Continuous annealing strip steel edge blocking early warning method based on whole-process big data

Abstract

The invention relates to the technical field of mechanical automation control, in particular to a continuous annealing strip steel blocking early warning method based on full-process big data, which comprises the following steps: acquiring width data of each position in the full-length direction of a strip steel hot rolling outlet in the production process; based on the hot-state width and the thermal expansion coefficient, converting the hot-state width and the thermal expansion coefficient into cold-state data to obtain the cold-state width of each position in the full-length direction of the strip steel; calculating the actual rolling mill reduced scale and the continuous retraction reduced scale; constructing a BP neural network model to obtain a rolling mill reduced scale and a continuous retraction reduced scale; the opening degree of the continuous annealing circle shear is read, and the total edge reduction amount of the strip steel on the two sides of the continuous annealing circle shear is calculated; all the positions of the whole length of the strip steel are judged, when the edge reduction amount of the position of the strip steel is smaller than a danger threshold value, it is judged that the edge blocking risk exists in the position, and the system conducts early warning; according to the method, whether the strip steel has the edge blocking risk at the position of the circle shear or not is accurately predicted by accurately predicting the strip steel reduced scale, stable operation of equipment is guaranteed, and unnecessary shutdown is reduced.

Description

Technical Field

[0001] The present invention relates to the field of mechanical automation control technology, and in particular to a continuous strip steel edge blocking early warning method based on full-process big data. Background Art

[0002] Circular shears, as essential edge shearing equipment, are widely used in the production of plate and strip. Their primary function is to align or trim the longitudinal edges of the moving strip to the customer's target width. As the final process in the cold rolling line, the continuous annealing unit produces the final product, directly destined for the customer. Unstable operation of this equipment can impact the overall production line's pace and quality. However, during production, the strip can become narrower in certain locations due to factors such as the width of the incoming hot-rolled material and strip reduction. When this condition occurs, the strip's edge width decreases as it reaches the circular shear position, significantly increasing the probability of edge blocking and impacting production.

[0003] Since there are no width gauges at the exits of the pickling production line and the continuous annealing production line, in order to accurately control the width of the finished strip, we can only rely on manual sampling inspection to evaluate the strip reduction and control the opening of the disc shear based on the reduction. However, the error in manual evaluation of the reduction is large, and it is impossible to accurately evaluate the reduction of strips of different specifications. Summary of the Invention

[0004] The present invention provides an early warning method for edge blocking of continuously unwound strip steel based on full-process big data. By training manually sampled data and constructing a neural network model, the strip reduction amount is accurately predicted, and then the presence of a risk of edge blocking of the strip steel at the disc shear is accurately predicted, thereby ensuring stable operation of the equipment and reducing unnecessary downtime.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for early warning of edge blocking of continuous strip steel stripping based on full-process big data includes the following steps:

[0007] S1. Obtain the width data of each position along the entire length of the hot-rolled strip outlet during the production process;

[0008] S2. Convert the hot width and thermal expansion coefficient into cold data to obtain the cold width at each position along the entire length of the strip;

[0009] S3. Calculate the actual mill reduction and continuous reduction based on the production data obtained from the manual inspection station;

[0010] S4. Construct a BP neural network model, using the width, thickness and carbon content of the cold-rolled exit strip as input to obtain the mill reduction and continuous reduction;

[0011] S5. Read the opening of the continuous annealing circular shear, and calculate the total edge reduction on both sides of the strip at the continuous annealing circular shear;

[0012] S6. Judge each position along the entire length of the strip. When the edge reduction at a certain position of the strip is less than the danger threshold, it is judged that there is a risk of edge blockage at that position, and the system issues an early warning.

[0013] Further, the cold widths at each position along the entire length of the strip are:

[0014]

[0015] where x is the position coordinate in the strip length direction; W h (x) is the hot rolling exit width at the x position along the entire length of the strip obtained in step 1; α is the strip thermal expansion coefficient; W c (x) is the cold width at the x position along the entire length of the strip.

[0016] Further, the actual mill scale reduction amount is calculated by selecting a production line with a circular shear at the cold rolling inlet, and the calculation method is as follows:

[0017] S r =W ri -W re (2)

[0018] where W ri is the opening width of the circular shear at the cold rolling inlet, that is, the measured width of the strip at the mill inlet; W re is the measured width of the strip at the mill outlet obtained from the electronic report entered at the manual inspection table at the mill outlet; S r is the actual mill scale reduction amount of the strip under this product specification;

[0019] The calculation method of the actual continuous annealing scale reduction amount is as follows:

[0020] S c =W ci -W ce =W re -W ce (3)

[0021] where W ci is the measured width of the strip at the continuous annealing inlet, which is equal to the measured width of the strip at the mill outlet; W ce is the measured width of the strip at the continuous annealing outlet obtained from the electronic report entered at the manual inspection table at the continuous annealing outlet; S c is the actual mill scale reduction amount of the strip under this product specification.

[0022] Further, calculate the total edge reduction on both sides of the strip at the continuous annealing circular shear:

[0023] D(x) = W c (x) - [S r (x) + S c (x) + W s (4)

[0024] Wherein, W c (i) is the cold width at the x position in the full length direction of the strip; S r (i), S c (i) are respectively the mill scale reduction and the continuous annealing scale reduction at the x position in the full length direction of the strip; W s is the opening width of the shear blades on both sides of the continuous annealing circular shear equipment, that is, the target outlet width; D(x) is the total edge reduction on both sides of the continuous annealing circular shear equipment.

[0025] Furthermore, the BP neural network model selects the actual mill scale reduction and continuous annealing scale reduction obtained in step S3 as the training set and the validation set to train and debug the model.

[0026] Compared with the prior art, the beneficial effects of the present invention are:

[0027] The obtained hot rolling outlet width data of the strip is converted into cold state data based on the hot state width and the thermal expansion coefficient, a BP neural network model is established to obtain the mill scale reduction and the continuous annealing scale reduction at this width, and according to the cold state width, the scale reduction and the opening width of the circular shear, the total edge trimming amount on both sides of the strip after continuous annealing is calculated. If the total edge trimming amount is less than the threshold, it is determined that there is a risk of edge blockage at this position of the strip, accurately predicting the scale reduction of the strip, and then accurately predicting whether there is a risk of edge blockage at the circular shear of the strip, preventing the narrow strip from entering the circular shear and causing edge blockage accidents, ensuring the stable operation of the equipment, and reducing unnecessary shutdowns. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 is the method flow chart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0029] The following further describes the specific embodiments of the present invention with reference to the drawings:

[0030] See Figure 1 , which is the method flow chart of the present invention. A method for warning of edge blockage of continuous annealing strip based on full-process big data of the present invention includes the following steps:

[0031] S1. Obtain the width data W h (x) at each position in the full length direction of the strip hot rolling outlet during the production process.

[0032] S2. Convert it into cold state data based on the hot state width and the thermal expansion coefficient to obtain the cold state width at each position in the full length direction of the strip:

[0033]

[0034] Among them, x is the position coordinate in the strip length direction; W h (x) is the hot rolling exit width at the x position in the full length direction of the strip obtained in step 1; α is the strip thermal expansion coefficient; W c (x) is the cold state width at the x position in the full length direction of the strip.

[0035] S3. Calculate the actual mill scale reduction and continuous annealing scale reduction according to the production data obtained from the manual inspection table;

[0036] Select the production line with a rotary shear at the cold rolling inlet to calculate the actual mill scale reduction. The calculation method is as follows:

[0037] S r =W ri -W re (2)

[0038] Among them, W ri is the opening width of the rotary shear at the cold rolling inlet, that is, the measured width of the strip at the mill inlet; W re is the measured width of the strip at the mill outlet obtained from the electronic report entered by the manual inspection table at the mill outlet; S r is the actual mill scale reduction of the strip under this product specification;

[0039] The calculation method of the actual continuous annealing scale reduction is as follows:

[0040] S c =W ci -W ce =W re -W ce (3)

[0041] Among them, W ci is the measured width of the strip at the continuous annealing inlet, which is equal to the measured width of the strip at the mill outlet; W ce is the measured width of the strip at the continuous annealing outlet obtained from the electronic report entered by the manual inspection table at the continuous annealing outlet; S c is the actual mill scale reduction of the strip under this product specification.

[0042] S4. Build a BP neural network model, with the cold rolling exit strip width, cold rolling exit strip thickness and carbon content as inputs, to obtain the mill scale reduction and continuous annealing scale reduction;

[0043] Build a BP neural network model, determine the specific parameters of the model, and train the model; preprocess the historical production data, take the cold width of the strip, the thickness of the strip at the cold rolling exit, and the carbon content of the strip as input parameters, and take the mill scaling amount and the continuous annealing scaling amount as the model output; select the actual mill scaling amount and the continuous annealing scaling amount obtained in step S3 as the training set and the verification set, train and debug the model, and use the trained neural network model to obtain the mill scaling amount S r (x) and the continuous annealing scaling amount S c (x);

[0044] Among them, preprocessing the historical production data includes normalization processing, removing noise and abnormal data, and the normalization processing here uses the maximum-minimum normalization processing;

[0045] Among them, the process of model training is as follows: the data propagates from the input layer through the hidden layer to the output layer. If the model prediction result has a large deviation, the error signal propagates backward to correct the weights of each unit. After continuous cycling until the output error is less than the preset precision value or the learning times are reached.

[0046] S5. Read the opening degree of the continuous annealing slitter, and calculate the total edge reduction amount on both sides of the strip at the continuous annealing slitter;

[0047] D(x) = W c (x) - [S r (x) + S c (x) + W s (4)

[0048] Among them, W c (i) is the cold width at the x position in the full length direction of the strip; S r (i), S c (i) are the mill scaling amount and the continuous annealing scaling amount at the x position in the full length direction of the strip respectively; W s is the opening width of the shear blades on both sides of the continuous annealing slitter equipment, that is, the target exit width; D(x) is the total edge reduction amount on both sides of the continuous annealing slitter equipment.

[0049] S6. Judge each position of the full length of the strip. When the edge reduction amount at this position of the strip is less than the danger threshold, it is judged that there is a risk of edge blockage at this place, and the system gives an early warning; set the danger thresholds D A , D B , where D A > D B ; when D(i) < D A , the system reminds that there is a risk of edge blockage; when D(i) < D B , the system recommends not to reduce the edge.

[0050] The following embodiments are implemented on the premise of the technical solution of the present invention, and detailed implementation manners and specific operation processes are given. However, the protection scope of the present invention is not limited to the following embodiments. The methods used in the following embodiments are all conventional methods unless otherwise specified.

[0051]

Embodiment

[0052] See Figure 1 , the present invention provides a method for warning of edge blocking of continuous annealing strip steel based on the whole-process big data, including the following steps:

[0053] S1. Obtain the width data of each position in the full length direction of the strip steel at the hot rolling outlet during the production process;

[0054] S2. Based on the hot state width and the coefficient of thermal expansion, convert it into cold state data to obtain the cold state width of each position in the full length direction of the strip steel;

[0055] S3. According to the production data obtained from the manual inspection table, calculate the actual mill scale reduction and continuous annealing scale reduction;

[0056] S4. Construct a BP neural network model, with the strip steel width, strip steel thickness and carbon content at the cold rolling outlet as inputs, to obtain the mill scale reduction and continuous annealing scale reduction;

[0057] S5. Read the opening degree of the continuous annealing circular shear, and calculate the total edge reduction amount on both sides of the strip steel at the continuous annealing circular shear;

[0058] S6. Judge each position in the full length of the strip steel. When the edge reduction amount at this position of the strip steel is less than the danger threshold, it is judged that there is a risk of edge blocking at this place, and the system gives an early warning.

[0059] Taking the circular shear at the outlet of a certain continuous annealing production line as an example, the method of this embodiment is described in detail.

[0060] The above S1 is specifically to obtain the hot rolling outlet width W h (x) of each position in the full length direction of the strip steel during the production process. Since there is a large amount of full length width data of a coil of strip steel, only the data of several positions are selected for calculation and description here, as shown in Table 1.

[0061] Table 1 Strip steel hot rolling outlet width data

[0062] Position x / mm 1 100 200 300 400 494 <![CDATA[W h (x) / mm]]> 1238.44 1237.56 1238.97 1238.15 1236.33 1221.04

[0063] S2. Obtain the coefficient of thermal expansion α for the strip steel to be converted from the hot state to the cold state. In this embodiment, the coefficient of thermal expansion α is taken as 1.015 to obtain the cold state width of each position in the full length direction of the strip steel;

[0064]

[0065] Among them, x is the position coordinate in the length direction of the strip steel; W h (x) is the hot rolling exit width at the position x in the entire length direction of the strip steel obtained in step 1; α is the thermal expansion coefficient of the strip steel; W c (x) is the cold state width at the position x in the entire length direction of the strip steel, and the calculation results are shown in Table 2.

[0066] Table 2 Cold state width data of the strip steel at some positions

[0067] Position x / mm 1 100 200 300 400 494 <![CDATA[W c (x) / mm]]> 1220.138 1219.271 1220.66 1219.852 1218.059 1202.995

[0068] S3. Calculate the actual mill scale reduction and continuous annealing scale reduction according to the production data obtained from the manual inspection table;

[0069] Select the production line with a slitting shear at the cold rolling entrance to calculate the actual mill scale reduction, and its calculation method is as follows:

[0070] S r =W ri -W re (6)

[0071] Among them, W ri is the opening width of the slitting shear at the cold rolling entrance, that is, the measured width of the strip steel at the mill entrance; W re is the measured width of the strip steel at the mill exit obtained from the electronic report form entered by the manual inspection table at the mill exit; S r is the actual mill scale reduction of the strip steel under this product specification;

[0072] The calculation method of the actual continuous annealing scale reduction is as follows:

[0073] S c =W ci -W ce =W re -W ce (7)

[0074] Among them, W ci is the measured width of the strip steel at the continuous annealing entrance, which is equal to the measured width of the strip steel at the mill exit; W ce is the measured width of the strip steel at the continuous annealing exit obtained from the electronic report form entered by the manual inspection table at the continuous annealing exit; S c is the actual mill scale reduction of the strip steel under this product specification.

[0075] In this implementation case, 2000 data are selected as the training set and 500 data are selected as the validation set.

[0076] S4. Construct a BP neural network model, determine the specific parameters of the model, and train the model. Preprocess the historical production data, taking the cold width of the strip, the thickness of the strip at the cold rolling exit, and the carbon content of the strip as input parameters, and taking the mill scale reduction and continuous annealing scale reduction as the model outputs. Select the actual mill scale reduction and continuous annealing scale reduction obtained in step 3 as the training set and validation set, and train and debug the model. Use the trained neural network model to obtain the mill scale reduction S r (x) and the continuous annealing scale reduction S c (x);

[0077] Among them, preprocessing the historical production data includes normalization processing, removing noise points and abnormal data. The normalization processing here uses the maximum-minimum normalization processing;

[0078] Among them, the process of model training is as follows: The data propagates backward from the input layer through the hidden layer to the output layer. If the model prediction result has a large deviation from the actual value, the error signal propagates backward to correct the weights of each unit. After continuous cycling until the output error is less than the preset precision value or the number of learning times is reached;

[0079] In this embodiment, the mill scale reduction of the strip obtained by this model is 3 mm, and the continuous annealing scale reduction is 1 mm.

[0080] S5. Read the opening of the continuous annealing slitter. In this embodiment, the opening of the slitter is 1188 mm, and calculate the total edge reduction of the strip at the continuous annealing slitter;

[0081] D(x) = W c (x) - [S r (x) + S c (x) + W s (8)

[0082] Among them, W c (i) is the cold width of the strip at the x position in the full length direction of the strip; S r (i), S c (i) are the mill scale reduction and continuous annealing scale reduction of the strip at the x position in the full length direction of the strip respectively; W s is the opening width of the shear blades on both sides of the continuous annealing slitter equipment, that is, the target exit width; D(x) is the total edge reduction on both sides of the continuous annealing slitter equipment, as shown in Table 3.

[0083] Table 3 Total edge reduction of the strip at the continuous annealing slitter at some positions.

[0084] Position x / mm 1 100 200 300 400 494 D(x) / mm 28.138 27.271 28.66 27.852 26.059 10.995

[0085] S6. Judge each position along the entire length of the strip steel. When the edge reduction at a certain position of the strip steel is less than the danger threshold, it is judged that there is a risk of edge blockage at this position, and the system gives an early warning; the danger threshold D is set according to experience. A , D B , where D A > D B ; when D(i) < D A , the system reminds that there is a risk of edge blockage; when D(i) < D B , the system suggests not reducing the edge. In this implementation case, D A is taken as 20 mm, and D B is taken as 16 mm. Then the prediction results of the edge blockage risk at some positions of the strip steel are shown in Table 4.

[0086] Table 4 Prediction Results of Edge Blockage Risk at Some Positions of the Strip Steel

[0087]

[0088] After applying this method to the actual on-site production, the total edge reduction curve graph at the continuous annealing slitting shear can be visually obtained through data recording, which can assist the operator to prevent the edge blockage accident caused by too narrow strip steel entering the slitting shear; based on the full-length information of the hot-rolled strip steel incoming material and combined with the scale reduction prediction model established by the neural network, for strip steels of different varieties and specifications, calculate its total edge reduction at the continuous annealing slitting shear, so as to judge whether there is a risk of edge blockage, and give corresponding suggestions to ensure the continuous and stable production process and reduce unnecessary shutdowns.

Claims

1. A method for predicting edge blockage of continuous annealing strip steel based on full-process big data, characterized in that, It includes the following steps: S1. Obtain the width data at each position in the full length direction of the strip hot rolling outlet during the production process; S2. Based on the hot width and the thermal expansion coefficient, convert it into cold state data to obtain the cold width at each position in the full length direction of the strip; S3. Calculate the actual mill scaling amount and continuous annealing scaling amount according to the production data obtained from the manual inspection table; S4. Construct a BP neural network model, with the strip width at the cold rolling outlet, the strip thickness at the cold rolling outlet, and the carbon content as inputs, to obtain the mill scaling amount and continuous annealing scaling amount; S5. Read the opening degree of the continuous annealing slitting shear, and calculate the total edge reduction amount on both sides of the strip at the continuous annealing slitting shear; S6. Judge each position in the full length of the strip. When the edge reduction amount at this position of the strip is less than the danger threshold, it is judged that there is a risk of edge blockage at this position, and the system gives an early warning.

2. The edge blocking warning method for continuous annealing strip steel based on the whole-process big data according to claim 1, wherein The cold width at each position in the full length direction of the strip is: where x is the position coordinate in the strip length direction; W h (x) is the hot rolling exit width at position x in the full length direction of the strip obtained in step 1; α is the strip thermal expansion coefficient; W c (x) is the cold width at position x in the full length direction of the strip.

3. A method for warning of edge blocking of continuous annealing strip steel based on full-process big data according to claim 1, characterized in that, The actual mill scaling amount is calculated by selecting a production line with a slitting shear at the cold rolling inlet, and the calculation method is as follows: S r = W ri - W re (2) Among them, W ri is the opening width of the slitting shear at the cold rolling inlet, that is, the measured width of the strip steel at the mill inlet; W re is the measured width of the strip steel at the mill outlet obtained from the electronic report form entered by the manual inspection table at the mill outlet; S r is the actual mill reduction of the strip steel under this product specification; The calculation method of the actual continuous annealing scaling amount is as follows: S c = W ci - W ce = W re - W ce (3) Among them, W ci is the measured width of the strip at the entry of the continuous annealing, which is equal to the measured width of the strip at the exit of the rolling mill; W ce is the measured width of the strip at the exit of the continuous annealing obtained from the electronic report form entered on the manual inspection table at the exit of the continuous annealing; S c is the actual rolling mill scaling amount of the strip under this product specification.

4. A method for warning of edge blocking of continuous annealing strip steel based on the whole-process big data according to claim 1, characterized in that, The calculation of the total edge reduction amount on both sides of the strip at the continuous annealing slitting shear: D(x) = W c (x) - [S r (x) + S c (x) + W s (4) Among them, W c (i) is the cold width at the x position in the longitudinal direction of the strip steel; S r (i), S c (i) are respectively the mill scale reduction amount and the continuous annealing scale reduction amount at the x position in the longitudinal direction of the strip steel; W s is the opening width of the shear blades on both sides of the continuous annealing slitting shear equipment, that is, the target outlet width; D(x) is the total edge reduction amount on both sides of the continuous annealing slitting shear equipment.

5. A method for warning of edge blocking of continuous annealing strip steel based on the whole-process big data according to claim 1, characterized in that, The BP neural network model selects the actual mill scaling amount and continuous annealing scaling amount obtained in step S3 as the training set and validation set to train and debug the model.

Images