality, but also enhances the transparency and traceability of the production process, and plays a positive role in stabilizing product performance and improving customer satisfaction.

[0004] However, in actual use, it still has some shortcomings, such as low efficiency of manual detection, time-consuming analysis process and reliance on operator experience and subjective judgment, which makes it difficult to adapt to the needs of real-time monitoring in large-scale continuous production; the lack of effective data tracking and feedback mechanism makes it difficult to quickly locate and solve the root cause of the problem. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a steel quality analysis and feedback system, which solves the problems raised in the above-mentioned background technology through the following scheme.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A steel quality analysis and feedback system, comprising:

[0008] Steel data acquisition module: used to obtain physical property data, chemical composition data and surface image data of the target steel through different acquisition devices, and transmit them to the steel data preprocessing module;

[0009] Steel data preprocessing module: used to clean and standardize the data transmitted by the steel data acquisition module and eliminate different dimensions and magnitudes, extract the quality characteristics of the target steel from the original image and data, and generate a characteristic data set of the steel; the characteristic data set of the steel includes physical property data, chemical composition data, and image data;

[0010] Physical feature extraction module: used to analyze the physical property data of the target steel transmitted by the steel data preprocessing module, calculate the mechanical properties of the target steel through the physical feature extraction formula, construct the physical feature matrix, and transmit it to the steel quality comprehensive analysis module;

[0011] Chemical composition quantitative analysis module: used to construct a matrix of quantitative concentrations of each element in the target steel using the chemical composition quantitative analysis formula, which is collected by spectral analysis technology and in a unified format by the steel data preprocessing module, and transmit it to the steel quality comprehensive analysis module;

[0012] Image analysis module: used to analyze the image data transmitted by the steel data preprocessing module, evaluate the surface quality of the steel through the image texture feature analysis formula, and transmit the results to the steel quality comprehensive analysis module;

[0013] Steel quality comprehensive analysis module: used to integrate the matrix data transmitted by the physical feature extraction module, chemical composition quantitative analysis module and image analysis module, conduct a comprehensive evaluation of the overall quality of the target steel, generate a comprehensive analysis index of steel quality, identify potential quality risks and transmit them to the steel quality judgment module;

[0014] Steel quality judgment module: used to judge the comprehensive analysis index of steel quality transmitted by the comprehensive evaluation module, realize automatic judgment of steel quality grade through predefined steel quality threshold, and generate corresponding signals to be transmitted to the report and output module;

[0015] Report and output module: After receiving the grade signal transmitted by the steel quality judgment module, it is used to generate a corresponding quality report according to the content of the signal, including the detailed evaluation results, quality grade and recommended measures of the target steel.

[0016] Preferably, the method for acquiring the physical property data of the target steel in the steel data acquisition module is as follows:

[0017] A data acquisition node that can accurately monitor the corresponding physical properties is deployed at intervals of 10 cm at each key position of the steel production line; the data acquisition node collects and transmits data in real time and performs appropriate signal processing to ensure that the data collected from different monitoring devices are synchronized in time; the sensor nodes that can accurately monitor the corresponding physical properties include using infrared sensors to monitor the temperature at points on the production line in real time, obtaining the pressure data of the rolling mill through pressure sensors, and continuously detecting the thickness of the target steel through ultrasonic thickness gauges;

[0018] The method for obtaining the chemical composition data of the target steel is as follows:

[0019] After confirming the length of the target steel, use a sawing machine to cut the starting and ending positions of the target steel into small pieces of 5cm×5cm respectively; use a polishing machine to grind the surface of the small piece of steel flat and remove the oxide layer and other impurities; grind the small piece of steel after grinding into powder and use cold pressing technology to press the powder into a film for spectral analysis, collect and record the emission or absorption spectral data, and identify and quantify various elements in the steel; the spectral data includes the carbon content that affects the strength and hardness of the steel, and the content of chemical elements that affect the toughness, weldability and corrosion resistance of the steel;

[0020] The specific method for acquiring the target steel surface image data is as follows:

[0021] Install high-definition cameras and lighting equipment on the steel production line to obtain surface images of the target steel at different positions and angles; pre-process the acquired images, including adjusting contrast, denoising, edge enhancement, etc., to improve the recognizability of defect features.

[0022] Preferably, the specific processing method corresponding to the steel data preprocessing module is:

[0023] The steel data preprocessing module preprocesses the physical property data, and the calculation formula is specifically expressed as:

[0024] ,

[0025] Among them, P i represents the preprocessed value corresponding to the ith physical property data after removing outliers, and m represents the value for calculating P i The data window size is , l is represented by the index of the physical property data point, ranging from (i-(m / 2)) to (i+(m / 2)), p i It is represented as the i-th physical property data after removing outliers;

[0026] The steel data preprocessing module preprocesses the chemical composition data, and the calculation formula is specifically expressed as:

[0027] ,

[0028] Among them, C j Represents the preprocessed value of the jth chemical composition data, c j Represented as the jth chemical composition data, min (c j ) represents the minimum value of all chemical composition data, s represents the standard deviation of chemical composition data, max(c j ) is represented by the maximum value of all chemical composition data, and μ is represented by the average value of chemical composition data;

[0029] The steel data preprocessing module preprocesses the image data of each surface, and the calculation formula is specifically expressed as:

[0030] ,

[0031] Among them, I k It is represented as the preprocessed value of the kth pixel in the image, k represents the pixel, f represents the image direction, a f Expressed as the weight coefficient related to direction f, I k+f It is expressed as the intensity value of the positive neighbor pixel of pixel k in direction f, I k-f It is expressed as the intensity value of the negative neighbor pixel of pixel k in direction f, I k+nf It is expressed as the intensity value of the positive neighbor pixel of pixel k in the diagonal direction nf, I k-nf It is expressed as the intensity value of the negative neighbor pixel of the pixel point k in the diagonal direction nf, and N is expressed as the offset in the diagonal direction.

[0032] Preferably, the physical feature extraction and calculation formula are specifically expressed as:

[0033] ,

[0034] Among them, PFE represents the physical feature matrix of the target steel, n represents the total number of physical property data, PDM represents the physical data matrix of the target steel, is represented as the regression coefficient vector, PTV is represented as the quality index of standard steel, λ is represented as the regularization parameter, represents the L1 norm, Refers to the sum of the absolute values ​​of the regression coefficient vector; represents the L2 norm, Refers to the size of the difference between the physical feature matrix and the actual value.

[0035] Preferably, the chemical composition quantitative analysis is performed and the calculation formula is specifically expressed as:

[0036] ,

[0037] Where Y represents the matrix composed of the chemical composition concentration of the target steel, X represents the matrix composed of the spectral data of the target steel, and t j It is represented as the j-th element score of matrix X, w j It is represented as the loading of the jth element of the matrix X, (B1, B2, ..., B j） It is represented as the score matrix shared by matrix X and matrix Y, q j Represented as the weight of the j-th element in the matrix Y.

[0038] Preferably, the image texture feature is calculated by the formula:

[0039] ,

[0040] Where ITFE is expressed as the conjugate matrix of grayscale under the condition of distance d and angle θ, α and β are the grayscale index of image length and width respectively, δ is the Kronecker function, η k and η k ' represents the grayscale value corresponding to pixel k and its adjacent pixels, g(α, β) represents the probability of grayscale appearing at position (α, β), g(α+(d×cosθ), β+(d×sinθ)) represents the probability of grayscale appearing at position (α, β) after being offset by distance d and angle θ, cos(θ) and sin(θ) represent the cosine and sine values ​​of angle θ.

[0041] Preferably, the calculation formula of the comprehensive analysis index of steel quality is specifically expressed as:

[0042] ,

[0043] Among them, CSQ represents the comprehensive analysis index of target steel quality, PFE represents the physical characteristic matrix of target steel, Y represents the matrix composed of chemical component concentration of target steel, ITFE represents the conjugate matrix of gray level under the conditions of distance d and angle θ, χ1, χ2, χ3 represent the weight coefficient of the physical characteristic matrix of target steel, the weight coefficient of the matrix composed of chemical component concentration of target steel and the weight coefficient of the conjugate matrix of gray level under the conditions of distance d and angle θ, respectively, f(PFE), f(Y), f(ITFE) represent the processing functions of the physical characteristic matrix of target steel, the processing functions of the matrix composed of chemical component concentration of target steel and the processing functions of the conjugate matrix of gray level under the conditions of distance d and angle θ, respectively.

[0044] Preferably, the specific evaluation methods corresponding to the steel quality judgment module include:

[0045] The data transmitted by the comprehensive evaluation module is compared with a series of pre-set steel quality risk thresholds, and combined with predefined scoring rules to determine the overall quality grade of the steel, convert the quality grade into a corresponding signal, and send the generated quality grade signal to the report and output module; the predefined scoring rules include that if the target steel quality comprehensive analysis index is between 90 and 100, the steel quality is excellent and the A signal is transmitted; if the target steel quality comprehensive analysis index is between 75 and 90, the steel quality is good and the B signal is transmitted; if the target steel quality comprehensive analysis index is between 60 and 75, the steel quality is qualified and the C signal is transmitted; if the target steel quality comprehensive analysis index is less than 60, the steel quality is unqualified and the D signal is transmitted.

[0046] Technical effects and advantages of the present invention:

[0047] The present invention analyzes the physical property data of the target steel through a physical feature extraction module, constructs a physical feature matrix, calculates the mechanical properties of the target steel, and ensures the fit to the known data;

[0048] The present invention quantifies the correlation between spectral data and element scores through a chemical composition quantitative analysis module, reflects the characteristic spectral intensity of the elements in the target steel, and realizes the effective conversion from spectral data to chemical composition concentration;

[0049] The present invention uses an image analysis module to perform quantitative analysis on the texture features of an image and evaluate the surface quality of the steel. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] As attached Figure 1 A steel quality analysis and feedback system shown includes a steel data acquisition module, a steel data preprocessing module, a physical feature extraction module, a chemical composition quantitative analysis module, an image analysis module, a steel quality comprehensive analysis module, a steel quality judgment module, and a report and output module.

[0053] The steel data acquisition module is used to obtain the physical property data, chemical composition data and surface image data of the target steel through different acquisition devices, and transmit them to the steel data preprocessing module.

[0054] It should be noted that the specific data of target steel collected in the steel data collection module include:

[0055] Physical property data: temperature during heating, rolling and cooling, pressure during rolling, thickness data of steel, width and length of steel;

[0056] Chemical composition data: carbon content in the target steel, manganese content in the target steel, silicon content in the target steel, phosphorus content in the target steel, sulfur content in the target steel, and other chemical element contents in the target steel;

[0057] Surface image data: relative positioning of each pixel point of the target steel image, and grayscale value of each pixel point of the target steel image.

[0058] It should be noted that the specific method for obtaining the physical property data of the target steel in the steel data acquisition module is as follows:

[0059] A data acquisition node that can accurately monitor the corresponding physical properties is deployed at intervals of 10 cm at each key position of the steel production line; the data acquisition node collects and transmits data in real time and performs appropriate signal processing to ensure that the data collected from different monitoring devices are synchronized in time; the sensor nodes that can accurately monitor the corresponding physical properties include using infrared sensors to monitor the temperature at points on the production line in real time, obtaining the pressure data of the rolling mill through pressure sensors, and continuously detecting the thickness of the target steel through ultrasonic thickness gauges.

[0060] It should be noted that the specific method for obtaining the chemical composition data of the target steel in the steel data acquisition module is as follows:

[0061] After confirming the length of the target steel, use a saw to cut the starting and ending positions of the target steel into small pieces of 5cm×5cm respectively; use a polishing machine to grind the surface of the small piece of steel flat and remove the oxide layer and other impurities; grind the ground small piece of steel into powder and use cold pressing technology to press the powder into a film for spectral analysis, collect and record the emission or absorption spectral data, and identify and quantify the various elements in the steel; the spectral data includes the carbon content that affects the strength and hardness of the steel, and the content of chemical elements that affect the toughness, weldability and corrosion resistance of the steel.

[0062] It should be noted that the specific method for acquiring the target steel surface image data in the steel data acquisition module is as follows:

[0063] Install high-definition cameras and lighting equipment on the steel production line to obtain surface images of the target steel at different positions and angles; pre-process the acquired images, including adjusting contrast, denoising, edge enhancement, etc., to improve the recognizability of defect features.

[0064] The steel data preprocessing module is used to clean and standardize the data transmitted by the steel data acquisition module and eliminate different dimensions and magnitudes, extract the quality characteristics of the target steel from the original image and data, and generate a characteristic data set of the steel; the characteristic data set of the steel includes physical property data, chemical composition data, and image data.

[0065] The specific processing method corresponding to the steel data preprocessing module is:

[0066] The steel data preprocessing module preprocesses the physical property data, and the calculation formula is specifically expressed as:

[0067] ,

[0068] Among them, P i represents the preprocessed value corresponding to the ith physical property data after removing outliers, and m represents the value for calculating P i The data window size is , l is represented by the index of the physical property data point, ranging from (i-(m / 2)) to (i+(m / 2)), p i It is represented as the i-th physical property data after removing outliers;

[0069] The steel data preprocessing module preprocesses the chemical composition data, and the calculation formula is specifically expressed as:

[0070] ,

[0071] Among them, C j Represents the preprocessed value of the jth chemical composition data, c j Represented as the jth chemical composition data, min (c j ) represents the minimum value of all chemical composition data, s represents the standard deviation of chemical composition data, max(c j ) is represented by the maximum value of all chemical composition data, and μ is represented by the average value of chemical composition data;

[0072] The steel data preprocessing module preprocesses the image data of each surface, and the calculation formula is specifically expressed as:

[0073] ,

[0074] Among them, I k It is represented as the preprocessed value of the kth pixel in the image, k represents the pixel, f represents the image direction, a f Expressed as the weight coefficient related to direction f, I k+f It is expressed as the intensity value of the positive neighbor pixel of pixel k in direction f, I k-fDenoted as the intensity value of the negative neighbor pixel of pixel point k in direction f, I k+nf Denoted as the intensity value that may represent the positive neighbor pixel of pixel point k in the diagonal direction nf, I k-nf Denoted as the intensity value that may represent the negative neighbor pixel of pixel point k in the diagonal direction nf, and N is denoted as the offset in the diagonal direction.

[0075] In this embodiment, the calculation formula for the preprocessing of each physical property data by the steel data preprocessing module will perform data cleaning and smoothing on the data points within the quartile range; (m / 2) in the calculation formula is denoted as half of the data window to ensure the symmetry of each physical data point; the data window size is denoted as a subset of the physical property data set, which contains the data points collected within the specified range; p i Only when it meets the value condition, calculate the preprocessing value P corresponding to the i-th physical property data after removing the outliers i , and the value condition is (Q1 - 1.5×IQR < p i < Q3 + 1.5×IQR), where Q1 is denoted as the first quartile of the physical property data, IQR is denoted as the degree of dispersion of the physical property data, denoted as Q3 - Q1, and Q3 is denoted as the third quartile of the physical property data; the quartiles are denoted as dividing all physical property data into four equal-sized point sets.

[0076] In this embodiment, the calculation formula for the preprocessing of each chemical composition data by the steel data preprocessing module will perform normalization and standardization on the chemical composition data. In the formula, (c j - min(c j )) / ((max(c j )) - min(c j )) is denoted as the normalization operation on the chemical composition data.

[0077] In this embodiment, the calculation formula for the preprocessing of each surface image data by the steel data preprocessing module performs weighted summation on the difference of pixel values in the direction, and then takes the square root to obtain a comprehensive metric, highlighting the texture features in the image to improve the recognizability of defect features. Among them, the weight coefficient related to direction f is used to weight the pixel values in different directions.

[0078] In this embodiment, the steel data preprocessing module can convert the original physical property data, chemical composition data, and surface image data into useful information for quality control and performance prediction, providing a solid data foundation for the optimization and quality improvement of the steel production process.

[0079] The physical feature extraction module is used to analyze the physical property data of the target steel transmitted by the steel data preprocessing module, calculate the mechanical properties of the target steel through the physical feature extraction formula, construct the physical feature matrix, and transmit it to the steel quality comprehensive analysis module.

[0080] It should be noted that the physical feature extraction and calculation formula are specifically expressed as:

[0081] ,

[0082] Among them, PFE represents the physical feature matrix of the target steel, n represents the total number of physical property data, PDM represents the physical data matrix of the target steel, is represented as the regression coefficient vector, PTV is represented as the quality index of standard steel, λ is represented as the regularization parameter, represents the L1 norm, Refers to the sum of the absolute values ​​of the regression coefficient vector; represents the L2 norm, Refers to the size of the difference between the physical feature matrix and the actual value.

[0083] It should be noted that the calculation formula for physical feature extraction in the physical feature extraction module minimizes the physical feature matrix of the target steel by finding the minimum regression coefficient vector, ensuring the fit to the known data and improving the generalization and interpretability of the formula. The physical data matrix of the target steel contains the physical property data of the steel, each row represents a location point, and each column represents a physical property. The regression coefficient vector is used to indicate the degree of influence of each physical property on the quality of the steel, and the regularization parameter value is used to control the degree of fit of the formula. It is used to use the L2 norm, that is, the Euclidean norm, to measure the difference between the physical feature matrix and the actual value. Used to use the L1 norm, which represents the sum of the absolute values ​​of the regression coefficient vector.

[0084] The chemical composition quantitative analysis module is used to construct a matrix of quantitative concentrations of each element in the target steel through the chemical composition quantitative analysis formula, using the chemical composition data of the target steel collected by spectral analysis technology and in a unified format by the steel data preprocessing module, and transmit it to the steel quality comprehensive analysis module.

[0085] It should be noted that the quantitative analysis of chemical composition is specifically expressed as follows:

[0086] ,

[0087] Where Y represents the matrix composed of the chemical composition concentration of the target steel, X represents the matrix composed of the spectral data of the target steel, and t jIt is represented as the j-th element score of matrix X, w j It is represented as the loading of the jth element of the matrix X, (B1, B2, ..., B j） It is represented as the score matrix shared by matrix X and matrix Y, q j Represented as the weight of the j-th element in the matrix Y.

[0088] It should be noted that the j-th element score of the matrix X in the calculation formula of the chemical composition quantitative analysis module represents the weighted linear combination of the original spectral data, and the calculation formula is expressed as:

[0089] t j =X×w j ',

[0090] Among them, w j ' represents the weight of the jth element in the matrix X; the load of the jth element in the matrix X represents the relationship between the original spectral data and the element score, and the calculation formula is expressed as:

[0091] ;

[0092] The weight of the jth element in the matrix Y represents the relationship between the chemical component concentration and the element score, and the calculation formula is expressed as:

[0093] ;

[0094] The score matrix shared by matrix X and matrix Y is calculated as follows:

[0095] .

[0096] In this embodiment, the chemical composition quantitative analysis module establishes a quantitative connection between chemical composition concentration and spectral data by accurately reflecting the characteristic spectral intensity of the elements in the target steel and the correlation between the quantified spectral data and the element scores, thereby achieving effective conversion from spectral data to chemical composition concentration.

[0097] The image analysis module is used to analyze the image data transmitted by the steel data preprocessing module, evaluate the surface quality of the steel through the image texture feature analysis formula, and transmit the results to the steel quality comprehensive analysis module.

[0098] It should be noted that the image texture feature calculation formula is specifically expressed as:

[0099] ,

[0100] Where ITFE is expressed as the conjugate matrix of grayscale under the condition of distance d and angle θ, α and β are the grayscale index of image length and width respectively, δ is the Kronecker function, ηk and η k ' represents the grayscale value corresponding to pixel k and its adjacent pixels, g(α, β) represents the probability of grayscale appearing at position (α, β), g(α+(d×cosθ), β+(d×sinθ)) represents the probability of grayscale appearing at position (α, β) after being offset by distance d and angle θ, cos(θ) and sin(θ) represent the cosine and sine values ​​of angle θ.

[0101] In this embodiment, the calculation formula in the image analysis module analyzes ITFE at different directions and distances to quantify the texture characteristics of the steel surface, thereby comprehensively evaluating the surface quality of the steel, wherein under the conditions of distance d and angle θ, the conjugate matrix of the gray level is used to store the probability of occurrence of different gray level pairs (α, β) at a specific direction θ and distance d; the Kronecker function is used to indicate whether two parameters are equal, and the value is 1 when the parameters are equal, otherwise it is 0.

[0102] In this embodiment, the image analysis module quantifies the roughness of the steel surface by analyzing the texture features of the image, and evaluates the surface quality of the steel, thereby improving the quality control level of the steel production process; by analyzing the grayscale co-occurrence matrix in different directions and distances, it can more carefully capture the subtle changes in the steel surface, determine whether there are defects on the steel surface, and improve the efficiency of quality control.

[0103] The steel quality comprehensive analysis module is used to integrate the matrix data transmitted by the physical feature extraction module, the chemical composition quantitative analysis module and the image analysis module, conduct a comprehensive evaluation of the overall quality of the target steel, generate a steel quality comprehensive analysis index, identify potential quality risks and transmit them to the steel quality judgment module.

[0104] It should be noted that the calculation formula of the comprehensive analysis index of steel quality is specifically expressed as:

[0105] ,

[0106] Among them, CSQ represents the comprehensive analysis index of target steel quality, PFE represents the physical characteristic matrix of target steel, Y represents the matrix composed of chemical component concentration of target steel, ITFE represents the conjugate matrix of gray level under the conditions of distance d and angle θ, χ1, χ2, χ3 represent the weight coefficient of the physical characteristic matrix of target steel, the weight coefficient of the matrix composed of chemical component concentration of target steel and the weight coefficient of the conjugate matrix of gray level under the conditions of distance d and angle θ, respectively, f(PFE), f(Y), f(ITFE) represent the processing functions of the physical characteristic matrix of target steel, the processing functions of the matrix composed of chemical component concentration of target steel and the processing functions of the conjugate matrix of gray level under the conditions of distance d and angle θ, respectively.

[0107] It should be noted that the weights of each matrix in the calculation formula of the comprehensive analysis index of steel quality are assigned by applying the MCDM method; the processing function of each matrix is ​​used to convert the raw data into a standardized score ranging from 0 to 100, so that data from different sources and types can be directly compared.

[0108] The steel quality judgment module is used to judge the comprehensive analysis index of steel quality transmitted by the comprehensive evaluation module. Through the predefined steel quality threshold, it realizes automatic judgment of the steel quality grade and generates corresponding signals to be transmitted to the report and output module.

[0109] It should be noted that the specific evaluation method corresponding to the steel quality judgment module is as follows:

[0110] The data transmitted by the comprehensive evaluation module is compared with a series of pre-set steel quality risk thresholds, and combined with predefined scoring rules to determine the overall quality grade of the steel, convert the quality grade into a corresponding signal, and send the generated quality grade signal to the report and output module; the predefined scoring rules include that if the target steel quality comprehensive analysis index is between 90 and 100, the steel quality is excellent and the A signal is transmitted; if the target steel quality comprehensive analysis index is between 75 and 90, the steel quality is good and the B signal is transmitted; if the target steel quality comprehensive analysis index is between 60 and 75, the steel quality is qualified and the C signal is transmitted; if the target steel quality comprehensive analysis index is less than 60, the steel quality is unqualified and the D signal is transmitted.

[0111] In this embodiment, the steel quality judgment module receives data from the comprehensive evaluation module, and classifies the steel into quality grades according to the established quality standards and risk thresholds, and the classification results are transmitted to the report and output module in the form of signals; the steel quality judgment module reduces the interference of human factors through an automated judgment mechanism, enhances the consistency and reliability of the evaluation results, and provides real-time and accurate data support, thereby optimizing the entire steel production and quality management process.

[0112] The report and output module is used to generate a corresponding quality report based on the content of the signal after receiving the grade signal transmitted by the steel quality judgment module, including the detailed evaluation results, quality grade and recommended measures of the target steel.

[0113] In this embodiment, the final report is saved in the form of an electronic document for quality management personnel to review; the detailed evaluation results of the target steel are based on preset quality standards and historical data analysis, suggestions for optimizing the production process, defining decision rules and providing targeted quality improvement plans and process adjustment suggestions, and the analysis results and suggestions are fed back to relevant controls so that corrective measures can be taken quickly.

[0114] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0115] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A steel quality analysis and feedback system, It is characterized in that include: Steel data acquisition module: used to obtain physical property data, chemical composition data and surface image data of the target steel through different acquisition devices, and transmit them to the steel data preprocessing module; The method for obtaining the physical property data of the target steel is as follows: A data acquisition node that can accurately monitor the corresponding physical properties is deployed at intervals of 10 cm at each key position of the steel production line; the data acquisition node collects and transmits data in real time and performs appropriate signal processing to ensure that the data collected from different monitoring devices are synchronized in time; the sensor nodes that can accurately monitor the corresponding physical properties include using infrared sensors to monitor the temperature in real time on the production line, obtaining the pressure data of the rolling mill through pressure sensors, and continuously detecting the thickness of the target steel through ultrasonic thickness gauges; The method for obtaining the chemical composition data of the target steel is as follows: After confirming the length of the target steel, use a sawing machine to cut the starting and ending positions of the target steel into small pieces of 5cm×5cm respectively; use a polishing machine to grind the surface of the small piece of steel flat and remove the oxide layer and other impurities; grind the small piece of steel after grinding into powder and use cold pressing technology to press the powder into a film for spectral analysis, collect and record the emission or absorption spectral data, and identify and quantify the elements in the steel; the spectral data includes the carbon content that affects the strength and hardness of the steel, and the content of chemical elements that affect the toughness, weldability and corrosion resistance of the steel; The specific method for acquiring the target steel surface image data is as follows: Install high-definition cameras and lighting equipment on the steel production line to obtain target steel surface images at different positions and angles; pre-process the acquired images, including adjusting contrast, denoising, and edge enhancement to improve the recognizability of defect features; Steel data preprocessing module: used to clean and standardize the data transmitted by the steel data acquisition module and eliminate different dimensions and magnitudes, extract the quality characteristics of the target steel from the original image and data, and generate a characteristic data set of the steel; the characteristic data set of the steel includes physical property data, chemical composition data, and image data; After preprocessing the data of each physical attribute, the calculation formula is specifically expressed as: ， Among them, P i represents the preprocessed value corresponding to the ith physical property data after removing outliers, and m represents the value for calculating P i The data window size is , l is represented by the index of the physical property data point, ranging from (i-(m / 2)) to (i+(m / 2)), p i It is represented as the i-th physical property data after removing outliers; The data of each chemical component is preprocessed and the calculation formula is specifically expressed as: ， Among them, C j Represents the preprocessed value of the jth chemical composition data, c j Represented as the jth chemical composition data, min (c j ) represents the minimum value of all chemical composition data, s represents the standard deviation of chemical composition data, max(c j ) is represented by the maximum value of all chemical composition data, and μ is represented by the average value of chemical composition data; The calculation formula for preprocessing the image data of each surface is specifically expressed as: ， Among them, I k It is represented as the preprocessed value of the kth pixel in the image, k represents the pixel, f represents the image direction, a f Expressed as the weight coefficient associated with direction f, I k+f It is expressed as the intensity value of the positive neighbor pixel of pixel k in direction f, I k-f It is represented by the intensity value of the negative neighbor pixel of pixel k in direction f, I k+Nf It is expressed as the intensity value of the positive neighbor pixel of pixel k in the diagonal direction Nf, I k-Nf It is expressed as the intensity value of the negative neighbor pixel of pixel k in the diagonal direction Nf, where N is the offset in the diagonal direction; Physical feature extraction module: used to analyze the physical property data of the target steel transmitted by the steel data preprocessing module, calculate the mechanical properties of the target steel through the physical feature extraction formula, construct the physical feature matrix, and transmit it to the steel quality comprehensive analysis module; The physical feature extraction in the physical feature extraction module is specifically expressed as follows: ， Among them, PFE represents the physical feature matrix of the target steel, n represents the total number of physical property data, PDM represents the physical data matrix of the target steel, ϵ represents the regression coefficient vector, PTV represents the quality index of the standard steel, λ represents the regularization parameter, refers to the sum of the absolute values ​​of the regression coefficient vector, Refers to the size of the difference between the physical feature matrix and the actual value; Chemical composition quantitative analysis module: used to construct a matrix of quantitative concentrations of each element in the target steel using the chemical composition quantitative analysis formula, which is collected by spectral analysis technology and in a unified format by the steel data preprocessing module, and transmit it to the steel quality comprehensive analysis module; The calculation formula for quantitative analysis of chemical components in the chemical composition quantitative analysis module is specifically expressed as: ， Where Y represents the matrix composed of the chemical composition concentration of the target steel, X represents the matrix composed of the spectral data of the target steel, and t j It is represented as the j-th element score of matrix X, w j It is expressed as the load of the jth element of the matrix X, (B 1 , B 2 , …, B j） It is represented as the score matrix shared by matrix X and matrix Y, q j Represented as the weight of the j-th element in the matrix Y; Image analysis module: used to analyze the image data transmitted by the steel data preprocessing module, evaluate the surface quality of the steel through the image texture feature analysis formula, and transmit the results to the steel quality comprehensive analysis module; The calculation formula of image texture features in the image analysis module is specifically expressed as: ， Where ITFE is expressed as the conjugate matrix of grayscale under the condition of distance d and angle θ, α and β are the grayscale index of image length and width respectively, δ is the Kronecker function, η k and η k ' represents the grayscale value corresponding to pixel k and its adjacent pixels, g(α, β) represents the probability of grayscale appearing at position (α, β), g(α+(d×cosθ), β+(d×sinθ)) represents the probability of grayscale appearing at position (α, β) after offsetting by distance d and angle θ, cos(θ) and sin(θ) represent the cosine and sine values ​​of angle θ; Steel quality comprehensive analysis module: used to integrate the matrix data transmitted by the physical feature extraction module, chemical composition quantitative analysis module and image analysis module, conduct a comprehensive evaluation of the overall quality of the target steel, generate a comprehensive analysis index of steel quality, identify potential quality risks and transmit them to the steel quality judgment module; Steel quality judgment module: used to judge the comprehensive analysis index of steel quality transmitted by the comprehensive evaluation module, realize automatic judgment of steel quality grade through predefined steel quality threshold, and generate corresponding signals to be transmitted to the report and output module; Report and output module: After receiving the grade signal transmitted by the steel quality judgment module, it is used to generate a corresponding quality report according to the content of the signal, including the detailed evaluation results, quality grade and recommended measures of the target steel.

2. A steel quality analysis and feedback system according to claim 1, Features: The calculation formula of the steel quality comprehensive analysis index in the steel quality comprehensive analysis module is specifically expressed as: ， Among them, CSQ is represented by the comprehensive analysis index of target steel quality, PFE is represented by the physical characteristic matrix of target steel, Y is represented by the matrix composed of chemical composition concentration of target steel, ITFE is represented by the conjugate matrix of gray level under the conditions of distance d and angle θ, χ 1 , χ 2 , χ 3 They are respectively represented as the weight coefficients of the physical characteristic matrix of the target steel, the weight coefficients of the matrix composed of the chemical component concentration of the target steel, and the weight coefficients of the conjugate matrix of the gray level under the conditions of distance d and angle θ. f(PFE), f(Y), and f(ITFE) are respectively represented as the processing functions of the physical characteristic matrix of the target steel, the processing functions of the matrix composed of the chemical component concentration of the target steel, and the processing functions of the conjugate matrix of the gray level under the conditions of distance d and angle θ.

3. A steel quality analysis and feedback system according to claim 1, Features: The evaluation method in the steel quality judgment module is as follows: The data transmitted by the comprehensive evaluation module is compared with a series of pre-set steel quality risk thresholds, and combined with the pre-defined scoring rules, the overall quality grade of the steel is determined, the quality grade is converted into a corresponding signal, and the generated quality grade signal is sent to the report and output module; the pre-defined scoring rules include that if the target steel quality comprehensive analysis index is between 90 and 100, the steel quality is excellent and the A signal is transmitted; If the comprehensive analysis index of target steel quality is between 75 and 90, the steel quality is good and the B signal is transmitted; If the comprehensive analysis index of target steel quality is between 60 and 75, the steel quality is qualified and the C signal is transmitted; If the target steel quality comprehensive analysis index is less than 60, the steel quality is unqualified and a D signal is transmitted.

Images