A strip steel cross-process data set construction method for quality control

Abstract

The application provides a strip steel cross-process data set construction method for quality control, and relates to the technical field of steel data analysis and processing, which comprises the following steps: collecting time domain process data of a target strip steel in each process; determining a cumulative length coordinate function of the target strip steel in each process based on an instantaneous linear velocity function of the target strip steel in each process; determining length domain process data of the target strip steel based on an inverse mapping relationship of the cumulative length coordinate function and the time domain process data, wherein the length domain process data represents a corresponding relationship between the target strip steel and process data in the length direction; and constructing a cross-process data set according to the length domain process data. The application can establish a data set that accurately reflects the characteristics of the whole strip steel production process.

Description

Technical Field

[0001] This invention relates to the field of steel data analysis and processing technology, specifically a method for constructing cross-process datasets for strip steel for quality control. Background Technology

[0002] The steel industry is a crucial foundation of the national economy, occupying a vital position within the economic system. With the continuous improvement of material performance requirements in the manufacturing industry, the demand for high-strength steel, high-performance steel, and other high-end steel products continues to grow, placing higher demands on the quality stability and consistency of the steel production process. The production process of strip steel products spans multiple stages and information systems, involving hot rolling, leveling, cold rolling, continuous annealing, and galvanizing. Complex process coupling and quality transfer relationships exist between these stages, and relevant process parameters and process conditions jointly determine the final product's microstructure, dimensional accuracy, and surface quality.

[0003] Traditional steel production quality control mainly focuses on optimizing single sections or processes, which makes it difficult to fully reflect the collaborative relationships between different production links. Therefore, it is impossible to establish a dataset that can accurately reflect the characteristics of the entire strip steel production process. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to propose a method for constructing a cross-process dataset for strip steel production oriented towards quality control, which can establish a dataset that accurately reflects the characteristics of the entire strip steel production process.

[0005] The technical solutions for the above-mentioned method for constructing cross-process datasets for strip steel for quality control include:

[0006] Collect time-domain process data of the target strip steel in each process;

[0007] Based on the instantaneous linear velocity function of the target strip in each process, the cumulative length coordinate function of the target strip in each process is determined;

[0008] Based on the inverse mapping relationship of the cumulative length coordinate function and the time domain process data, the length domain process data of the target strip is determined. The length domain process data represents the correspondence between the target strip and the process data in the length direction.

[0009] A cross-process dataset is constructed based on the process data in the length field.

[0010] Furthermore, the above-mentioned method for constructing a cross-process dataset for strip steel for quality control also includes:

[0011] Obtain the target strip steel into the process The length of the cut not removed, the length of the inlet cut head, the length of the inlet cut tail, the length of the outlet cut head, and the length of the outlet cut tail;

[0012] The target strip is determined in the process based on the uncut length, the inlet cut length, the inlet cut tail length, the outlet cut length, and the outlet cut tail length. The effective length of time is as follows:

[0013] ;

[0014] in, Indicates the process The effective length, Indicates the process Unresected length Indicates the process The length of the inlet cut-off head, Indicates the process The length of the exit cut-off head, Indicates the process The length of the cut-off at the entrance. Indicates the process The length of the cut-off tail at the outlet;

[0015] The effective interval of the cumulative length coordinate function corresponding to the target process is determined, and the effective interval is expressed as:

[0016] ;

[0017] in, Indicates the process The corresponding cumulative length coordinate function.

[0018] Furthermore, the above-mentioned method for constructing a cross-process dataset for strip steel for quality control also includes:

[0019] The process is constructed within the effective range. The corresponding normalized cumulative length coordinate function;

[0020] According to the process The effective length, inlet cutter length, and outlet cutter length, and the process described herein. The corresponding normalized cumulative length coordinate function is used to establish the process. With the process The length mapping relationship between them, the process This refers to the process described. The preceding adjacent process, the process The corresponding effective length is equal to the process. The corresponding uncut lengths, the length mapping relationship is as follows:

[0021] ;

[0022] in, Indicates the process The corresponding normalized cumulative length coordinate function, Indicates the process The effective length, Indicates the process The length of the inlet cut-off head, Indicates the process The length of the cut-off head at the outlet.

[0023] Furthermore, the construction is within the effective range, and the process... The corresponding normalized cumulative length coordinate function includes:

[0024] ;

[0025] in, This represents the normalized cumulative length coordinate function. .

[0026] Furthermore, within the effective range of the construction, the process... Following the normalized cumulative length coordinate function corresponding to the cumulative length coordinate function, it also includes:

[0027] Determined in the process and said process Between these points, the number of times the target strip reverses along its length;

[0028] The process is based on the number of reversals. The corresponding normalized cumulative length coordinate function is corrected as follows:

[0029] ;

[0030] in, This refers to the revised process described above. The corresponding normalized cumulative length coordinate function.

[0031] Further, determining the cumulative length coordinate function of the target strip in each process based on the instantaneous linear velocity function of the target strip in each process includes:

[0032] ;

[0033] in, Indicates the process The corresponding cumulative length coordinate function, This indicates that the target strip steel is in the process. The instantaneous linear velocity function in Indicates the process The starting time;

[0034] The determination of the length domain process data of the target strip based on the inverse mapping relationship of the cumulative length coordinate function and the time domain process data includes:

[0035] ;

[0036] in, Indicates the process The corresponding inverse function of the cumulative length coordinate function or the numerical equivalent mapping of the inverse function, A function representing the time-domain process data. A function representing the process data in the length field.

[0037] Furthermore, constructing a cross-process dataset based on the length field process data includes:

[0038] The length domain process data of the target strip in each process are concatenated;

[0039] Obtain the smelting composition data of the molten steel used to smelt the target strip, and associate the smelting composition data with the target strip;

[0040] Acquire performance test data of the target strip at various length positions, and correlate the performance test data with the length positions of the target strip;

[0041] Using the steel coil identifier corresponding to the target strip and the coordinates of the target strip in the length direction as indexes, the cross-process dataset is constructed based on the length domain process data, the smelting composition data, and the performance testing data.

[0042] The technical solution provided by this invention collects time-domain process data of the target strip steel in each process, determines the cumulative length coordinate function of the target strip steel in each process based on the instantaneous linear velocity function of the target strip steel in each process, and determines the length-domain process data of the target strip steel based on the inverse mapping relationship of the cumulative length coordinate function and the time-domain process data. The length-domain process data represents the correspondence between the target strip steel and the process data in the length direction. A cross-process dataset is constructed based on the length-domain process data. The method for constructing a cross-process dataset for strip steel for quality control provided by this invention establishes a unified length coordinate system to accurately align the process data of strip steel collected in different processes and at different times, thereby constructing a cross-process dataset that accurately reflects the characteristics of the entire strip steel production process. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of a method for constructing a cross-process dataset for strip steel for quality control, as described in an embodiment of the present invention.

[0044] Figure 2 This is a diagram illustrating the overall architecture for collecting and integrating cross-process data of strip steel in an embodiment of the present invention.

[0045] Figure 3 This is a schematic diagram illustrating the process of converting time-domain data into length-domain data in an embodiment of the present invention.

[0046] Figure 4 This is a schematic diagram of strip steel cross-process length mapping in an embodiment of the present invention;

[0047] Figure 5 This is a schematic diagram of a cross-process data association structure in an embodiment of the present invention;

[0048] Figure 6 This is a schematic diagram of the results of a cross-process data set for strip steel for quality control in an embodiment of the present invention. Detailed Implementation

[0049] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0050] To address the problems existing in the prior art, this invention provides a method for constructing a cross-process dataset for strip steel for quality control. Figure 1 This is a schematic diagram of a method for constructing a cross-process dataset for strip steel for quality control, as described in an embodiment of the present invention. Figure 1 The method provided by this invention may include:

[0051] Step 1: Collect time-domain process data of the target strip steel in each process.

[0052] For example, time-domain process data refers to the process data of strip steel being recorded sequentially in each process of production, with time as the dimension. This process data may include parameters such as rolling speed, rolling force, temperature, and tension, which change over time and are recorded sequentially in this step.

[0053] In some examples, Figure 2 This is a diagram illustrating the overall architecture for the acquisition and integration of cross-process data for strip steel in an embodiment of the present invention. Figure 2 As shown, step 1 may include:

[0054] Step 1.1: Obtain the time-domain process data through the database of the production control system corresponding to each process and the production application.

[0055] For example, by obtaining the database connection method and application programming interface connection method of the production control system corresponding to each production process, unified access to multi-source production data can be achieved.

[0056] It should be noted that the production process of strip steel usually includes multiple continuous and sequential production steps such as hot rolling, leveling, cold rolling, and continuous annealing or galvanizing. The production process data of each step are stored in the corresponding production control system.

[0057] It should be noted that the production data for strip steel includes historical production data and real-time production data. Historical process data can be read from the production control database of each process through database connection, while real-time process data can be obtained through application programming interfaces or other data communication methods.

[0058] This step enables the unified collection of process data for strip steel in different processes, thereby providing a basic data source for subsequent data processing and cross-process data alignment.

[0059] Step 1.2: Perform ETL processing on the time-domain process data.

[0060] For example, ETL processing includes data extraction, data transformation, and data loading processes. By cleaning, transforming, and storing data from different sources, it enables the data to meet the needs of subsequent cross-process data alignment and dataset construction.

[0061] The above ETL processing may include:

[0062] (1) Batch extraction and preprocessing of historical process data:

[0063] The historical process data stored in the aforementioned production control database is preprocessed and cleaned, including removing redundant data, eliminating outliers, filling in missing values, and standardizing data, to obtain structured data, and the processed data is written into the data storage system.

[0064] (2) Streaming processing of real-time process data:

[0065] For real-time process data, the data stream is buffered using a message queue and processed online using a streaming ETL approach. This streaming ETL approach is based on a streaming computing framework, such as Apache Flink. This processing performs noise reduction, anomaly detection, and data format standardization on high-frequency time-domain process data, thereby ensuring consistent data structure and quality across different production processes.

[0066] (3) Data loading and categorized storage:

[0067] Depending on the data type of the process data, it is loaded into different data storage systems. For example, high-frequency time-domain process data is written to an industrial time-series database (such as Tdengine) to support compressed storage and fast querying of high-frequency sampling point data; while mixed non-time-domain data is written to a relational database (such as MySQL) to support structured storage and subsequent correlation analysis of process parameter data, test data, and production status data.

[0068] Step 2: Based on the instantaneous velocity function of each process, determine the cumulative length coordinate function of the target strip in each process.

[0069] For example, the processing time and content of the strip steel differ in different processes, and correspondingly, the length of the strip steel processed in different processes also differs. Based on the instantaneous velocity function of each process, the function of the strip steel's length coordinate versus time in different processes is derived.

[0070] In some examples, step 2 may include:

[0071] ;

[0072] in, Indicates the process The corresponding cumulative length coordinate function, This indicates that the target strip steel is in the process. The corresponding instantaneous linear velocity function, Indicates the process The starting moment.

[0073] It should be noted that in the actual production of strip steel, process data is typically collected using high-frequency sampling to form discrete time series data. In different processes, this discrete time series data forms the instantaneous linear velocity function of the strip steel at the corresponding process. Based on the operating time of each process, the instantaneous linear velocity function corresponding to each process is integrated to obtain the cumulative length coordinate function for that process.

[0074] Step 3: Based on the inverse mapping relationship of the cumulative length coordinate function and the time domain process data, determine the length domain process data of the target strip. The length domain process data represents the correspondence between the target strip and the process data in the length direction.

[0075] For example, the cumulative length coordinate function reflects the mapping from the time series to the length coordinate, while the inverse mapping relationship reflects the mapping from the length coordinate to the time series. By constructing a composite function relationship between this inverse mapping relationship and the collected time series process data, the functional relationship between the length coordinate and the process data can be obtained.

[0076] It should be noted that the length direction of the strip refers to the direction of the process.

[0077] In some examples, step 3 may include:

[0078] ;

[0079] in, Indicates the process The corresponding inverse function of the cumulative length coordinate function or the numerical equivalent mapping of the inverse function, A function representing the time-domain process data. A function representing the process data in the length field.

[0080] For example, the above inverse mapping relationship can be established by the inverse function of the cumulative length coordinate function or the numerical equivalent mapping of the inverse function.

[0081] Figure 3 This is a schematic diagram illustrating the process of converting time-domain data into length-domain data in an embodiment of the present invention, as shown below. Figure 3 As shown, the embodiment of the present invention completes the conversion of time domain data to length domain data through steps 2 and 3, converting the time series process parameter data into data distributed along the length direction of the strip, so that a one-to-one correspondence is established between various process parameters and the physical position of the strip in the length direction, thereby providing a unified length coordinate reference for subsequent cross-process length domain data alignment and multi-process data fusion.

[0082] Understandably, in actual production, strip steel is typically trimmed at the beginning and end when entering and leaving each process to remove sections with unstable quality. This results in variations in the effective length of the strip steel between different processes, leading to inaccurate physical positional correspondences between them.

[0083] In view of this, the present invention also includes step 4, which provides a normalization method for the cumulative length coordinate function, and on this basis, establishes the alignment of length domain process data across processes. Figure 4 This is a schematic diagram of strip steel cross-process length mapping in an embodiment of the present invention, combined with... Figure 4 Step 4 above can be performed as follows:

[0084] Step 4.1: Obtain the uncut length, inlet cut length, inlet cut tail length, outlet cut length, and outlet cut tail length of the target strip when it enters process i.

[0085] For example, parameters such as uncut length, inlet cut length, inlet cut tail length, outlet cut length, and outlet cut tail length can be obtained from the production control system or process record data of the corresponding production process.

[0086] Step 4.2: Determine the target strip length in the process based on the uncut length, the inlet cut length, the inlet cut tail length, the outlet cut length, and the outlet cut tail length. The effective length of time is as follows:

[0087] ;

[0088] in, Indicates the process The effective length, Indicates the process Unresected length Indicates the process The length of the inlet cut-off head, Indicates the process The length of the exit cut-off head, Indicates the process The length of the cut-off at the entrance. Indicates the process The length of the cut-off tail at the exit.

[0089] Step 4.3, determine the effective interval of the cumulative length coordinate function corresponding to the target process, wherein the effective interval is represented as:

[0090] ;

[0091] in, Indicates the process The corresponding cumulative length coordinate function.

[0092] Through the processing in steps 4.1 to 4.3, it is ensured that the corresponding cumulative length coordinate function in each process falls only within the valid range, thus avoiding invalid data and improving the reliability of the obtained length domain process data.

[0093] In some examples, considering the different operational dimensions of strip steel in different processes, embodiments of the present invention construct normalized length coordinates to facilitate data alignment between different processes. The method provided by the present invention also includes:

[0094] Step 4.4, construct within the effective range, the process The corresponding normalized cumulative length coordinate function.

[0095] It should be noted that after normalization, the cumulative length coordinate function of all processes is within the same scale.

[0096] In some examples, step 4.4 includes:

[0097] ;

[0098] in, This represents the normalized cumulative length coordinate function. .

[0099] In the continuous production of strip steel, the strip length between adjacent production processes has a succession relationship. Specifically, the process... After processing and trimming, the effective strip length output will be transferred to the next process. and as a process The initial strip length upon entry satisfies the following relationship:

[0100] ;

[0101] in, This indicates the process. The effective length of the strip steel, This indicates that the strip steel enters the process. The length not removed at that time.

[0102] Under the premise that the above formula holds true, construct the coordinate mapping relationship of strip length between two adjacent processes.

[0103] Step 4.5 According to the process The effective length, inlet cutter length, and outlet cutter length, and the process described herein. The corresponding normalized cumulative length coordinate function is used to establish the process. With the process The length mapping relationship between them, the process This refers to the process described. The length mapping relationship for the preceding adjacent process is as follows:

[0104] ;

[0105] in, Indicates the process The corresponding normalized cumulative length coordinate function, Indicates the process The effective length, Indicates the process The length of the inlet cut-off head, Indicates the process The length of the cut-off head at the outlet.

[0106] It should be noted that step 4.5 achieves an accurate correspondence of the physical position of the strip between adjacent processes, which can then lead to the correspondence of the physical position of the strip between all processes, thereby achieving length field alignment of cross-process data.

[0107] In some examples, considering that in actual production processes, the strip steel may move in opposite directions between different processes, for example, due to the reversal of the strip steel length direction caused by the coiling or uncoiling process, it is necessary to correct the orientation of the normalized length coordinates. Therefore, after step 4.4, the method provided by the present invention further includes:

[0108] Step 4.41, determine in the process and said process The number of times the target strip reverses along its length.

[0109] Step 4.42, adjust the process according to the number of reversals. The corresponding normalized cumulative length coordinate function is corrected as follows:

[0110] ;

[0111] in, This refers to the revised process described above. The corresponding normalized cumulative length coordinate function, This indicates the number of reversals.

[0112] In some examples, after steps 4.41 and 4.42, the corrected normalized cumulative length coordinate function is obtained. Substituting this corrected normalized cumulative length coordinate function into the equation, the length mapping relationship in step 4.5 can then be expressed as follows:

[0113] ;

[0114] The method provided by this invention, through steps 1 to 4, completes the process from collecting time domain data of strip steel in production operations, to converting time domain data into length domain data, and then unifying the length domain data of each process under the same scale, providing data support for subsequent cross-process data fusion and dataset construction.

[0115] Step 5: Construct a cross-process dataset based on the length field process data.

[0116] It should be noted that, based on various identification information in the production process, the process data obtained from multiple processes can be further correlated and organized to construct a cross-process related data structure covering the entire strip steel production process, thereby realizing unified management and analysis of process parameters, composition data and performance data.

[0117] In summary, the method for constructing a cross-process dataset for strip steel for quality control provided by this invention establishes a unified length coordinate system to accurately align the process data collected from different processes and at different times for strip steel, thereby constructing a cross-process dataset that accurately reflects the characteristics of the entire strip steel production process.

[0118] In some examples, step 5 above may include:

[0119] Step 5.1: Connect the length domain process data of the target strip in each process.

[0120] For example, the length field process data of the target strip in each process can be concatenated by the process number of each process, where the process number is used to indicate the order of each process.

[0121] Figure 5 This is a schematic diagram of a cross-process data association structure in an embodiment of the present invention, such as... Figure 5 As shown, process parameter data for the same strip steel in different production processes are linked together based on the process number. During strip steel production, the same strip steel typically undergoes multiple processes sequentially, such as hot rolling, leveling, cold rolling, and continuous annealing or galvanizing. The production control system for each process records the corresponding process parameter data. By obtaining the process number corresponding to each process and using the strip steel identifier as an index, the process parameter data for the same strip steel in different production processes can be sequentially linked together, thus forming a cross-process parameter data chain.

[0122] The aforementioned process numbers are used to identify the processing sequence of the strip steel in each production process and can be used to track special situations that may occur during the production process, such as slitting, coiling, or bypassing of the strip steel. Through step 5.1, the evolution relationship of process parameters for the same strip steel at different production stages can be established.

[0123] It should be noted that steel coils, the subsequent product of strip steel, can also be used as the recording object, and the subsequent steps are the same, so they will not be repeated here.

[0124] Step 5.2: Obtain the smelting composition data of the molten steel used to smelt the target strip, and associate the smelting composition data with the target strip.

[0125] For example, during the production process, strip steel typically inherits the tapping mark of its corresponding molten steel. This tapping mark allows for the association of the steel coil with the smelting composition data from the smelting process, thereby establishing a mapping relationship between the strip steel and its corresponding chemical composition. This smelting composition data may include information such as carbon content, manganese content, silicon content, and the content of other alloying elements. Step 5.2 achieves a unified organization between process parameters and material composition.

[0126] Step 5.3: Obtain performance test data of the target strip at various length positions, and associate the performance test data with the length positions of the target strip.

[0127] For example, the aforementioned performance test data may include performance indicators such as tensile strength, yield strength, elongation, and hardness.

[0128] In some examples, the normalized length coordinates constructed in step 4.4 can be used to correlate the performance test data at each of the above locations with the strip and its length position.

[0129] It should be noted that in actual production, the strip steel needs to be correlated with the performance test data through the inspection and testing batch number. The aforementioned inspection and testing batch number refers to a unique number used to identify the target strip steel during the inspection, testing, or experimentation process. Alternatively, by obtaining the inspection and testing batch number and combining it with the normalized length coordinates constructed in step 4.4, the performance test data at each of the above locations can be associated with the strip steel and its length position.

[0130] It should be noted that by establishing a correspondence between process data and performance testing data, the correlation between the process state and final performance of the strip at different length positions can be realized.

[0131] Step 5.4: Using the steel coil identifier corresponding to the target strip and the coordinates of the target strip in the length direction as indexes, construct the cross-process dataset based on the length domain process data, the smelting composition data, and the performance testing data.

[0132] For example, Figure 6 This is a schematic diagram of a cross-process data set result for strip steel for quality control in an embodiment of the present invention, as shown below. Figure 6 As shown, the markings are on the steel coil. Figure 6 Using the slab number as an index, the table header is further constructed based on the aforementioned steel tapping mark and test batch number, recording material information, mechanical property information, composition information, and process parameters (process data) based on a unified length coordinate system.

[0133] When the detection position of a certain sample is obtained, the corresponding position of the sample in the strip length direction can be determined based on the normalized length coordinates. According to the mapping relationship established in step 4.5 above, the corresponding length position of the position in each preceding production process can be determined, thereby obtaining the process parameter data of the position in each production process and realizing the accurate correlation between the sample performance data and the process parameters of multiple processes.

[0134] To verify the technical effects of the present invention, specific examples are given here:

[0135] The mechanical properties of strip steel were selected as prediction targets from the aforementioned cross-process dataset, including sample yield strength, sample tensile strength, and sample elongation, totaling three prediction target dimensions. The model input features comprise 58 feature dimensions, mainly including chemical composition features, rolling process parameter features, and temperature process parameter features. Specifically, chemical composition features include the content of major alloying elements in the molten steel; rolling process parameter features include process parameters such as the reduction rate of each stand and the tension between stands; and temperature process parameter features include key production temperature parameters such as the finishing mill inlet temperature, finishing mill outlet temperature, and coiling temperature.

[0136] After obtaining the cross-process dataset, various common machine learning models were used to predict and analyze the aforementioned quality indicators, including Linear Regression (LR), Support Vector Regression (SVR), Random Forest (RF), and Extreme Gradient Boosting Regression (XGBR). A comparative experiment was conducted between the dataset constructed using the method of this invention and the original dataset without cross-process length domain alignment. To more intuitively illustrate the impact of the method of this invention on prediction performance, root mean square error (RMSE) and mean absolute error (MAE) were selected as evaluation indicators for the prediction method. The statistical analysis results of the model prediction errors are shown in Table 1.

[0137] Table 1 Comparison of Model Prediction Errors Before and After Cross-Process Length Alignment Tensile strength (MPa) 20.39 13.79 12.12 9.73 Elongation of specimen (%) 2.57 2.20 1.66 1.60 Specimen yield strength (MPa) 25.84 15.86 14.03 10.90

[0138] Experimental results show that after using the method of this invention to align the length domain and correlate cross-process data for multi-process data, the prediction errors of each model are significantly reduced. As shown in Table 1, for the tensile strength prediction task, the average mean square error of the model is reduced by approximately 32.37%, and the average absolute error is reduced by approximately 19.71%; for the elongation prediction task, the average mean square error is reduced by approximately 14.40%, and the average absolute error is reduced by approximately 3.61%; for the yield strength prediction task, the average mean square error is reduced by approximately 38.62%, and the average absolute error is reduced by approximately 22.31%. These results indicate that by aligning the length domain and correlating cross-process data for multi-process production, the consistency between data from different production processes can be effectively improved, enabling process parameters to accurately correspond at the same physical location of the strip, thereby improving data quality and significantly enhancing the prediction accuracy of the quality prediction model. The machine learning model used in this invention is merely an example; those skilled in the art can select other data-driven models as needed.

[0139] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for constructing a cross-process dataset for strip steel for quality control, characterized in that, include: Collect time-domain process data of the target strip steel in each process; Based on the instantaneous linear velocity function of the target strip in each process, the cumulative length coordinate function of the target strip in each process is determined; Based on the inverse mapping relationship of the cumulative length coordinate function and the time domain process data, the length domain process data of the target strip is determined. The length domain process data represents the correspondence between the target strip and the process data in the length direction. Construct a cross-process dataset based on the length field process data; The method for constructing a cross-process data set for strip steel for quality control also includes a normalization method for the cumulative length coordinate function, and based on this, an alignment of the length domain process data across processes is established, specifically: Obtain the target strip steel into the process The length of the cut-off head, the length of the cut-off tail, the length of the cut-off head, and the length of the cut-off tail are as follows: The target strip is determined in the process based on the uncut length, the inlet cut length, the inlet cut tail length, the outlet cut length, and the outlet cut tail length. The effective length of time is as follows: ； in, Indicates the process The effective length, Indicates the process Unresected length Indicates the process The length of the inlet cut-off head, Indicates the process The length of the exit cut-off head, Indicates the process The length of the cut-off at the entrance. Indicates the process The length of the cut-off tail at the outlet; The effective interval of the cumulative length coordinate function corresponding to the target process is determined, and the effective interval is expressed as: ； in, Indicates the process The corresponding cumulative length coordinate function; The process is constructed within the effective range. The corresponding normalized cumulative length coordinate function; ； in, This represents the normalized cumulative length coordinate function. ; According to the process The effective length, inlet cutter length, and outlet cutter length, and the process described herein. The corresponding normalized cumulative length coordinate function is used to establish the process. With the process The length mapping relationship between them, the process This refers to the process described. The preceding adjacent process, the process The corresponding effective length is equal to the process. The corresponding uncut lengths, the length mapping relationship is as follows: ； in, Indicates the process The corresponding normalized cumulative length coordinate function, Indicates the process The effective length, Indicates the process The length of the inlet cut-off head, Indicates the process The length of the exit cut-off head; When the detection position of a certain sample is obtained, the corresponding position of the sample in the strip length direction is determined based on the normalized length coordinates, and the corresponding length position of the position in each preceding production process is determined according to the established length mapping relationship, thereby obtaining the process parameter data of the position in each production process. The process is constructed within the effective range. Following the normalized cumulative length coordinate function corresponding to the cumulative length coordinate function, it also includes: Determined in the process and said process Between these points, the number of times the target strip reverses along its length; The process is based on the number of reversals. The corresponding normalized cumulative length coordinate function is corrected as follows: ； in, This refers to the revised process described above. The corresponding normalized cumulative length coordinate function.

2. The method for constructing a cross-process dataset for strip steel for quality control according to claim 1, characterized in that, The step of determining the cumulative length coordinate function of the target strip in each process based on the instantaneous linear velocity function of the target strip in each process includes: ； in, Indicates the process The corresponding cumulative length coordinate function, This indicates that the target strip steel is in the process. The instantaneous linear velocity function in Indicates the process The starting time; The determination of the length domain process data of the target strip based on the inverse mapping relationship of the cumulative length coordinate function and the time domain process data includes: ； in, Indicates the process The corresponding inverse function of the cumulative length coordinate function or the numerical equivalent mapping of the inverse function, A function representing the time-domain process data. A function representing the process data in the length field.

3. The method for constructing a cross-process dataset for strip steel for quality control according to claim 1, characterized in that, The step of constructing a cross-process dataset based on the length field process data includes: The length domain process data of the target strip in each process are concatenated; Obtain the smelting composition data of the molten steel used to smelt the target strip, and associate the smelting composition data with the target strip; Acquire performance test data of the target strip at various length positions, and correlate the performance test data with the length positions of the target strip; Using the steel coil identifier corresponding to the target strip and the coordinates of the target strip in the length direction as indexes, the cross-process dataset is constructed based on the length domain process data, the smelting composition data, and the performance testing data.

Images