CN111266405B - Plate and strip hot rolling process control method and control device - Google Patents
Plate and strip hot rolling process control method and control device Download PDFInfo
- Publication number
- CN111266405B CN111266405B CN202010123629.0A CN202010123629A CN111266405B CN 111266405 B CN111266405 B CN 111266405B CN 202010123629 A CN202010123629 A CN 202010123629A CN 111266405 B CN111266405 B CN 111266405B
- Authority
- CN
- China
- Prior art keywords
- self
- rolling process
- hot rolling
- learning
- solution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
- B21B—ROLLING OF METAL
- B21B1/00—Metal-rolling methods or mills for making semi-finished products of solid or profiled cross-section; Sequence of operations in milling trains; Layout of rolling-mill plant, e.g. grouping of stands; Succession of passes or of sectional pass alternations
- B21B1/22—Metal-rolling methods or mills for making semi-finished products of solid or profiled cross-section; Sequence of operations in milling trains; Layout of rolling-mill plant, e.g. grouping of stands; Succession of passes or of sectional pass alternations for rolling plates, strips, bands or sheets of indefinite length
- B21B1/24—Metal-rolling methods or mills for making semi-finished products of solid or profiled cross-section; Sequence of operations in milling trains; Layout of rolling-mill plant, e.g. grouping of stands; Succession of passes or of sectional pass alternations for rolling plates, strips, bands or sheets of indefinite length in a continuous or semi-continuous process
- B21B1/26—Metal-rolling methods or mills for making semi-finished products of solid or profiled cross-section; Sequence of operations in milling trains; Layout of rolling-mill plant, e.g. grouping of stands; Succession of passes or of sectional pass alternations for rolling plates, strips, bands or sheets of indefinite length in a continuous or semi-continuous process by hot-rolling, e.g. Steckel hot mill
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
- B21B—ROLLING OF METAL
- B21B37/00—Control devices or methods specially adapted for metal-rolling mills or the work produced thereby
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Feedback Control In General (AREA)
- Control Of Metal Rolling (AREA)
Abstract
The invention discloses a method for controlling a plate and strip hot rolling process, which comprises the following steps: acquiring n values { Z of self-learning coefficients of a hot rolling process control model corresponding to preset control parameters in a preset time periodiAnd with each ZiCorresponding update time { Ti}; identifying threshold ranges for self-learning coefficients, and in { ZiThe number m of values outside the threshold range; the processor judges whether the ratio of m/n is within a preset range; if yes, for { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}; according to { ZjCorresponding update time TjGet at each TjInput quantities of the control model on the time node; analyzing the input quantity, and adjusting the input quantity in the subsequent rolling process; controlling the hot rolling process according to the adjusted input quantity; by the method, the reduction of the control precision of the plate and strip hot rolled product caused by the faulty operation of the control model program in the rolling process can be avoided.
Description
Technical Field
The application relates to the technical field of hot-rolled plate strip production control, in particular to a plate and strip hot-rolling process control method and a plate and strip hot-rolling process control device.
Background
A rolling process control system formed by combining process control model programs is arranged in a hot-rolled plate strip production line, and various indexes such as the size of a plate and strip product, a control process and the like in the production process are calculated, measured, fed back and controlled. Taking the size control in the hot rolling of the strip steel as an example, the control process of the process control system is as follows: firstly, inputting data such as the size of a supplied material and the target size of a finished product into a model program for a series of calculations according to the input data, wherein the calculation results comprise two types: the first kind of data is directly used for controlling the action of each device on the production line, such as the width control and the thickness control of the strip steel; and the second type of data is predicted value data of each index of the strip steel size after the actual production process is controlled according to the first type of data. Various detecting instruments are also arranged on the hot rolling production line and are used for measuring various indexes such as the size of the strip steel and feeding back actual data. The model program will compare the measured data of these meters with the second type of data predicted by itself, and if there is a deviation, will automatically add a coefficient to the calculation process to compensate for the calculated deviation. The process is called model program self-learning, and the coefficients of the model which are automatically increased are called model program self-learning coefficients.
At present, when the control of a certain plate and strip produced by a rolling line is not proper, the fault factors generating problems in a certain time period need to be researched, and the cause and the solution of the problems are found by analyzing the process control data of the plate and strip in the time period, including the input quantity and the output quantity of a control model. However, the data of the rolling process is collected in real time along with the rolling process, a large amount of data can be generated by rolling one plate blank, and the hot rolling production process of the plate blank is continuously carried out, so that the data of the hot rolling process has the characteristics of large data total amount and difficult regular analysis. Therefore, a solution is needed that can quickly determine key data from a large amount of process control data to improve data analysis accuracy and problem resolution efficiency.
Disclosure of Invention
The invention provides a plate and strip hot rolling process control method and a plate and strip hot rolling process control device, which aim to solve or partially solve the technical problems of insufficient pertinence of control data analysis, large analysis workload and low efficiency of the existing hot rolling process.
In order to solve the technical problem, the invention provides a method for controlling a plate and strip hot rolling process, which specifically comprises the following steps:
the processor obtains n values { Z of self-learning coefficients of the hot rolling process control model corresponding to the preset control parameters in a preset time periodiAnd with each ZiCorresponding update time { Ti}; n is more than or equal to 2 and is a positive integer, and i sequentially takes the values of 1,2, … and n;
the processor identifies the threshold range of the self-learning coefficients, and is set at { ZiThe number m of values outside the threshold range;
the processor judges whether the ratio of m/n is within a preset range;
if so, the processor pair { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}，j∈[1,n]；
The processor is according to { ZjCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node;
the processor analyzes the input quantity and adjusts the input quantity in the subsequent rolling process;
and the controller controls the hot rolling process according to the adjusted input quantity.
Optionally, the processor determines a threshold range of the self-learning coefficient, and specifically includes:
confirming threshold value range lim of self-learning coefficient belongs to [ P ∈ [ [ P ]l×Set,Pu×Set]；
Wherein, PlAnd PuAnd Set is the output value of the preset control parameter output by the hot rolling process control model when the self-learning coefficient is 0.
Further, PlHas a value range of [ -0.3, -0.05 [)]；PuHas a value range of [0.05,0.3 ]]。
According to the technical scheme, the preset range is [0.05,0.6 ].
As with the above-described solution, processor pairs { Z }iPerforming regression analysis to confirm the period value-taking points { Z }jThe method specifically comprises the following steps:
with { ZiIs the y-axis, { TiThe sequence of integers [ N ] determined in ascending order of timei]Is the x-axis, NiSequentially taking values of 1,2, …, n; to { ZiPerforming polynomial fitting regression analysis to obtain a corresponding regression equation, which is as follows:
y＝cpxp+cp-1xp-1+……+c1x+c0(ii) a Wherein x has a domain of [0, n ]]P is not less than 3 and is a positive integer;
performing first order differentiation on the regression equation to obtain a first order differential equation, and determiningSense Domain [0, n]Solving set of first order differential equations { x }jThe first order differential equation is specifically as follows:
p·cpxp-1+(p-1)·cp-1xp-2+……+c2x+c1＝0；
judging solution set { xjSolution x injWhether the number of (a) is more than two;
if yes, confirming that the values of the self-learning coefficients have periodic distribution, and collecting a solution set { xjEach solution x injRounding to obtain integer set Xj]；
According to [ X ]j]From { N }iDetermining X in the sequencej＝NjA point of (a); then according to { NjFrom { Z }iDetermine { Z }jWill { Z }jEach of the points is determined as a period value-taking point.
Further, a solution set { x is judgedjSolution x injWhether the number of (2) is more than two, further comprising:
if xjThe number of the regression equation is less than two, the regression equation is subjected to second order differentiation to obtain a second order differential equation, and the second order differential equation is positioned in a definition domain [0, n ]]Solving set of second order differential equations { x }kThe second order differential equation is specifically as follows:
p(p-1)·cpxp-2+(p-1)(p-2)·cp-1xp-3+……+c3x+c2＝0；
judging solution set { xkSolution x inkWhether the number of (a) is more than two;
if yes, confirming that the values of the self-learning coefficients have periodic distribution, and collecting a solution set { xkEach solution x inkRounding to obtain integer set Xk]；
According to [ X ]k]From { N }iDetermining X in the sequencek＝NkA point of (a); then according to { NkFrom { Z }iDetermine { Z }kWill { Z }kDetermining each point in the data as a period value taking point;
the processor is according to { ZjCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node specifically comprises the following steps:
the processor is according to { ZkCorresponding update time TkGet at each TkThe input of the hot rolling process control model at the time node.
Based on the same inventive concept of the above technical scheme, the invention also provides a plate and strip hot rolling process control device, comprising:
a first obtaining module, configured to obtain n values { Z ] of self-learning coefficients of the hot rolling process control model corresponding to the preset control parameters in a preset time periodiAnd with each ZiCorresponding update time { Ti}; n is more than or equal to 2 and is a positive integer, and i sequentially takes the values of 1,2, … and n;
a validation module for validating a threshold range of the self-learning coefficients, and in { Z }iThe number m of values outside the threshold range;
the judging module is used for judging whether the ratio of m/n is within a preset range or not;
if yes, for { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}，j∈[1,n]；
A second obtaining module for obtaining ZjCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node;
the adjusting module is used for analyzing the input quantity and adjusting the input quantity in the subsequent rolling process;
and the control module is used for controlling the hot rolling process according to the adjusted input quantity.
Based on the same inventive concept of the above technical solutions, the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and when the processor executes the program, the steps of any one of the control methods in the above technical solutions are implemented.
Based on the same inventive concept of the above technical solutions, the present invention further provides a computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, implements the steps of any one of the control methods of the above technical solutions.
Through one or more technical schemes of the invention, the invention has the following beneficial effects or advantages:
the invention provides a method for performing hot rolling control by judging and analyzing a process control model self-learning coefficient, which determines a period point of the change of the self-learning coefficient through regression analysis, performs targeted extraction and analysis on process control model input data of the period point, can obviously reduce the analysis workload compared with the method of nondifferential extraction and analysis of all hot rolling process data, extracts the model input data of the corresponding period point by combining the analysis of the change period of the self-learning coefficient, can more quickly and accurately discover implicit rules in the data, thereby more timely and accurately adjusting the hot rolling process control, and avoids the reduction of the control precision of a strip steel product caused by the faulty operation of a rolling process control model program.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 shows a schematic flow diagram of a method for controlling a hot rolling process of a sheet and strip according to an embodiment of the invention;
FIG. 2 illustrates a sample data scatter plot of coiling temperature self-learning coefficients according to one embodiment of the present invention;
FIG. 3 shows a schematic diagram of a polynomial regression fitting of the coiling temperature self-learning coefficients according to one embodiment of the invention;
fig. 4 is a schematic structural view showing a hot rolling process control apparatus for a plate and strip according to an embodiment of the present invention.
Detailed Description
In order to make the present application more clearly understood by those skilled in the art to which the present application pertains, the following detailed description of the present application is made with reference to the accompanying drawings by way of specific embodiments.
The application process principle of the self-learning coefficient in the process control model is introduced in the background technology, and in the existing control scheme, in order to avoid the problem that the self-learning coefficient determined by the process control model through autonomous calculation is too large, the self-learning coefficient of the model program is limited by an upper limit and a lower limit. The inventor finds out through research that the problems in the rolling process control can be reflected by the special monitoring, analysis and problem diagnosis of the self-learning coefficient of the model program. This is because many seemingly high-precision model program calculation results actually have a large value of model program self-learning coefficients in coordinating the control model, which is an abnormal phenomenon. On the other hand, when a problem occurs in the rolling line, the self-learning coefficients regularly and periodically fluctuate, and the problem existing in the hot rolling line can be revealed by analyzing the change rule of the self-learning coefficients, however, the problem cannot be found from the calculation result of the process control model program, because the model program makes the input quantity appear to be in the normal control range by adjusting the self-learning coefficients.
Based on the research findings, the inventor provides a control method for diagnosing the rolling line problem by utilizing the self-learning coefficient through a large amount of data tracking and analysis, diagnoses the fault by monitoring and analyzing the self-learning coefficient of the model program, finds the problem in time and processes the problem, and can avoid the problem that the process control precision of the plate and strip product is reduced due to faulty operation of the model program, wherein the overall thought of the control method is as follows:
the invention provides a plate and strip hot rolling process control method, which specifically comprises the following steps:
s1: the processor obtains n values { Z of self-learning coefficients of the hot rolling process control model corresponding to the preset control parameters in a preset time periodiAnd with each ZiCorresponding update time { Ti}; n is more than or equal to 2 and is a positive integer, and i sequentially takes the values of 1,2, … and n;
s2: the processor identifies the threshold range of the self-learning coefficients, and is set at { ZiThe number m of values outside the threshold range;
s3: the processor judges whether the ratio of m/n is within a preset range;
if so, the processor pair { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}，j∈[1,n]；
S4: the processor is according to { ZjCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node;
s5: the processor analyzes the input quantity and adjusts the input quantity in the subsequent rolling process;
s6: and the controller controls the hot rolling process according to the adjusted input quantity.
In this embodiment, { ZiDenotes a data set composed of n values of a self-learning coefficient Z corresponding to a certain control parameter, and { T }iIs a data set of self-learning coefficient update times, TiIs a reaction of with ZiA corresponding update time; { ZjIs analyzed by regression from { Z }iThe data set formed by the self-learning coefficient periodic value points determined in the step (f) { T }jIs equal to { Z }jThe corresponding update time data set.
In summary, the control method provided in this embodiment is to obtain the self-learning coefficients of the model program, determine whether the number proportion of the self-learning coefficients exceeding the preset threshold range is within the preset range, if so, indicate that the calculation of the process control model is valid, perform regression analysis on the values of all the self-learning coefficients, determine the value-taking points from the periodic variation of the learning coefficients, extract the process model input quantity data at the update time nodes according to the update time corresponding to the period value-taking points, analyze the process model input quantity data, adjust the input quantity according to the analysis result, and control the hot rolling. The period points of the self-learning coefficient change are determined through regression analysis, process control model input data of the period points are extracted and analyzed in a targeted mode, compared with the mode that all hot rolling process data are extracted and analyzed in a non-differential mode, the analysis workload can be reduced remarkably, the model input data of the corresponding period points are extracted in combination with the change period analysis of the self-learning coefficient, the hidden rules in the data can be found more quickly and accurately, and therefore the hot rolling process control can be adjusted more timely and accurately. The implementation of each step is described in detail below:
for S1, the data acquisition process may be performed in the secondary system of the hot rolling line: and writing a corresponding program in the process control system for implementation. Specifically, the computer room terminal connected to the secondary system may be programmed in C + + language, and the model program self-learning coefficients and corresponding update time points stored in the secondary system may be written into a data recording or extracting program, and extracted or recorded into a local log file or a local database for analysis. Here, various data acquisition and storage programs can be written as needed, as long as the functions are guaranteed to be realized and data can be acquired from the secondary system to the local terminal or the database.
The preset control parameters refer to determined rolling control parameters to be analyzed, such as the width and thickness of the plate and strip, the hot rolling finishing temperature of the plate and strip, the coiling temperature and the like; each rolling control parameter corresponds to a process control model and also corresponds to a model program self-learning coefficient, for example, the coiling temperature corresponds to a coiling temperature process control model and also has a coiling temperature self-learning coefficient. In the preset time period, the time range to be analyzed can be determined according to actual requirements, and in the time range, the value of the self-learning coefficient Z changes along with the actually rolled plate and strip, so that the self-learning coefficient in the preset time period needs to be acquiredAll values of ZiAnd each of ZiCorresponding update time TiForm a self-learning coefficient value data set { ZiAnd update time data set Ti}; wherein ZiAnd TiThe sorting can be carried out in ascending time sequence, for example, the preset time period is 0: 00-24: 00 of a certain day, and the self-learning coefficient { Z of the temperature is coiled in the time rangeiAnd the corresponding { T }iThe total number of data in the data is 20, which shows that the values of self-learning coefficients of the coiling temperature control model are 20 in total within one day, wherein Z10Is shown at T10Value of the coiling temperature self-learning coefficient on the time node, T10Line 10 in time series.
For S2 and S3, the self-learning coefficients of the model program obtained in S1 are initially judged according to the determined threshold range; if the number proportion of the self-learning coefficients exceeding the threshold range reaches above a given preset value, the calculation process of the model program is considered to be invalid, at the moment, the problem root should be directly searched according to the input quantity required by the calculation of the model program, and the self-learning coefficients of the model program are not further analyzed; if the number proportion of the self-learning coefficients exceeding the threshold range is within the given preset value, the calculation process is effective, and the self-learning coefficients should be analyzed at the moment.
In some alternative embodiments, S2: the processor confirms the threshold range of the self-learning coefficient, and specifically comprises the following steps: confirming threshold value range lim of self-learning coefficient belongs to [ P ∈ [ [ P ]l×Set,Pu×Set](ii) a Wherein, PlAnd PuAnd Set is the output value of the preset control parameter output by the hot rolling process control model when the self-learning coefficient is 0.
Lim in the scheme is a threshold range of a self-learning coefficient of the model program, Set is a calculation result of the process control model program in which the self-learning coefficient participates if and only if the value of the self-learning coefficient is 0; taking the coiling temperature as an example, when the value of the coiling temperature self-learning coefficient is 0, the value output by calculation of a certain mark of the hot rolled coil is 600 ℃, and then the Set value is 600. PlIs the lower percentage of the threshold range, PuIs the upper percentage of the threshold range, as measured by the Set value and Pl、PuThe range of lim can be determined.
Optionally, PlHas a value range of [ -0.3, -0.05 [)]；PuHas a value range of [0.05,0.3 ]]. For example, if Pl＝-0.05，Pu0.05, Set 600; the threshold range lim of the self-learning coefficient of the grade hot rolled coil is [ -30,30 [ -30 [ ]]。
For the preset range in S3, the collected self-learning coefficients { Z ] are characterizediWhether the proportion of the number m of the self-learning coefficients exceeding the threshold range lim is within a judgment interval or not is judged; if m is lower than the lower limit of the preset range (judgment interval), it is indicated that the process control model is normal and does not need to analyze the self-learning coefficient, and if m is higher than the upper limit of the preset range (judgment interval), it is indicated that the calculation process of the process control model is invalid, the problem root should be directly searched from the input quantity required by the model program calculation, and further analysis on the self-learning coefficient of the model program is unnecessary; if the number proportion of the self-learning coefficients exceeding the threshold range lim falls within the preset range (judgment interval), the calculation process of the process control model is effective, but the calculation and the value of the self-learning coefficients reflect that certain problems exist in process control, and at the moment, the model program self-learning coefficients should be further analyzed as follows. Optionally, the preset range is [0.05,0.6]](ii) a Preferably, the lower limit value may be 0.1 or 0.15; the preferred upper range may be 0.4 or 0.5.
When the calculation process of the process control model is valid, the periodicity of the values of the self-learning coefficients should be analyzed, and based on the same inventive concept as the foregoing embodiment, in other alternative embodiments, a specific analysis method is proposed: processor pair { ZiPerforming regression analysis to confirm the period value-taking points { Z }jThe method specifically comprises the following steps:
s31: with { ZiIs the y-axis, { TiThe sequence of integers [ N ] determined in ascending order of timei]Is the x-axis, NiSequentially taking values of 1,2, …, n; to { ZiPerforming polynomial fitting regression analysis to obtain pairsThe regression equation should be as follows:
y＝cpxp+cp-1xp-1+……+c1x+c0(ii) a Wherein x has a domain of [0, n ]]P is not less than 3 and is a positive integer;
s32: performing first order differentiation on the regression equation to obtain a first order differential equation, and defining the domain [0, n ]]Solving set of first order differential equations { x }jThe first order differential equation is specifically as follows:
p·cpxp-1+(p-1)·cp-1xp-2+……+c2x+c1＝0；
s33: judging solution set { xjSolution x injWhether the number of (a) is more than two;
s34: if yes, confirming that the values of the self-learning coefficients have periodic distribution, and collecting a solution set { xjEach solution x injRounding to obtain integer set Xj]；
S35: according to [ X ]j]From { N }iDetermining X in the sequencej＝NjA point of (a); then according to { NjFrom { Z }iDetermine { Z }jWill { Z }jEach of the points is determined as a period value-taking point.
In general, regression analysis of the values of all self-learning coefficients using polynomial fitting regression can be programmed to { T } in ascending time orderjSorting is carried out, and according to the sorting sequence, each self-learning coefficient Z is sortediGiving a sequence value of an integer to obtain an integer sequence data set { N } corresponding to the values of all self-learning coefficientsi}; for example, Z10The value of the self-learning coefficient is at the 10 th time point T10Is obtained, and the corresponding sequence value is N1010, set { N in integersiIs an abscissa structure, { ZiAnd (4) constructing a coordinate system on the total left side, putting the values of all self-learning coefficients into the coordinate system, and performing polynomial regression fitting on the values to obtain a regression equation: y ═ cpxp+cp-1xp-1+……+c1x+c0Which isWherein y is the value of the self-learning coefficient, and the definition domain of x is [0, n](ii) a And then carrying out differential calculation on the regression equation, firstly carrying out first-order differentiation on the regression equation, and finding out all stagnation points or stable points existing in the regression equation by solving the solution of the first-order differential equation, wherein the stagnation points indicate that the values of the self-learning coefficients have some special critical points, and extreme points (necessary and insufficient conditions) possibly having the self-learning coefficients are found out at the critical points. Combining a large amount of data statistics and analysis, for the field of hot rolling of plate and strip, if the solution set { x ] of the first order differential equation of the self-learning coefficient regression equationjThe number of solutions in the method is more than two, according to the probability statistics theory, the original regression function can be considered to be a periodic function, and then the self-learning coefficient of the model program is considered to have periodic fluctuation. In combination with accumulated experience in the field of hot rolling process control, the critical point of the periodic variation of the self-learning coefficient is usually located on the stagnation point, so that the input quantity or other parameters of the process control model on the time node corresponding to the stagnation point need to be focused, and the key factor of abnormal process control is often implied. By analyzing the input quantity or the model at the stagnation point (namely, the period point), the workload of data analysis is remarkably reduced, and more importantly, the root cause of the process control abnormity can be more accurately found, so that the accuracy of process control diagnosis is remarkably improved.
Thus, a solution set { x } of the first order differential equation is foundjAfter that, the normalization process of rounding is performed, and a rounding method can be adopted, namely, a rule that each solution is closest to which natural number and is recorded as the natural number is used for solution set { x }jStandardizing. Such as the solution x19.2, then normalized to 9; if the solution is x2Normalized to 12, constituting a standard solution set { X ═ 11.6j}. Then, on an equal basis, from the integer set { N }iDetermining a self-learning coefficient set (Z) corresponding to the stagnation pointj}, e.g. XjWhen it is 9, then it corresponds to Z9(ii) a Then will { ZjAll points in the sequence are determined as period value taking points, and Z is obtainedjCorresponding TjAll inputs to the process control model on the time nodeVolume, or other parameters for further analysis or fault diagnosis.
Through a large amount of data statistical analysis, preferably, the order of polynomial regression is 6, and regression fitting can be more accurately performed on the value of the self-learning coefficient in the hot rolling field, so that the concrete form of the regression equation is as follows:
y＝c6x6+c5x5+c4x4+c3x3+c2x2+c1x+c0；
the first order differential equation of the regression equation to solve is as follows:
6c6x5+5c5x4+4c4x3+3c3x2+2c2x+c1＝0。
if the number of solutions of the first-order differential equation in the definition domain [0, n ] is less than two, it is indicated that the value of the self-learning coefficient is not periodic, and at this time, another periodic analysis method is provided, which is specifically as follows:
in alternative embodiments, the decision set { x } is determined based on the same inventive concept as the previous embodimentjSolution x injWhether the number of (2) is more than two, further comprising:
s36: if xjThe number of the regression equation is less than two, the regression equation is subjected to second order differentiation to obtain a second order differential equation, and the second order differential equation is positioned in a definition domain [0, n ]]Solving set of second order differential equations { x }kThe second order differential equation is specifically as follows:
p(p-1)·cpxp-2+(p-1)(p-2)·cp-1xp-3+……+c3x+c2＝0；
s37: judging solution set { xkSolution x inkWhether the number of (a) is more than two;
s38: if yes, confirming that the values of the self-learning coefficients have periodic distribution, and collecting a solution set { xkEach solution x inkRounding to obtain integer set Xk]；
S39: according to [ X ]k]From { N }iDetermining X in the sequencek＝NkA point of (a); then according to { NkFrom { Z }iDetermine { Z }kWill { Z }kDetermining each point in the data as a period value taking point;
s4: the processor is according to { ZjCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node specifically comprises the following steps:
s41: the processor is according to { ZkCorresponding update time TkGet at each TkThe input of the hot rolling process control model at the time node.
Specifically, when the solution of the first-order differential equation is less than two, the value of the self-learning coefficient has no periodicity, the regression equation is further subjected to second-order differentiation, and the second-order differential equation is solved in the definition domain [0, n ]]Solution set of { xk}; or, taking 6 th-order polynomial regression as an example, the second order differential equation of the regression equation to be solved is as follows:
30c6x4+20c5x3+12c4x2+6c3x+2c2＝0。
if the number of the second-order differential equation in the solution is more than or equal to two, the change degree of the self-learning coefficient shows periodicity, and further the change of the self-learning coefficient of the model program is considered to have periodic fluctuation, and the solution set { x ] of the second-order differential equationkThe point where the self-learning coefficient is located is the periodic point of the variation trend of the self-learning coefficient. In principle as in the above embodiment, will { x }kRounding each solution to the norm and rounding according to xk＝NkFrom { N }iIn this step, { N } is confirmedkFind out the corresponding ZkAnd Tk(ii) a Finally, according to { TkAnd acquiring all input quantities of the process control model on the corresponding time node, or further analyzing or diagnosing faults of other parameters.
If the solution set of the second order differential equation { x }kThe number of solutions in this case is still less than two from the production experience, which is the hot rolling hardware setupIn preparation for the serious problem, rather than the abnormality occurring in the process control calculation process, the operation of the line rolling equipment should be checked at the first time.
After all the model input quantities obtained by periodic value taking are obtained, the specific problems hidden in the model input quantities need to be analyzed. For the input analysis in S5, the input data may be imported into statistical analysis software, such as Minitab, MatLab, SPSS, etc., or the data may be cleaned to construct a data set, pattern recognition and prediction may be performed using a preset hot rolling production data mining algorithm, the machine learning or data mining result may be intelligently output, the input of the model may be adjusted according to the output result, and then the adjusted input may be fed back to the secondary system according to step S6, so as to control the subsequent hot rolling process, thereby eliminating the anomaly of model calculation in the production process of the hot rolling production line.
By the analysis method of the self-learning coefficient in the embodiment, the purposes of timely diagnosing faults, finding problems and processing are achieved, and the problem that the control precision of plate and strip products is reduced due to faulty operation of a model program is avoided.
In the following examples, steel grades SPHC-W were produced during the period of 2019, month 11 to 2019, month 9, month 15, thickness specification: 3.0mm, target temperature: the above method will be described in detail by taking an example of abnormal analysis of the coiling temperature process control of 580 ℃ series products:
1. the self-learning coefficient of the coiling temperature model program and the corresponding updating time are monitored, recorded and obtained from the secondary system, and the process is mainly realized by writing a corresponding program in a control system. A program commonly used in the current industry is a C + + language program, and the model program self-learning coefficients and the corresponding updating time points are recorded into a log document or a database by writing a commonly used recording program. Here, various programs can be written as necessary, but as long as this function is ensured;
an alternative way is to use the C + + language program to output the relevant data records of each strip steel roll to the file name CTC _ MMDD _ year. The file name defines the rules, MMDD denotes month and date, and YEAR denotes YEAR. For example: the file for CTC _ Sep15_2019.csv represents a 9 month 15 log file in 2019. All steel grades from 11 th 9 th 2019 to 15 th 9 th 2019 are SPHC-W, and the thickness specification is as follows: 3.0mm, coiling target temperature: 580 ℃; the recorded data of the self-learning coefficients are 84 pieces, and a scatter diagram of the self-learning coefficients and the update time is shown in fig. 2.
2. Performing initial judgment on the obtained self-learning coefficient: whether the number proportion of the self-learning coefficients exceeding the set threshold reaches a given preset numerical value or more is determined by the following specific method:
obtaining a self-learning coefficient set of a model program: { Zi-84, number n:
{Zi-14.861, -15.86, -18.668, -14.124, -11.34, … …,27.828,21.473,19.058,14.778}, i having values 1,2, …, 84;
obtaining a time point set corresponding to the updating of the self-learning coefficient of the model program: { TiAnd corresponding the integer set (or natural number set) of the number of self-learning coefficients of the model program according to the time sequence: { Ni}；
{TiThe value of i is 1,2, … and 84 in sequence, wherein { 2019-Sep-1102: 50:31,2019-Sep-1102: 52:48,2019-Sep-1102: 55:04,2019-Sep-1102: 57:24, … …, 2019-Sep-1501: 10:07,2019-Sep-1501: 12:57,2019-Sep-1501: 15:57 };
{Ni}＝{1,2,3,4,5,6,……,80,81,82,83,84}；
setting a threshold value of a self-learning coefficient of a model program: lim is an element of [ P ∈ ]l·Set,Pu·Set]Where Lim is the threshold value of the self-learning coefficient of the model program, Set is the calculation result of the model program in which the self-learning coefficient of the model program participates if and only if the self-learning coefficient of the model program is 0, PlIs the percentage and the value range is [ -30%, -5% ]]，PuIs the percentage and the value range is (5 percent, 30 percent)](ii) a In this embodiment, a threshold Lim ∈ [ -30,30] of a coiling temperature self-learning coefficient is set]；
The value range of the preset value as the determination condition may be 5% to 60%, where 5% is the lower limit and 60% is the upper limit, the determination range of the number D exceeding the threshold is: 0.05 · n,0.6 · n ═ 4.2,50.4,;
then, the following judgment is made: if the number proportion of the obtained model program self-learning coefficients exceeding the threshold lim is within the judgment range D, further analyzing the model program self-learning coefficients: if the number proportion of the self-learning coefficients exceeding the threshold lim exceeds the upper limit of the judgment range D, searching the problem source according to the input quantity required by the calculation of the model program without further analysis on the self-learning coefficients of the model program;
in the present embodiment, the coiling temperature learning coefficient is set at the threshold lim: the number outside the range of [ -30,30] is 7, and within the judgment range D, the model calculation process is effective, and then the periodicity of the self-learning coefficients should be analyzed;
3. if the calculation process of the model program is not invalid, the obtained self-learning coefficients of the model program are regressed to be { N }iThe value of the self-learning coefficient is ZiEstablishing a coordinate system for the vertical coordinate, performing regression on the self-learning coefficient, and performing multiple differential calculation on the regressed function, wherein the specific process is as follows:
to { ZiCarry out 6 th order polynomial regression, as shown in fig. 3, to obtain the following original regression function:
y＝-4E-08·x6+1E-05·x5-0.0011·x4+0.0501·x3-1.1377·x2+11.232 · x-37.102; at this time, y represents a self-learning coefficient of the coiling temperature, and the domain of the independent variable x is [0,84 ]]；
The original regression function is differentiated 1 time and 2 times respectively to obtain the following differential functions:
y(1)＝-2.4E-07·x5+5E-05·x4-0.0044·x3+0.1503·x2-2.2754·x+11.232；
y(2)＝-1.2E-06·x4+2E-04·x3-0.0132·x2+0.3006·x-2.2754；
4. and periodically analyzing the calculated multiple differentials, if periodicity exists, finding out the periodic time point, and further performing further fault diagnosis on input quantity required by calculation of all model programs at the time point, wherein the specific method comprises the following steps:
in [0,84 ]]Within the interval, solve the differential equation y (1)0, a solution set containing 5 solutions is obtained: { xj}＝{9.512,22.471,49.586,58.031,78.9}；
Because the number of the solution sets is more than or equal to 2, the original regression function is considered to be a periodic function according to probability statistics knowledge, and the self-learning coefficient of the model program is further considered to have periodic fluctuation; in this case, it is not necessary to solve the quadratic differential equation y(2)＝0；
The solution set { x obtained abovejNormalizing the solutions into a standard solution set { N } according to a rule that the natural number is closest to each solution {9.512,22.471,49.586,58.031,78.9}, and recording the natural number as the natural numberj}＝{10,22,50,58,79}；
Self-learning natural number set of coefficient numbers { N) according to the model programjTime point T corresponding to updating of self-learning coefficient of model programiThe above standard solutions are collected into { N } corresponding relationsjOne-to-one correspondence is made into a standard time point solution set { T }j}＝{2019-Sep-11 04:55:55,2019-Sep-12 04:08:48,2019-Sep-13 05:52:30,2019-Sep-13 15:18:56,2019-Sep-14 15:31:22}；
Next, the standard time point solution set { T } is obtained as described abovejDetermining the value change period of the coiling temperature self-learning coefficient to be 24h, and determining the period time point, namely { T }jExtracting input quantities of the coiling temperature process control models of the corresponding five time nodes, and performing further fault diagnosis;
5. problem analysis and adjustment: through analysis, when the SPHC-W is rolled in the period from 11 days in 9 months to 15 days in 2019, 9 and 15 days, the cross influence of two input parameters exists between a coiling temperature control model used by the SPHC-W and a control model used by the SPHC-DC of the same specification steel type of the same steel family, and the input parameters comprise coiling cooling water quantity, cooling water temperature, residence time of hot rolled coils on a production line, rough rolling and finish rolling passes, rolling reduction and the like; the cross mixing of the input quantities leads the self-learning coefficient value of the coiling temperature to gradually tend to be larger when the SPHC-W is rolled, and has a divergence trend of breaking through a threshold Lim which belongs to the range of-30 and 30, so the input quantities of the SPHC-W and the SPHC-DC steel are further played, the cross influence of the two steel coils is avoided, and the control of the coiling temperature process is not abnormal in the subsequent SPHC-W production process.
Based on the same inventive concept of the previous embodiments, in other alternative embodiments, as shown in fig. 4, there is provided a plate and strip hot rolling process control apparatus including:
a first obtaining module 41, configured to obtain n values { Z ] of a self-learning coefficient of the hot rolling process control model corresponding to the preset control parameter in a preset time periodiAnd with each ZiCorresponding update time { Ti}; n is more than or equal to 2 and is a positive integer, and i sequentially takes the values of 1,2, … and n;
a validation module 42 for validating the threshold range of the self-learning coefficients, and in { Z }iThe number m of values outside the threshold range;
a judging module 43, configured to judge whether the m/n ratio is within a preset range;
if yes, for { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}，j∈[1,n]；
A second obtaining module 44 for obtaining a value according to { Z }jCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node;
the adjusting module 45 is used for analyzing the input quantity and adjusting the input quantity in the subsequent rolling process;
and a control module 46 for controlling the hot rolling process according to the adjusted input amount.
Based on the same inventive concept of the foregoing embodiments, in other alternative embodiments, there is also provided an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor implements the steps of the control method of any one of the foregoing embodiments when executing the program.
In other alternative embodiments, based on the same inventive concept as the previous embodiments, a computer-readable storage medium is further provided, on which a computer program is stored, which when executed by a processor implements the steps of the control method of any one of the previous embodiments.
Through one or more embodiments of the present invention, the present invention has the following advantageous effects or advantages:
the invention provides a method for performing hot rolling control by judging and analyzing a process control model self-learning coefficient, which determines a period point of the change of the self-learning coefficient through regression analysis, performs targeted extraction and analysis on process control model input data of the period point, can obviously reduce the analysis workload compared with the method of nondifferential extraction and analysis of all hot rolling process data, extracts the model input data of the corresponding period point by combining the analysis of the change period of the self-learning coefficient, can more quickly and accurately discover implicit rules in the data, thereby more timely and accurately adjusting the hot rolling process control, and avoids the reduction of the control precision of a strip steel product caused by the faulty operation of a rolling process control model program.
While the preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.
Claims (6)
1. A plate and strip hot rolling process control method is characterized by comprising the following steps:
the processor acquires the self-setting time period of the hot rolling process control model corresponding to the preset control parametersN values of the learning coefficient { ZiAnd with each of said ZiCorresponding update time { Ti}; n is more than or equal to 2 and is a positive integer, and i sequentially takes the values of 1,2, … and n;
the processor identifies the threshold range of the self-learning coefficients, and in the { ZiThe number m of values outside the threshold range;
the processor judges whether the ratio of m/n is within a preset range;
if yes, the processor compares the { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}，j∈[1,n](ii) a The method comprises the following steps: with the { ZiIs the y-axis, said { T }iThe sequence of integers [ N ] determined in ascending order of timei]Is the x-axis, NiSequentially taking values of 1,2, …, n; for the { ZiPerforming polynomial fitting regression analysis to obtain a corresponding regression equation, which is as follows: y ═ cpxp+cp-1xp-1+……+c1x+c0(ii) a Wherein the definition domain of x is [0, n]P is not less than 3 and is a positive integer; performing first order differentiation on the regression equation to obtain a first order differential equation, and defining the domain [0, n]Solving a solution set { x) of the first order differential equationjThe first order differential equation is specifically as follows: p.cpxp-1+(p-1)·cp-1xp-2+……+c2x+c10; judging the solution set { xjSolution x injWhether the number of (a) is more than two; if yes, confirming that the values of the self-learning coefficients have periodic distribution, and collecting the solution set { xjEach solution x injRounding to obtain integer set Xj](ii) a According to [ X ] abovej]From { N }iDetermining X in the sequencej＝NjA point of (a); then according to { NjFrom said { Z }iDetermine { Z }jH, the { Z } isjDetermining each point in the data as a period value taking point;
the processor is according to the { ZjCorresponding update time TjGet at each TjInput quantities of the hot rolling process control model on time nodes;
the processor analyzes the input quantity and adjusts the input quantity in the subsequent rolling process;
and the controller controls the hot rolling process according to the adjusted input quantity.
2. The control method of claim 1, wherein the processor determining the threshold range of the self-learning coefficients comprises:
confirming threshold value range lim epsilon [ P ] of self-learning coefficientl×Set,Pu×Set]；
Wherein, the PlAnd PuAnd the Set is an output value of a preset control parameter output by the hot rolling process control model when the self-learning coefficient is 0.
3. The control method according to claim 2, wherein P islHas a value range of [ -0.3, -0.05 [)](ii) a The P isuHas a value range of [0.05,0.3 ]]。
4. The control method according to claim 1, wherein the preset range is [0.05,0.6 ].
5. The control method of claim 1, wherein said determining said solution set { xjSolution x injWhether the number of (2) is more than two, further comprising:
if xjIs less than two, the regression equation is subjected to second order differentiation to obtain a second order differential equation, and the second order differential equation is positioned in a definition domain [0, n ]]Solving a solution set { x) of the second order differential equationkAnd the second order differential equation is concretely as follows:
p(p-1)·cpxp-2+(p-1)(p-2)·cp-1xp-3+……+c3x+c2＝0；
judging the solution set { xkSolution x inkWhether the number of (a) is more than two;
if so, the station is confirmedThe values of the self-learning coefficients are periodically distributed, and the solution set { xkEach solution x inkRounding to obtain integer set Xk]；
According to [ X ] abovek]From { N }iDetermining X in the sequencek＝NkA point of (a); then according to { NkFrom said { Z }iDetermine { Z }kH, the { Z } iskDetermining each point in the data as a period value taking point;
the processor is according to the { ZjCorresponding update time TjGet at each TjThe input quantity of the hot rolling process control model on the time node specifically comprises the following steps:
the processor is according to the { ZkCorresponding update time TkGet at each TkAn input quantity of the hot rolling process control model at a time node.
6. A hot rolling process control apparatus for a plate and strip, comprising:
a first obtaining module, configured to obtain n values { Z ] of self-learning coefficients of the hot rolling process control model corresponding to the preset control parameters in a preset time periodiAnd with each of said ZiCorresponding update time { Ti}; n is more than or equal to 2 and is a positive integer, and i sequentially takes the values of 1,2, … and n;
a validation module for validating a threshold range of the self-learning coefficients, and at the { Z }iThe number m of values outside the threshold range;
the judging module is used for judging whether the m/n ratio is within a preset range or not;
if yes, for the { ZiPerforming regression analysis to confirm the period value-taking points { Z }j}，j∈[1,n](ii) a The method comprises the following steps: with the { ZiIs the y-axis, said { T }iThe sequence of integers [ N ] determined in ascending order of timei]Is the x-axis, NiSequentially taking values of 1,2, …, n; for the { ZiPerforming polynomial fitting regression analysis to obtain corresponding regression equationThe following were used: y ═ cpxp+cp-1xp-1+……+c1x+c0(ii) a Wherein the definition domain of x is [0, n]P is not less than 3 and is a positive integer; performing first order differentiation on the regression equation to obtain a first order differential equation, and defining the domain [0, n]Solving a solution set { x) of the first order differential equationjThe first order differential equation is specifically as follows: p.cpxp-1+(p-1)·cp-1xp-2+……+c2x+c10; judging the solution set { xjSolution x injWhether the number of (a) is more than two; if yes, confirming that the values of the self-learning coefficients have periodic distribution, and collecting the solution set { xjEach solution x injRounding to obtain integer set Xj](ii) a According to [ X ] abovej]From { N }iDetermining X in the sequencej＝NjA point of (a); then according to { NjFrom said { Z }iDetermine { Z }jH, the { Z } isjDetermining each point in the data as a period value taking point;
a second obtaining module for obtaining ZjCorresponding update time TjGet at each TjInput quantities of the hot rolling process control model on time nodes;
the adjusting module is used for analyzing the input quantity and adjusting the input quantity in the subsequent rolling process;
and the control module is used for controlling the hot rolling process according to the adjusted input quantity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010123629.0A CN111266405B (en) | 2020-02-27 | 2020-02-27 | Plate and strip hot rolling process control method and control device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010123629.0A CN111266405B (en) | 2020-02-27 | 2020-02-27 | Plate and strip hot rolling process control method and control device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111266405A CN111266405A (en) | 2020-06-12 |
| CN111266405B true CN111266405B (en) | 2021-11-02 |
Family
ID=70994203
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010123629.0A Active CN111266405B (en) | 2020-02-27 | 2020-02-27 | Plate and strip hot rolling process control method and control device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111266405B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104281088A (en) * | 2014-05-27 | 2015-01-14 | 南京钢铁股份有限公司 | Second-level system progress control method and device of hot rolled strip rolling mill |
| CN104815848A (en) * | 2014-12-19 | 2015-08-05 | 中冶南方（武汉）自动化有限公司 | Roll eccentricity control method based on thickness detection signals and self-adaptive neural network |
| JP2016131975A (en) * | 2015-01-15 | 2016-07-25 | Ｊｆｅスチール株式会社 | Plate thickness control method in thick plate rolling |
| CN107045585A (en) * | 2017-05-17 | 2017-08-15 | 北京科技大学 | A kind of rolling model self study coefficient determines method |
| CN107470372A (en) * | 2017-09-05 | 2017-12-15 | 北京首钢冷轧薄板有限公司 | A kind of method, apparatus and electronic equipment for being used to control rolled piece sideslip |
| CN107794335A (en) * | 2016-09-06 | 2018-03-13 | 鞍钢股份有限公司 | Method for improving precision of automatic steelmaking dynamic model |
-
2020
- 2020-02-27 CN CN202010123629.0A patent/CN111266405B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104281088A (en) * | 2014-05-27 | 2015-01-14 | 南京钢铁股份有限公司 | Second-level system progress control method and device of hot rolled strip rolling mill |
| CN104815848A (en) * | 2014-12-19 | 2015-08-05 | 中冶南方（武汉）自动化有限公司 | Roll eccentricity control method based on thickness detection signals and self-adaptive neural network |
| JP2016131975A (en) * | 2015-01-15 | 2016-07-25 | Ｊｆｅスチール株式会社 | Plate thickness control method in thick plate rolling |
| CN107794335A (en) * | 2016-09-06 | 2018-03-13 | 鞍钢股份有限公司 | Method for improving precision of automatic steelmaking dynamic model |
| CN107045585A (en) * | 2017-05-17 | 2017-08-15 | 北京科技大学 | A kind of rolling model self study coefficient determines method |
| CN107470372A (en) * | 2017-09-05 | 2017-12-15 | 北京首钢冷轧薄板有限公司 | A kind of method, apparatus and electronic equipment for being used to control rolled piece sideslip |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111266405A (en) | 2020-06-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109992921B (en) | On-line soft measurement method and system for thermal efficiency of boiler of coal-fired power plant | |
| CN110794093B (en) | Precision compensation method for discharged caustic alkali concentration measuring device in evaporation process | |
| US20240184956A1 (en) | Prediction method of crown of steel plates and strips based on data driving and mechanism model fusion | |
| DE102021201296A1 (en) | Prediction model for predicting product quality parameter values | |
| CN108171142B (en) | A kind of causal method of key variables in determining complex industrial process | |
| Ma et al. | A novel hierarchical detection and isolation framework for quality-related multiple faults in large-scale processes | |
| CN109261724B (en) | Method for improving preset model precision in multi-variety rolling mode | |
| CN115308562A (en) | Chip testing method and related equipment | |
| CN105868164B (en) | A kind of soft-measuring modeling method based on the linear dynamic system model for having supervision | |
| CN113102516B (en) | Hot continuous rolling strip steel head width prediction method integrating rolling mechanism and deep learning | |
| CN114077876B (en) | Strip steel hot continuous rolling multi-mode process monitoring method and device | |
| CN112287550B (en) | Strip steel head thickness difference process parameter optimization method based on principal component analysis controller | |
| CN115993856B (en) | Multi-region environmental condition management and control method and system for factory building | |
| Dong et al. | Quality monitoring and root cause diagnosis for industrial processes based on Lasso-SAE-CCA | |
| CN114580151A (en) | Water demand prediction method based on gray linear regression-Markov chain model | |
| CN106845825A (en) | It is a kind of to be traced to the source and control method based on the cold rolling of strip steel quality problems for improving PCA | |
| CN111266405B (en) | Plate and strip hot rolling process control method and control device | |
| CN117610195A (en) | Strip steel expansion prediction method, device, medium and electronic equipment | |
| CN103389360A (en) | Probabilistic principal component regression model-based method for soft sensing of butane content of debutanizer | |
| CN106647274B (en) | Operating condition stable state method of discrimination in a kind of continuous flow procedure | |
| CN103279030A (en) | Bayesian framework-based dynamic soft measurement modeling method and device | |
| CN117290752A (en) | Prediction method for coal fines fineness of coal mill | |
| CN113780852B (en) | Diagnosis method for quality defects in plate and strip rolling process | |
| CN115619280A (en) | Process quality prediction method based on process standard map and CNN-GRU network model | |
| CN115034443A (en) | Boiler performance prediction method based on working condition division and neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |