Hot rolling product thickness control process analysis method

Abstract

The invention relates to an analysis method for a thickness control process of a hot-rolled product, which comprises the following steps of: classifying thickness abnormity into head, tail and belt bodies through a standardized defect classification mechanism, respectively matching a targeted analysis method, and establishing a multivariable correlation analysis and dynamic coupling analysis method by combining key parameters such as rolling force, temperature, roll gap compensation and loop control; and the self-adaptive threshold adjustment is realized, and the judgment threshold is dynamically adjusted according to the steel type and specification, so that the generalization ability is improved. The method solves the problems of incomplete analysis, inaccurate classification and poor dynamic adaptability in the prior art, can remarkably improve the thickness control precision of the hot-rolled product and reduce the rejection rate, is suitable for various hot-rolling production lines, and has wide industrial application prospects.

Description

TECHNICAL FIELD

[0001] The patent application belongs to the technical field of hot rolling mill automation in the metallurgical industry, and more particularly relates to a hot rolling product thickness control process analysis method. BACKGROUND

[0002] In the hot rolling production process, thickness control is one of the core links that determines product quality and performance. The thickness precision of hot rolling products directly affects their mechanical properties, processing forming ability, and the use reliability of the final products. If the thickness control deviation is large, it may lead to uneven product strength, surface defects (such as warping or wrinkling), and even cause fracture problems in subsequent processing procedures, seriously affecting production efficiency and economic benefits. Therefore, how to achieve accurate thickness control under complex rolling conditions and quickly locate the cause and make corrections when there is a deviation is a key technical challenge in hot rolling production.

[0003] Currently, hot rolling thickness control mainly relies on the accuracy of rolling model predictions of key parameters such as roll gap, temperature, rolling force, and the automation system of the rolling mill equipment (such as the AGC automatic thickness control system), which adjusts the roll gap, rolling force model parameters, and real-time rolling force and roll gap to stabilize the thickness output. However, in actual production, thickness deviation still occurs frequently, and the cause cannot be quickly locked, resulting in poor thickness control effect, which has a great impact on product quality and production cost.

[0004] Patent CN201910001234.5 proposes a hot rolling thickness deviation analysis method based on multi-source data fusion, which integrates multi-dimensional process parameters such as rolling force, temperature distribution, and cooling water state to establish a dynamic correlation model between thickness deviation and process variables. However, its limitation is that it does not establish classification judgment rules for different types of thickness abnormalities (such as head and tail deviation, local fluctuation), resulting in insufficient pertinence of the analysis results. In addition, this method has high requirements for data quality and computing resources, which may affect its real-time performance.

[0005] Patent CN202010567890.1 develops a thickness deviation root cause analysis method based on machine learning, which automatically identifies the main causes of thickness deviation such as rolling temperature fluctuation, roll wear, and uneven cooling by training a neural network model. Its innovation lies in using historical data to construct a nonlinear mapping relationship between thickness deviation and process parameters, avoiding the dependence on physical assumptions of traditional methods. However, the practical application effect of this method is limited by the coverage and quality of the training data, especially in extreme working conditions or new steel production, the model's generalization ability is insufficient. In addition, this method does not provide a classification and tracing mechanism for thickness abnormalities, making it difficult to achieve fast and accurate correction.

[0006] Patent CN202110654321.X designs a hot rolling thickness deviation analysis method based on abnormal classification, which divides the thickness deviation into three categories of head and tail deviation, local fluctuation and overall deviation, respectively establishes the judgment rules and root cause analysis model. The core is to combine the process parameters (such as rolling speed, temperature gradient, cooling efficiency) with the characteristics of the thickness curve (such as fluctuation amplitude, frequency, spatial distribution), to realize the rapid classification and tracing of thickness deviation. The advantage of this method is that it can provide differentiated analysis strategies for different types of thickness abnormalities, but its limitation is that it does not fully consider the coupling effect between multiple variables, which limits the accuracy of the analysis results. In addition, this method has high requirements for the real-time and accuracy of data collection, which may increase the complexity of the system.

[0007] The above technologies have different characteristics in the field of hot rolling thickness deviation cause analysis, but they all have the problem of insufficient multivariate coupling analysis. In order to fully tap the dynamic relationship between thickness deviation and process parameters, the abnormal classification and tracing is not perfect, and it is difficult to dynamically adjust the analysis strategy according to the change of steel grade and specification.

[0008] Therefore, in order to solve the above problems, the present application provides a hot rolling product thickness control process analysis method, which identifies and classifies the thickness deviation curve, combines the key parameters in the rolling process, and systematically analyzes the causes of thickness control deviation, so as to realize rapid correction and improve the thickness control precision. SUMMARY

[0009] The technical problem to be solved by the present application is to provide a hot rolling product thickness control process analysis method, which solves the problem of lack of standardized method in thickness control analysis in the prior art, and analyzes comprehensively and thoroughly, so as to effectively identify the problem points in the control process, and finally improve the thickness control precision of hot rolling products.

[0010] In order to solve the above problems, the technical scheme adopted by the present application is:

[0011] A hot rolling product thickness control process analysis method, comprising the following steps:

[0012] S101, obtaining the product thickness deviation curve, processing the curve data and judging the validity of the curve;

[0013] S102, obtaining the thickness upper and lower tolerance, judging the product thickness defect type, the product thickness defect type including head thickness abnormality, tail thickness abnormality and strip thickness abnormality;

[0014] S103, obtaining the rolling piece related control information, and executing the corresponding analysis steps according to the defect type;

[0015] S104-1. Calculate the rolling force setting deviation rate, and analyze the head thickness anomaly by combining the head temperature prediction deviation, alloy composition deviation rate and roll gap self-learning change amount.

[0016] S104-2. Analyze the tail thickness anomaly by combining the tail temperature deviation and the tail roll gap compensation value.

[0017] S104-3. Analyze the anomalies in belt thickness by combining the furnace dwell time and the standard deviation of the looper angle.

[0018] Further, in S101, the set of product thickness deviation curve data is denoted as T, T = {T1, T2, ..., T}. i ,…,T N}, where T i The thickness deviation at the i-th meter position is expressed in micrometers, and N is the number of product thickness deviation curve data points. The product thickness deviation curve is sampled once per meter.

[0019] Furthermore, to determine whether the curve is valid, calculate the average value T of the data set T. ave Maximum value T max and minimum value T min If all conditions are met, the curve is considered valid; otherwise, the curve is considered invalid, and the analysis process ends.

[0020] Specifically, the conditions include:

[0021] 1. T max -T min >3μm;

[0022] 2. T max -T ave >3μm;

[0023] 3. T ave -T min >3μm;

[0024] 4. T max <200μm;

[0025] 5. T min >-200μm;

[0026] 6. T max -T min <300μm;

[0027] 7. N>100;

[0028] Furthermore, in S102, the thickness tolerance includes an upper thickness tolerance and a lower thickness tolerance, wherein the upper thickness tolerance is denoted as Tol. up The thickness tolerance is denoted as Tol.low , unit: microns;

[0029] Further, in S102, the product thickness defect types include head thickness abnormality, tail thickness abnormality and strip thickness abnormality, and the judgment method is performed according to the following steps.

[0030] The thickness deviation data T is divided into rejection data, head data, tail data and strip data;

[0031] The rejection data is the front and rear S data of the thickness deviation data set T, S is determined according to the target thickness H of the strip steel targ Set, when H targ <2.5mm, S=5; when 2.5mm≤H targ <6.0mm, S=4; when 6.0mm≤H targ , S=3;

[0032] The head data is the first 3 meters of data after removing the head S data from the thickness deviation data set T, that is, head data={T S+1 ,T S+2 ,T S+3};

[0033] The tail data is the last 3 meters of data after removing the tail S data from the thickness deviation data set T, that is, tail data={T N-S-2 ,T N-S-1 ,T N-S};

[0034] The strip data is the data obtained by removing the rejection data, head data and tail data from the thickness deviation data set T, that is, strip data={T S+4 ,T S+5 ,…,T N-S-3};

[0035] Further, the head thickness deviation T h , If T h >Tol up or T h <Tol low , it is judged as head thickness abnormality, otherwise it is judged as head thickness normality;

[0036] Further, the tail thickness deviation T t , If T t >Tol up or T t <Tol low , it is judged as tail thickness abnormality, otherwise it is judged as tail thickness normality;

[0037] Further, a qualified rate P of the strip data is calculated,

[0038] wherein, is an indicator function, when T i is 1, otherwise 0; low up

[0039] Further, if P < P lim , it is determined that the strip thickness is abnormal, otherwise it is determined that the strip thickness is normal, wherein P lim is a strip thickness qualified rate threshold, and the value range is [95%, 99%], preferably, P lim = 98%;

[0040] Further, in S103, rolling piece related control information is obtained, and subsequent steps are executed according to the defect type.

[0041] The rolling piece related control information includes a steel grade SGF, a thickness layer Grt_index, and a roll period sequence number Num.

[0042] The steel grade SGF is determined according to the steel grade mark of the rolling piece;

[0043] The thickness layer Grt_index is determined according to the target thickness H, and is defined as 23, including:

[0044] When H < 1.20mm, the thickness layer Grt_index is 0;

[0045] When 1.20mm≤H < 1.40mm, the thickness layer Grt_index is 1;

[0046] When 1.40mm≤H < 1.60mm, the thickness layer Grt_index is 2;

[0047] When 1.60mm≤H < 1.80mm, the thickness layer Grt_index is 3;

[0048] When 1.80mm≤H < 2.00mm, the thickness layer Grt_index is 4;

[0049] When 2.00mm≤H < 2.25mm, the thickness layer Grt_index is 5;

[0050] When 2.25mm≤H < 2.50mm, the thickness layer Grt_index is 6;

[0051] When 2.50mm≤H < 3.00mm, the thickness layer Grt_index is 7; ​​

[0052] 3.00mm≤H<3.50mm, thickness layer Grt_index is 8;

[0053] 3.50mm≤H<4.00mm, thickness layer Grt_index is 9;

[0054] 4.00mm≤H<5.00mm, thickness layer Grt_index is 10;

[0055] 5.00mm≤H<6.00mm, thickness layer Grt_index is 11;

[0056] 6.00mm≤H<7.50mm, thickness layer Grt_index is 12;

[0057] 7.50mm≤H<9.00mm, thickness layer Grt_index is 13;

[0058] 9.00mm≤H<10.50mm, thickness layer Grt_index is 14;

[0059] 10.50mm≤H<12.00mm, thickness layer Grt_index is 15;

[0060] 12.00mm≤H<14.00mm, thickness layer Grt_index is 16;

[0061] 14.00mm≤H<16.00mm, thickness layer Grt_index is 17;

[0062] 16.00mm≤H<18.00mm, thickness layer Grt_index is 18;

[0063] 18.00mm≤H<20.00mm, thickness layer Grt_index is 19;

[0064] 20.00mm≤H<22.00mm, thickness layer Grt_index is 20;

[0065] 22.00mm≤H<24.00mm, thickness layer Grt_index is 21;

[0066] 24.0mm≤H, thickness layer Grt_index is 22;

[0067] The roll period sequence number Num is the sequence of the rolled piece after the finishing rolling roll change, the roll period sequence number is cleared after the finishing rolling roll change, the roll period sequence number of the first rolled piece produced after the roll change is set to 1, and is sequentially accumulated;

[0068] Further, if the defect type is head thickness abnormality, step 104-1 is performed; if the defect type is tail thickness abnormality, step 104-2 is performed; if the defect type is body thickness abnormality, step 104-3 is performed.

[0069] Further, the head thickness abnormality is analyzed in 104-1, and the following steps are performed.

[0070] The rolling force setting deviation rate of the finishing mill stand is denoted as δ i ,

[0071] Wherein, FS i is the rolling force setting value of the i-th stand, unit: KN;

[0072] FA i is the actual rolling force value of the i-th stand, unit: KN;

[0073] If δ i > 15%, it is further judged whether the steel grade and thickness layer are first produced, and the rolling coil number is obtained from the database according to the steel grade and thickness layer. If the rolling coil number is 0, it is judged that the steel grade is first produced, and the rolling force setting deviation is too large, otherwise the temperature prediction deviation is continuously judged.

[0074] Further, the average value of the head temperature at the finishing mill outlet is obtained, and the average value of the head temperature at the finishing mill outlet is the average value of the temperature of the head 3-5 meters at the finishing mill outlet. If the deviation of the average value of the head temperature at the finishing mill outlet from the target value of the finishing mill outlet is > 20℃, it is judged that the large rolling force setting deviation is caused by the large finishing mill outlet temperature prediction deviation, otherwise the alloy composition of the rolled piece is continuously judged.

[0075] Further, the standard value of the alloy composition is obtained according to the steel grade of the rolled piece, and the actual value of the alloy composition is obtained according to the rolled piece ID. The alloy composition includes carbon (C), silicon (Si), and manganese (Mn).

[0076]

[0077] If the deviation rate of any alloy composition is > 30%, it is judged that the alloy composition is abnormal, causing the rolling force setting deviation to be too large.

[0078] If δ i ≤ 15%, if the roll period sequence number Num of the rolled piece is ≤ 3, the roll gap self-learning change amount of the finishing mill stand is judged,

[0079] ZBS_DIF i = |ZBS_NEW i -ZBS_OLD i |

[0080] wherein, i is the finishing stand number

[0081] ZBS DIF i is the roll gap self-learning change amount of the i-th finishing stand;

[0082] ZBS NEW i is the roll gap self-learning update value of the i-th finishing stand;

[0083] ZBS OLD i is the roll gap self-learning initial value of the i-th finishing stand;

[0084] if ZBS DIF i > 0.1, it is judged that the roll gap self-learning of the starting-up material is inconsistent;

[0085] if the roll period sequence number Num > 3, it is judged whether the thickness specification transition is excessively large, the deviation of the target thickness of the current rolled piece and the previous rolled piece is compared, and if the absolute value of the deviation > 2.5 mm, the roll gap self-learning change amount of the finishing stand of the current rolled piece is obtained,

[0086] ZBS DIF i = |ZBS NEW i - ZBS OLD i |

[0087] wherein, i is the finishing stand number

[0088] ZBS DIF i is the roll gap self-learning change amount of the i-th finishing stand;

[0089] ZBS NEW i is the roll gap self-learning update value of the i-th finishing stand;

[0090] ZBS OLD i is the roll gap self-learning initial value of the i-th finishing stand;

[0091] if ZBS DIF i > 0.1, it is judged that the thickness specification transition is excessively large;

[0092] Further, in 104-2, the tail thickness abnormality is analyzed according to the following steps.

[0093] Further, the average value of the tail temperature at the finishing stand exit of the rolled piece is obtained, the average value of the tail temperature at the finishing stand exit of the rolled piece is the average value of the temperature of the tail 3-5 meters at the finishing stand exit, if the deviation of the average value of the tail temperature at the finishing stand exit and the target value at the finishing stand exit > 20℃, it is judged that the tail thickness abnormality is caused by the large deviation of the tail temperature prediction at the finishing stand exit, otherwise, it is continuously judged whether the tail roll gap compensation control is abnormal.

[0094] Further, the roll gap compensation values of the three stands after finishing rolling are obtained, denoted as gc1, gc2 and gc3, and the average value gc of gc1, gc2 and gc3 is calculated a ,

[0095] Further, if the thickness deviation T t >Tol up and gc a >0, or T t <Tol low and gc a <0, it is judged that the tail thickness is abnormal due to the tail roll gap compensation control.

[0096] Further, the strip thickness abnormality is analyzed in 104-3, and the following steps are performed.

[0097] The residence time is the heating time of the slab in the heating furnace, denoted as t h , in hours, and if t h >5, it is determined that the strip thickness is abnormal due to long residence time of the rolling piece, otherwise the loop control condition is continued to be judged.

[0098] Further, the loop angle curve data set of the finishing stand is obtained, denoted as L_a j ,

[0099] L_a j ={L_a j1 ,L_a j2 ,…,L_a jk ,…,L_a jm}

[0100] Wherein, j is the loop number, and L_a j is the angle curve data set of the loop j;

[0101] L_a jk is the kth data sampling value of the loop j;

[0102] m is the total number of data points, and one data point is collected per second for the loop angle curve data;

[0103] Further, the standard deviation of the loop angle curve data set is calculated, and to prevent noise data from affecting the calculation results when the head and tail loops are set and released, the first and last 0.1m data points are removed when calculating the standard deviation and mean value.

[0104]

[0105] Wherein, σ j is the standard deviation of the angle curve data of the loop j;

[0106] is the mean value of the data set;

[0107] Further, it is judged whether the standard deviation of each loop angle curve data is out of limit, if there is σ j U, it is judged that the strip thickness abnormality is caused by loop control problem, U is the standard deviation threshold of loop angle, the value range is [0.3, 0.8], preferably, U=0.5.

[0108] Due to the adoption of the above technical solutions, the present application has the following beneficial effects:

[0109] The present application classifies the thickness abnormality into three categories of head, tail and strip body through the standardized defect classification mechanism, respectively matches the targeted analysis method, combines the key parameters of rolling force, temperature, roll gap compensation and loop control, establishes the multivariate correlation analysis and dynamic coupling analysis method, and adjusts the threshold value adaptively, so that the threshold value is dynamically adjusted according to the steel grade and specification, and the generalization ability is improved.

[0110] The method of the present application solves the problems of incomplete analysis, inaccurate classification and poor dynamic adaptability in the prior art, can significantly improve the thickness control precision of hot-rolled products, reduce the scrap rate, is suitable for various hot-rolling production lines, and has a wide industrial application prospect. BRIEF DESCRIPTION OF DRAWINGS

[0111] Figure 1 A flow chart of a hot-rolled product thickness control process analysis method provided by the embodiment of the present application.

[0112] Figure 2 A thickness deviation curve of a rolled piece 24C3363E52 provided by the embodiment of the present application.

[0113] Figure 3 A thickness deviation curve of a rolled piece 38D3471F09 provided by the embodiment of the present application.

[0114] Figure 4 A 6# loop angle curve of a rolled piece 38D3471F09 provided by the embodiment of the present application. DETAILED DESCRIPTION

[0115] The present application will be further described in detail below in combination with the embodiments.

[0116] Referring to Figure 1 The embodiment of the present application provides a hot-rolled product thickness control process analysis method, which comprises the following steps:

[0117] S101, obtaining a product thickness deviation curve, processing the curve data and judging the validity of the curve;

[0118] Further, in S101, the product thickness deviation curve data set is denoted as T, T={T1, T2, …, TN}, where T is the thickness deviation of the i-th meter position, unit: microns, N is the number of product thickness deviation curve data, and the product thickness deviation curve is sampled at a frequency of once per meter. i ,…,T N} where T i is the thickness deviation of the i-th meter position, unit: microns, N is the number of product thickness deviation curve data, and the product thickness deviation curve is sampled at a frequency of once per meter.

[0119] Further, it is judged whether the curve is valid, the average value T of the data set T, the maximum value T and the minimum value T are calculated. ave max min When all conditions are met, the curve is judged to be valid, otherwise the curve is judged to be invalid, and the analysis process is ended.

[0120] Specifically, the conditions include:

[0121] 1, T max -T min > 3 μm;

[0122] 2, T max -T ave > 3 μm;

[0123] 3, T ave -T min > 3 μm;

[0124] 4, T max <200 μm;

[0125] 5, T min >-200 μm;

[0126] 6, T max -T min <300 μm;

[0127] 7, N > 100;

[0128] S102, get the thickness upper and lower tolerances, judge the product thickness defect type, the product thickness defect type includes head thickness abnormality, tail thickness abnormality and belt body thickness abnormality;

[0129] Further, in S102, the thickness upper and lower tolerances include thickness upper tolerance and thickness lower tolerance, where the thickness upper tolerance is denoted as Tol up , and the thickness lower tolerance is denoted as Tol low , unit: microns;

[0130] Further, in S102, the product thickness defect type includes head thickness abnormality, tail thickness abnormality and belt body thickness abnormality, and the judgment method is performed according to the following steps. ​​

[0131] The thickness deviation data T is divided into rejection data, head data, tail data and body data;

[0132] The rejection data is the first and last S data of the thickness deviation data set T, S is determined according to the target thickness H of the strip steel targ When H targ When 2.5mm≤H targ When 6.0mm≤H targ When 6.0mm≤H

[0133] The head data is the first 3 meters of data after removing the head S data of the thickness deviation data set T, that is, head data={T S+1 ,T S+2 ,T S+3};

[0134] The tail data is the last 3 meters of data after removing the tail S data of the thickness deviation data set T, that is, tail data={T N-S-2 ,T N-S-1 ,T N-S};

[0135] The body data is the data obtained by removing the rejection data, head data and tail data from the thickness deviation data set T, that is, body data={T S+4 ,T S+5 ,…,T N-S-3};

[0136] Further, the head thickness deviation T h is calculated If T h >Tol up or T h <Tol low , it is judged that the head thickness is abnormal, otherwise it is judged that the head thickness is normal;

[0137] Further, the tail thickness deviation T t is calculated If T t >Tol up or T t <Tol low , it is judged that the tail thickness is abnormal, otherwise it is judged that the tail thickness is normal;

[0138] Further, the qualified rate P of the body data is calculated

[0139] Wherein, is an indicator function, when T i is in [Tol low, Tolup ] is 1, otherwise 0;

[0140] Further, if P < P lim , it is determined that the strip thickness is abnormal, otherwise it is determined that the strip thickness is normal, wherein P lim is a strip thickness qualification rate threshold, and the value range is [95%, 99%], preferably, P lim = 98%.

[0141] S103, obtaining rolling piece related control information, and performing corresponding analysis steps according to the defect type;

[0142] The rolling piece related control information includes a steel grade SGF, a thickness layer Grt_index, and a roll period sequence number Num.

[0143] The steel grade is determined according to the steel grade mark of the rolling piece.

[0144] The thickness layer Grt_index is determined according to the target thickness H, and is defined as 23, including:

[0145] When H < 1.20 mm, the thickness layer Grt_index is 0.

[0146] When 1.20 mm≤H < 1.40 mm, the thickness layer Grt_index is 1.

[0147] When 1.40 mm≤H < 1.60 mm, the thickness layer Grt_index is 2.

[0148] When 1.60 mm≤H < 1.80 mm, the thickness layer Grt_index is 3.

[0149] When 1.80 mm≤H < 2.00 mm, the thickness layer Grt_index is 4.

[0150] When 2.00 mm≤H < 2.25 mm, the thickness layer Grt_index is 5.

[0151] When 2.25 mm≤H < 2.50 mm, the thickness layer Grt_index is 6.

[0152] When 2.50 mm≤H < 3.00 mm, the thickness layer Grt_index is 7.

[0153] When 3.00 mm≤H < 3.50 mm, the thickness layer Grt_index is 8.

[0154] When 3.50 mm≤H < 4.00 mm, the thickness layer Grt_index is 9.

[0155] 4.00mm≤H<5.00mm, the thickness layer Grt_index is 10;

[0156] 5.00mm≤H<6.00mm, the thickness layer Grt_index is 11;

[0157] 6.00mm≤H<7.50mm, the thickness layer Grt_index is 12;

[0158] 7.50mm≤H<9.00mm, the thickness layer Grt_index is 13;

[0159] 9.00mm≤H<10.50mm, the thickness layer Grt_index is 14;

[0160] 10.50mm≤H<12.00mm, the thickness layer Grt_index is 15;

[0161] 12.00mm≤H<14.00mm, the thickness layer Grt_index is 16;

[0162] 14.00mm≤H<16.00mm, the thickness layer Grt_index is 17;

[0163] 16.00mm≤H<18.00mm, the thickness layer Grt_index is 18;

[0164] 18.00mm≤H<20.00mm, the thickness layer Grt_index is 19;

[0165] 20.00mm≤H<22.00mm, the thickness layer Grt_index is 20;

[0166] 22.00mm≤H<24.00mm, the thickness layer Grt_index is 21;

[0167] 24.0mm≤H, the thickness layer Grt_index is 22;

[0168] The roll period sequence number Num is the sequence of the rolled piece after the finishing rolling roll change, the roll period sequence number is cleared after the finishing rolling roll change, the roll period sequence number of the first rolled piece after the roll change is set to 1, and is sequentially accumulated;

[0169] Further, if the defect type is head thickness abnormality, step 104-1 is performed; if the defect type is tail thickness abnormality, step 104-2 is performed; and if the defect type is belt thickness abnormality, step 104-3 is performed.

[0170] S104-1, calculate the rolling force setting deviation rate, combine the head temperature prediction deviation, alloy composition deviation rate and roll gap self-learning change to analyze the head thickness abnormality;

[0171] Further, the head thickness abnormality is analyzed in 104-1, and the following steps are performed.

[0172] Calculate the rolling force setting deviation rate δ of the finishing mill stand i ,

[0173] Wherein, FS i is the rolling force setting value of the i-th stand, unit KN;

[0174] FA i is the actual value of the rolling force of the i-th stand, unit KN;

[0175] If δ i > 15%, further judge whether the steel grade and thickness layer are first produced, obtain the rolling coil number from the database according to the steel grade and thickness layer, if the rolling coil number is 0, it is judged that the steel grade is first produced, the rolling force setting deviation is too large, otherwise continue to judge the temperature prediction deviation;

[0176] Further, the average value of the head temperature at the finishing mill outlet is obtained, and the average value of the head temperature at the finishing mill outlet is the average value of the temperature of the head 3-5 meters at the finishing mill outlet, if the deviation of the average value of the head temperature at the finishing mill outlet and the target value of the finishing mill outlet is > 20℃, it is judged that the large rolling force setting deviation is caused by the large deviation of the finishing mill outlet temperature prediction, otherwise continue to judge whether the alloy composition of the rolled piece is abnormal;

[0177] Further, the standard value of the alloy composition is obtained according to the steel grade of the rolled piece, and the actual value of the alloy composition is obtained according to the rolled piece ID, the alloy composition includes carbon (C), silicon (Si), manganese (Mn);

[0178]

[0179] If the deviation rate of any alloy composition is > 30%, it is judged that the alloy composition is abnormal, which causes the rolling force setting deviation to be too large.

[0180] If δ i ≤ 15%, if the roll period sequence number Num of the rolled piece is ≤ 3, it is judged that the roll gap self-learning change of the finishing mill stand,

[0181] ZBS_DIF i = |ZBS_NEW i -ZBS_OLD i |

[0182] Wherein, i is the finishing mill stand number

[0183] ZBS_DIF i is the roll gap self-learning change amount of the i-th finishing stand;

[0184] ZBS_NEW i is the roll gap self-learning update value of the i-th finishing stand;

[0185] ZBS_OLD i is the roll gap self-learning initial value of the i-th finishing stand;

[0186] If ZBS_DIF i > 0.1, it is judged that the roll gap self-learning of the starting-up material is inconsistent.

[0187] If the roll period sequence number Num > 3, it is judged whether the thickness specification transition is too large, and the deviation of the target thickness of the current rolled piece and the previous rolled piece is compared. If the absolute value of the deviation is > 2.5 mm, the roll gap self-learning change amount of the i-th finishing stand of the current rolled piece is obtained,

[0188] ZBS_DIF i = |ZBS_NEW i - ZBS_OLD i |

[0189] wherein i is the finishing stand number

[0190] ZBS_DIF i is the roll gap self-learning change amount of the i-th finishing stand;

[0191] ZBS_NEW i is the roll gap self-learning update value of the i-th finishing stand;

[0192] ZBS_OLD i is the roll gap self-learning initial value of the i-th finishing stand;

[0193] If ZBS_DIF i > 0.1, it is judged that the thickness specification transition is too large.

[0194] S104-2, the tail thickness abnormality is analyzed in combination with the tail temperature deviation and the tail roll gap compensation value;

[0195] Further, in 104-2, the tail thickness abnormality is analyzed according to the following steps.

[0196] Further, the average value of the tail temperature at the finishing rolling outlet of the rolled piece is obtained, and the average value of the tail temperature at the finishing rolling outlet of the rolled piece is the average value of the temperature of the tail 3-5 meters at the finishing rolling outlet of the rolled piece. If the deviation of the average value of the tail temperature at the finishing rolling outlet from the target value of the tail temperature at the finishing rolling outlet is greater than 20 DEG C, it is judged that the tail thickness is abnormal due to the tail temperature prediction deviation. Otherwise, it is continued to judge whether the tail roll gap compensation control is abnormal.

[0197] Further, the roll gap compensation values of the rolled piece after three stands of finishing rolling are obtained, denoted as gc1, gc2 and gc3, and the average value gc of gc1, gc2 and gc3 is calculated. a ,

[0198] Further, if the thickness deviation T t >Tol up and gc a >0, or T t <Tol low and gc a <0, it is judged that the tail thickness is abnormal due to the tail roll gap compensation control.

[0199] S104-3, the strip thickness abnormality is analyzed in combination with the residence time in the furnace and the standard deviation of the loop angle.

[0200] Further, the strip thickness abnormality is analyzed in 104-3, and the following steps are performed.

[0201] Specifically, the residence time of the rolled piece in the heating furnace is denoted as t h , and the unit is hour. If t h >5, it is judged that the residence time of the rolled piece is long, which leads to the strip thickness abnormality. Otherwise, the loop control condition is continued to be judged.

[0202] Further, the loop angle curve data set of the finishing rolling stand is obtained, denoted as L_a j ,

[0203] L_a j ={L_a j1 ,L_a j2 ,…,L_a jk ,…,L_a jm}

[0204] Wherein, j is the loop serial number, and L_a j is the angle curve data set of the loop j;

[0205] L_a jk is the kth data sampling value of the loop j;

[0206] m is the total number of data points, and one data point is collected per meter of the loop angle curve data.

[0207] Further, the standard deviation of the loop angle curve data set is calculated, and when calculating the standard deviation and mean value, to prevent the noise data from affecting the calculation results during the head and tail loop, the first and last 0.1*m data points are removed.

[0208]

[0209] Wherein, σ j is the standard deviation of the loop j angle curve data;

[0210] is the mean value of the data set;

[0211] Further, it is judged whether the standard deviation of each loop angle curve data is out of limit, if there is σ j >U, it is judged that the strip thickness is abnormal due to the loop control problem, U is the standard deviation threshold of the loop angle, the value range is [0.2, 0.8], preferably, U=0.5.

[0212] The present application relates to a kind of hot rolling product thickness control process analysis method, by standardization defect classification mechanism, it is divided into head, tail and strip three categories to thickness anomaly, respectively match targeted analysis method, combine rolling force, temperature, roll gap compensation, loop control etc. Key parameters, establish multivariate correlation analysis and dynamic coupling analysis method, self-adaptive threshold adjustment, according to steel grade, specification dynamic adjustment determination threshold, improve generalization ability.

[0213] The method of the present application solves the problems of incomplete analysis, inaccurate classification and poor dynamic adaptability in the prior art, significantly improves the thickness control precision of hot rolling products, reduces the scrap rate, and is suitable for various hot rolling production lines, and has wide industrial application prospect.

[0214] The technical scheme of the present embodiment will be further described below through two typical application examples:

[0215] Example one:

[0216] Slab number: 24C3363E52

[0217] Steel grade: S420X (SGF=5)

[0218] Target thickness: H=3.2mm (corresponding Grt_index=8)

[0219] Rolling sequence number: Num=12 (the 12th rolled piece after roll change in finishing rolling)

[0220] Analysis process:

[0221] S101 curve validity verification:

[0222] Figure 2 For the thickness deviation curve of the rolled piece 24C3363E52, the length N=590 of the thickness deviation data set T is obtained, which satisfies N>100.

[0223] T max = 142 μm, T min = -23 μm, T ave = 5.15 μm

[0224] Verification condition:

[0225] T max -T min = 142-(-23)=165 μm>3 μm;

[0226] T max -T ave = 142-5.15=136.85 μm>3 μm;

[0227] T ave -T min = 5.15-(-23)=28.15 μm>3 μm;

[0228] T max = 142 μm<200 μm;

[0229] T min = -23 μm>-200 μm;

[0230] T max -T min = 142-(-23)=165 μm<300 μm;

[0231] Conclusion: the curve is effective

[0232] S102 defect classification:

[0233] Target thickness H=3.2 mm (S=4)

[0234] Head thickness Head thickness anomaly

[0235] Head thickness Tail thickness is normal;

[0236] Strip qualified rate P=99.14%>P lim =98%, the strip thickness is normal;

[0237] S104-1 head anomaly analysis:

[0238] 1. Rolling force analysis:

[0239] Rolling force setting deviation rate of finishing mill F1 rack

[0240] Rolling force setting is normal;

[0241] Rolling force setting deviation rate of F2 stand in finishing mill

[0242] Rolling force setting is normal;

[0243] Rolling force setting deviation rate of F3 stand in finishing mill

[0244] Rolling force setting is normal;

[0245] Rolling force setting deviation rate of F4 stand in finishing mill

[0246] Rolling force setting is normal;

[0247] Rolling force setting deviation rate of F5 stand in finishing mill

[0248] Rolling force setting is normal;

[0249] Rolling force setting deviation rate of F6 stand in finishing mill

[0250] Rolling force setting is normal;

[0251] Rolling force setting deviation rate of F7 stand in finishing mill

[0252] Rolling force setting is normal;

[0253] 2, rolling sequence number Num = 12 > 3, the target thickness of the previous rolled piece is 6.0 mm, and the absolute value of the deviation from the target thickness of the current rolled piece is = |6.0-3.2| = 2.7 mm > 2.5 mm, check the roll gap self-learning:

[0254] F4 stand ZBS_DIF4 = |ZBS_NEW4-ZBS_OLD4| = |0.263-0.132| = 0.131 > 0.1 mm (the thickness specification transition is too large);

[0255] Conclusion: the thickness specification transition is too large, resulting in the head thickness being out of tolerance.

[0256] Example two:

[0257] Slab number: 38D3471F09

[0258] Steel grade: TGW550 (SGF = 12)

[0259] Target thickness: H = 2.75 mm (Grt_index = 7)

[0260] Analysis process:

[0261] S101 curve validity verification:

[0262] Figure 3 For the thickness deviation curve of the rolled piece 38D3471F09, the length N=895 of the thickness deviation data set T is obtained, which satisfies N>100.

[0263] T max =143μm, T min =-29μm, T ave =9.72μm

[0264] Verification condition:

[0265] T max -T min =143-(-29)=172μm>3μm;

[0266] T max -T ave =143-9.72=133.28μm>3μm;

[0267] T ave -T min =9.72-(-29)=38.72μm>3μm;

[0268] T max =143μm<200μm;

[0269] T min =-29μm>-200μm;

[0270] T max -T min =143-(-29)=172μm<300μm;

[0271] Conclusion: Curve is valid

[0272] S102 defect classification:

[0273] Target thickness H=2.75mm (S=4)

[0274] Head thickness and T h >Tol low =-50μm, head thickness is normal

[0275] Head thickness and T t >Tol low =-50μm, tail thickness is normal

[0276] Belt body qualification rate P=96.73%<Plim = 98%, strip thickness abnormality;

[0277] S104-3 strip thickness abnormality analysis:

[0278] Residence time (hours) t h = 3.35 < 5, residence time normal;

[0279] Loop angle analysis (No. 1 loop-No. 6 loop):

[0280] 1#Loop standard deviation σ1 = 0.231 < U = 0.5, 1#Loop control is normal;

[0281] 2#Loop standard deviation σ2 = 0.366 < U = 0.5, 2#Loop control is normal;

[0282] 3#Loop standard deviation σ3 = 0.351 < U = 0.5, 3#Loop control is normal;

[0283] 4#Loop standard deviation σ4 = 0.383 < U = 0.5, 4#Loop control is normal;

[0284] 5#Loop standard deviation σ5 = 0.431 < U = 0.5, 5#Loop control is normal;

[0285] Figure 4 For the 6#Loop angle curve of the rolled piece 38D3471F09, the 6#Loop standard deviation σ6 = 0.538 > U = 0.5, the 6#Loop control fluctuation is large;

[0286] Conclusion: The 6#Loop control abnormality leads to strip thickness fluctuation, and the loop tension control parameters need to be adjusted.

[0287] The present application relates to a kind of hot-rolled product thickness control process analysis method, by standardization defect classification mechanism, thickness abnormality is divided into head, tail and strip three categories, respectively match targeted analysis method, combine rolling force, temperature, roll gap compensation, loop control etc. Key parameters, establish multivariate correlation analysis and dynamic coupling analysis method, adaptive threshold adjustment, according to steel grade, specification dynamic adjustment determination threshold, improve generalization ability.

[0288] The method of the present application solves the problems of incomplete analysis, inaccurate classification and poor dynamic adaptability in the prior art, significantly improves the thickness control precision of hot-rolled products, reduces the scrap rate, and is suitable for various hot-rolling production lines, and has wide industrial application prospect.

[0289] Finally, it should be noted that the above detailed description of the preferred embodiments is merely intended to illustrate the technical solutions of the present application, rather than limit the same. Even though the present application has been described in detail with reference to the examples, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalent replaced, without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.

[0290] The above describes the preferred embodiments of the present application. It should be noted that the present application is not limited to the specific embodiments described above, and the devices and structures not described in detail should be understood as implemented in the ordinary way in the art. Any simple modification, equivalent change and modification of the above embodiments according to the technical essence of the present application, without departing from the scope of the technical solutions of the present application, still belongs to the scope of protection of the technical solutions of the present application.

Claims

1. A method for analyzing the thickness control process of hot-rolled products, characterized in that, Includes the following steps: S101. Obtain the product thickness deviation curve, process the curve data of the product thickness deviation curve, and determine the validity of the curve. S102. Obtain the upper and lower tolerances of thickness and determine the type of product thickness defect. The types of product thickness defects include abnormal thickness at the head, abnormal thickness at the tail, and abnormal thickness of the belt body. S103. Obtain relevant control information of the rolled piece, and perform the corresponding analysis steps according to the product thickness defect type. If the defect type is abnormal thickness at the head, perform step 104-1; if the defect type is abnormal thickness at the tail, perform step 104-2; if the defect type is abnormal thickness of the strip, perform step 104-3. S104-1 Calculate the rolling force setting deviation rate δ i The head thickness anomaly was analyzed by combining the head temperature prediction deviation, alloy composition deviation rate, and roll gap self-learning change. S104-2. Analyze the tail thickness anomaly by combining the tail temperature deviation and the tail roll gap compensation value. S104-3, combined with furnace dwell time t h and the standard deviation of the loop angle σ j An analysis was conducted on the abnormal thickness of the belt.

2. The method for analyzing the thickness control process of hot-rolled products according to claim 1, characterized in that, The curve data of the product thickness deviation curve in step S101 is denoted as the thickness deviation data set T, where T = {T1, T2, ..., T}. i ,…,T N }, where T i Let represent the thickness deviation at the i-th meter position, in micrometers, and N be the number of data points for the product thickness deviation curve. The product thickness deviation curve is sampled once per meter. To determine if the product thickness deviation curve is valid, calculate the average value T. ave Maximum value T max Minimum value T min The product thickness deviation curve is considered valid if the following conditions are met simultaneously; otherwise, it is considered invalid and the analysis process ends. condition include: T max -T min >3μm; T max -T ave >3μm; T ave -T min >3μm; T max <200μm; T min >-200μm; T max -T min <300μm; N>100。 3. The method for analyzing the thickness control process of hot-rolled products according to claim 2, characterized in that, In step S102, the thickness tolerance includes the upper thickness tolerance and the lower thickness tolerance, wherein the upper thickness tolerance is denoted as Tol. up The thickness tolerance is denoted as Tol. low The unit is micrometer; The thickness deviation data set T is divided into discarded data, head data, tail data, and belt body data; The data to be removed are the first and last S data points of the thickness deviation data set T; The head data mentioned above is the data for the first 3 meters after removing the S head data points from the thickness deviation data set T, i.e., head data = {T} S+1 ,T S+2 ,T S+3 }; The tail data refers to the data for the last 3 meters after removing the last S data points from the thickness deviation data set T, i.e., tail data = {T} N-S-2 ,T N-S-1 ,T N-S }; The tape body data is the thickness deviation data set T after removing discarded data, head data, and tail data, i.e., tape body data = {T} S+4 ,T S+5 ,…,T N-S-3 }; Calculate head thickness deviation T h T h The average of the first 3 meters of data after removing the first S data points from the thickness deviation dataset T is given by: If T h >Tol up or T h <Tol low If the thickness is abnormal, it is considered abnormal; otherwise, the thickness is considered normal. Calculate the tail thickness deviation T t T t The average of the last 3 meters of data after removing the last S data points from the thickness deviation dataset T is given by: If T t >Tol up or T t <Tol low If the tail thickness is abnormal, it is judged as abnormal; otherwise, the tail thickness is judged as normal. Calculate the pass rate P of the body data. It is an indicator function, when T i In [Tol low Tol up The value is 1 if the condition is met, otherwise it is 0. If P <P lim If the thickness is abnormal, it is considered abnormal; otherwise, it is considered normal. Where P... lim The threshold value for the thickness of the belt is [95%, 99%].

4. The method for analyzing the thickness control process of hot-rolled products according to claim 3, characterized in that, S is based on the target thickness of the strip. targ Settings, when H targ When <2.5mm, S=5; when 2.5mm≤H targ When <6.0mm, S=4; when 6.0mm≤H targ At that time, S = 3.

5. The method for analyzing the thickness control process of hot-rolled products according to claim 4, characterized in that, In step S103, the relevant control information for the rolled piece includes steel class SGF, thickness layer Grt_index, and roll sequence number Num; The steel grade SGF is determined based on the steel grade of the rolled piece; The thickness level Grt_index is determined based on the target thickness H, and 23 levels are defined, including: When H < 1.20 mm, the thickness layer Grt_index is 0; When 1.20mm≤H<1.40mm, the thickness layer Grt_index is 1; When 1.40mm≤H<1.60mm, the thickness layer Grt_index is 2; When 1.60mm≤H<1.80mm, the thickness layer Grt_index is 3; When 1.80mm≤H<2.00mm, the thickness layer Grt_index is 4; When 2.00mm≤H<2.25mm, the thickness layer Grt_index is 5; When 2.25mm≤H<2.50mm, the thickness layer Grt_index is 6; When 2.50mm≤H<3.00mm, the thickness layer Grt_index is 7; When 3.00mm≤H<3.50mm, the thickness layer Grt_index is 8; When 3.50mm≤H<4.00mm, the thickness layer Grt_index is 9; When 4.00mm≤H<5.00mm, the thickness layer Grt_index is 10; When 5.00mm≤H<6.00mm, the thickness layer Grt_index is 11; When 6.00mm≤H<7.50mm, the thickness layer Grt_index is 12; When 7.50mm≤H<9.00mm, the thickness layer Grt_index is 13; When 9.00mm≤H<10.50mm, the thickness layer Grt_index is 14; When 10.50mm≤H<12.00mm, the thickness layer Grt_index is 15; When 12.00mm≤H<14.00mm, the thickness layer Grt_index is 16; When 14.00mm≤H<16.00mm, the thickness layer Grt_index is 17; When 16.00mm≤H<18.00mm, the thickness layer Grt_index is 18; When 18.00mm≤H<20.00mm, the thickness layer Grt_index is 19; When 20.00mm≤H<22.00mm, the thickness layer Grt_index is 20; When 22.00mm≤H<24.00mm, the thickness layer Grt_index is 21; When 24.0mm≤H, the thickness layer Grt_index is 22; The roll sequence number Num refers to the production sequence of the rolled piece after the finishing mill roll change. After the finishing mill roll change, the roll sequence number Num is cleared to zero. The roll sequence number Num of the first rolled piece produced after the roll change is set to 1, and so on.

6. The method for analyzing the thickness control process of hot-rolled products according to claim 1, characterized in that, The head thickness anomaly analysis in step 104-1 includes: The deviation rate of the rolling force setting of the finishing mill stand is denoted as δ. i , Among them, FS i The rolling force setting for the i-th stand, in kN; FA i The actual rolling force of the i-th stand is expressed in kN. When the rolling force is set to a deviation rate δ i When the deviation is >15%, determine whether it is the first production of a steel grade specification or the temperature prediction deviation is >20℃, or the alloy deviation rate of carbon, silicon and manganese in the alloy composition is >30%. The alloy composition deviation rate is: When the rolling force is set to a deviation rate δ i When the roll gap is ≤15% and the roll sequence number Num≤3, determine the roll gap self-learning change amount ZBS_DIF. i , ZBS_DIF i =|ZBS_NEW i -ZBS_OLD i |, Where i represents the finishing mill stand number; ZBS_DIF i The roll gap self-learning change of the i-th stand in the finishing mill; ZBS_NEW i The roll gap self-learning update value for the i-th stand of the finishing mill; ZBS_OLD i The initial value for the roll gap self-learning of the i-th stand in the finishing mill; If ZBS_DIF i >0.1 indicates a mismatch between the self-learning function of the material roller gap and the actual operation. When the rolling force is set to a deviation rate δ i When the thickness is ≤15% and the roll sequence number Num>3, determine whether the thickness specification transition is too large. Compare the target thickness deviation between the current workpiece and the previous workpiece. If the absolute value of the deviation is >2.5mm, obtain the roll gap self-learning change of the current workpiece finishing mill stand. ZBS_DIF i =|ZBS_NEW i -ZBS_OLD i | Where i is the finishing mill stand number ZBS_DIF i The roll gap self-learning change of the i-th stand in the finishing mill; ZBS_NEW i The roll gap self-learning update value for the i-th stand of the finishing mill; ZBS_OLD i The initial value for the roll gap self-learning of the i-th stand in the finishing mill; If ZBS_DIF exists i If the value is greater than 0.1, it is determined that the thickness specification is too large.

7. The method for analyzing the thickness control process of hot-rolled products according to claim 1, characterized in that, The tail thickness anomaly analysis in step 104-2 includes: If the average temperature at the tail end of the finishing mill exit deviates from the target temperature at the finishing mill exit by more than 20°C, it is judged as an abnormal temperature prediction. Otherwise, obtain the roll gap compensation values ​​of the workpiece in the three stands after finishing rolling, denoted as gc1, gc2, and gc3, and calculate the average value gc of gc1, gc2, and gc3. a , When the average value of the roll gap compensation at the tail of the last three stands of the finishing mill is gc a When the direction is consistent with the thickness deviation, it is determined to be an abnormality in compensation control.

8. The method for analyzing the thickness control process of hot-rolled products according to claim 7, characterized in that, Average value gc a When the direction is consistent with the thickness deviation, it is judged as an abnormality in compensation control, which means: If the thickness deviation T t >Tol up And GC a >0, or T t <Tol low And GC a <0 indicates that the abnormal thickness at the tail end is caused by tail roll gap compensation control.

9. A method for analyzing the thickness control process of hot-rolled products according to any one of claims 1-8, characterized in that, The anomaly analysis of belt thickness in step 104-3 includes: The furnace dwell time is the heating time of the slab in the heating furnace, denoted as t. h The unit is hours; if the furnace dwell time is t h If the heating time exceeds 5 hours, it is considered an abnormal heating condition. Otherwise, calculate the standard deviation σ of the loop angle. j When σ exists j A value greater than 0.5 indicates an abnormality in the looper control.

10. The method for analyzing the thickness control process of hot-rolled products according to claim 9, characterized in that, Loop angle standard deviation σ j The calculation process is as follows: Obtain the set of looper angle curve data for the finishing mill stand, denoted as L_a. j , L_a j ={L_a j1 ,L_a j2 ,…,L_a jk ,…,L_a jm } Where j is the loop number, L_a j This is a set of angle curve data for loop j; L_a jk This represents the k-th data sample value of the loop j; m is the total number of data points, and one data point of the loop angle curve is collected per second; To calculate the standard deviation of the looper angle curve dataset, and to prevent noise data from the beginning and end of the looper operation from affecting the calculation results, 0.1m data points at each end were removed when calculating the standard deviation and mean. Where, σ j The standard deviation of the angle curve data for the looper j; It is the mean of the data set; Determine if the standard deviation of each loop angle curve data exceeds the limit. If σ exists... j If the value is >U, it is determined that the abnormal thickness of the belt is caused by a looper control problem. U is the standard deviation threshold of the looper angle, and its value range is [0.3, 0.8].

Images