CN115081798A - PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method - Google Patents
PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method Download PDFInfo
- Publication number
- CN115081798A CN115081798A CN202210495337.9A CN202210495337A CN115081798A CN 115081798 A CN115081798 A CN 115081798A CN 202210495337 A CN202210495337 A CN 202210495337A CN 115081798 A CN115081798 A CN 115081798A
- Authority
- CN
- China
- Prior art keywords
- variable
- rack
- data
- monitoring
- quality
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 79
- 238000012544 monitoring process Methods 0.000 title claims abstract description 72
- 238000005096 rolling process Methods 0.000 title claims abstract description 68
- 238000003745 diagnosis Methods 0.000 title claims abstract description 43
- 238000004519 manufacturing process Methods 0.000 claims abstract description 34
- 230000002159 abnormal effect Effects 0.000 claims abstract description 25
- 238000010586 diagram Methods 0.000 claims abstract description 16
- 238000005098 hot rolling Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 36
- 239000013598 vector Substances 0.000 claims description 19
- 230000002068 genetic effect Effects 0.000 claims description 16
- 238000012360 testing method Methods 0.000 claims description 16
- 230000001419 dependent effect Effects 0.000 claims description 10
- 238000009826 distribution Methods 0.000 claims description 8
- 230000001364 causal effect Effects 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 4
- 150000001875 compounds Chemical class 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims description 2
- 238000003908 quality control method Methods 0.000 abstract description 5
- 230000007547 defect Effects 0.000 abstract description 2
- 238000007689 inspection Methods 0.000 abstract description 2
- 230000007774 longterm Effects 0.000 abstract description 2
- 230000005856 abnormality Effects 0.000 description 4
- 241000277275 Oncorhynchus mykiss Species 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 3
- 238000007730 finishing process Methods 0.000 description 3
- 229910000831 Steel Inorganic materials 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 208000013166 Abnormality of the head Diseases 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 229910052731 fluorine Inorganic materials 0.000 description 1
- 125000001153 fluoro group Chemical group F* 0.000 description 1
- 239000013072 incoming material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000013000 roll bending Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000012109 statistical procedure Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Operations Research (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- Entrepreneurship & Innovation (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Manufacturing & Machinery (AREA)
- Life Sciences & Earth Sciences (AREA)
- Algebra (AREA)
- Game Theory and Decision Science (AREA)
- Databases & Information Systems (AREA)
- General Factory Administration (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- Evolutionary Biology (AREA)
Abstract
The invention discloses a strip hot finish rolling process sub-frame combined monitoring and diagnosis method based on a PLS algorithm, relating to the technical field of quality control in a hot rolling strip finish rolling processing process, and comprising the following steps: according to the hot rolled plate strip finish rolling production stage, selecting the process variables related to the thickness and quality in the finish rolling process of each rack, constructing production data subblock samples of each rack, and carrying out standardization processing on original data; acquiring the accumulated contribution rate of each rack sub-block by using PRESS inspection, and determining the number of the principal elements; constructing a data subblock model of each rack by using a PLS algorithm, calculating subblock statistics, and acquiring monitoring data state information of each rack in the production process; and constructing a sub-frame data sub-block variable contribution diagram, and performing quality-related variable anomaly reasoning to obtain an anomaly root variable. The invention improves the defect of root source diagnosis error caused by quality inheritance in the multi-stand rolling process, and realizes the aims of accurate source tracing of abnormal quality in the multi-stand rolling process and long-term stable production in the rolling process.
Description
Technical Field
The invention relates to the technical field of quality control in a hot rolling strip finish rolling process, in particular to a strip hot finish rolling process sub-rack combined monitoring and diagnosis method based on a PLS algorithm, wherein the quality genetic characteristic of the hot finish rolling multi-rack rolling process is considered.
Background
The monitoring and diagnosis of the quality-related fault variables in the multi-frame hot finish rolling process is an important component of intelligent control and steady-state operation of the multi-frame hot finish rolling process, and field workers can master the operation state information and the quality control condition of equipment in time by monitoring the fluctuation condition and the quality abnormal signal of the process variables. Through monitoring and diagnosis, misjudgment caused by manual detection can be improved, the utilization rate of industrial fields on production data can be effectively improved, and deep features of the data are mined.
The monitoring and diagnosis of the quality related variables are an important guarantee in the production process of the hot rolled strip, and the diagnosis speed and the identification precision are two important indexes in the diagnosis process. In the monitoring and diagnosis of the quality related variables in the plate strip production process by the traditional method, the method for solving the problems of strong coupling and nonlinearity among data multivariable and the related method aiming at multivariate quality diagnosis are widely applied. However, since the processes in the finish rolling stage are complex and changeable and the frames are matched with each other, most of the monitoring and diagnosing methods ignore the composition module of a single process for a certain variable, so that the genetic characteristics of each processing module of the quality in the single process are also ignored, and the misjudgment of the diagnosis process on the source variable affecting the quality abnormity is caused. The quality control can not be carried out from the root, so that the phenomenon that the production can not be stably carried out for a long time often occurs in field production.
Disclosure of Invention
In view of the above, the invention provides a partial-rack combined monitoring and diagnosis method for a plate strip hot finish rolling process based on a PLS algorithm, so as to realize root cause diagnosis of quality-related faults in the existing process industrial process and reduce misjudgment of root cause variables influencing quality abnormality in the diagnosis process.
Therefore, the invention provides the following technical scheme:
the invention provides a partial-rack combined monitoring and diagnosis method for a plate strip hot finish rolling process based on a PLS algorithm, which comprises the following steps of:
according to the hot rolled plate strip finish rolling production process flow, production data related to thickness quality are collected, the collected data are subjected to standardized processing, and sub-block division is carried out on the data according to each sub-frame;
determining the accumulated contribution rate of each rack variable according to the sum of the square of the forecast errors and a cross test model to obtain the number of principal elements of the monitoring variable of each data subblock;
establishing each rack data subblock model by using a partial least square method PLS based on the principal element number of each data subblock monitoring variable, and extracting a variable principal component vector;
calculating monitoring statistics of each data subblock according to the data subblock model, and determining the control limit of each rack monitoring process;
obtaining the stable condition of monitoring the process variable of each rack according to the monitoring statistic of each data sub-block and the monitoring process control limit of each rack, and obtaining the state information whether the production process is stable;
calculating the contribution value of each rack process variable according to the projection size of each rack data submodule to the variable principal component, obtaining a contribution diagram of each variable to the quality-related fault, and determining the cause of the rack causing the quality abnormal state by combining the contribution diagram and the state information of whether the production process is stable or not;
and constructing a causal topological structure of the continuous rolling process based on the Glange causal relationship idea, establishing a genetic topological structure of each rack sub-monitoring module, and reasoning a fault source according to the genetic characteristics of the quality among the racks and the maximum contribution variable of the contribution diagram of each rack.
Further, determining the accumulated contribution rate of each rack variable according to the sum of squared prediction errors and a cross-checking model to obtain the number of principal elements of each data subblock monitoring variable, wherein the method comprises the following steps:
the number A of main elements of each rack is given i Modeling and checking to obtain a prediction error; divide the sample data into two groups, n 2 One sample for regression modeling, n 1 The samples are used for testing and inspection;
according to the sampling test, the total prediction error is obtained as follows:
in the formula y i Is n 1 The ith sample point of the test sample points;
predicting a value for the ith sample point of the test sample;
calculating n 1 Prediction error S of a sample point pair to A-1 components A-1 Then, the cross validity of the dependent variable at the a-th principal element is:
when in use
If the model is not accurate, the added A-th principal component is discarded.
Further, establishing a model of each rack data subblock by using a Partial Least Squares (PLS) based on the principal element number of the monitoring variable of each data subblock, wherein the model comprises the following steps:
taking the process variables of each frame of the hot finish rolling as independent variables as an input matrix X of the model, taking the thickness quality of a finish rolling outlet as a dependent variable as an output matrix Y of the model, and obtaining an external model:
wherein T is a process variable score matrix, and P is a process variable load vector; u is a quality variable score matrix; q is a mass variable load vector; e is a process variable residual error matrix; f is a quality variable residual error matrix; a is the PLS model principal element number; t is t j ,u j Principal components extracted in the process variable and the quality variable, respectively.
Further, calculating monitoring statistics of each data sub-block according to the data sub-block model, including:
calculating score vector t of variable data in ith frame production process new,i And residual error
t new,i ＝R T x new,i ；
Wherein x is new,i For a new sample of the strip thickness quality related variable data needing to be monitored, R is a projection direction matrix, and R is W (P) T W) -1 ；P T R＝R T P＝W T W＝I A ；R T Is the transpose of the projection direction matrix, and I is the identity matrix; i is A Is a unit matrix corresponding to the number of the pivot elements; w is a weight matrix of the input matrix;
calculating T according to the score vector and the residual error 2 And Q statistics:
wherein ^ is the covariance of the scoring matrix in the training model,
is a transpose of the score vector,
is the transpose of the new sample.
Further, the calculating the control limit of each rack comprises:
statistic T 2 And Q, respectively obeying F distribution and standard normal distribution, and calculating the control limits of the two variables
And
wherein the content of the first and second substances,
as a statistic T 2 The control limit of (a) is set,
is the control limit of the statistic Q, n is the number of sample points, and alpha is the control limit confidence level;
s is the variance of the Q statistic, and mu is the mean of the Q statistic;
F A,n-A；α and
is a number ofAccording to the distribution mode.
Further, according to the projection size of each rack data submodule to the variable principal component, calculating the contribution value of each rack process variable to obtain the contribution graph of each variable to the quality-related fault, comprising:
calculating the contribution value of the jth variable of the overrun sample point i to the square prediction error;
in the formula (I), the compound is shown in the specification,
is the total contribution value; x is a radical of a fluorine atom ij The detection value of the jth variable in the ith sample point is obtained;
the predicted value of the jth variable in the ith sample point is obtained;
adding the expected factors:
and forming a contribution graph based on the contribution value of each variable of the overrun sample point to the square prediction error after the expected factor is added.
The invention has the following beneficial effects:
the invention provides a partial-frame joint monitoring and diagnosing method for a plate strip finish rolling process based on a PLS algorithm, which comprises the following steps: the method comprises the steps of obtaining industrial data of a strip rolling process, dividing data sub-blocks of each rack in the rolling process according to a process system of each rack in a finish rolling process, carrying out modeling analysis on the data sub-blocks according to PRESS cross test and PLS algorithm, constructing model statistics, and obtaining quality related variable monitoring statistical information and monitoring results of the sub-blocks. And analyzing the contribution degree of each sub-block variable to the quality abnormity according to a contribution graph method to obtain main variables causing the quality abnormity. And (3) according to the genetic characteristic of the quality variable among the multiple racks, combining the diagnosis result of each subblock, reasoning the main reason of the abnormal variable of each rack to obtain the root cause abnormal variable and the passive abnormal variable caused by the genetic characteristic, and reasoning according to each subblock to realize the root cause diagnosis of the quality-related fault of the production process of the multiple racks. The technical scheme provided by the invention overcomes the defect of root source diagnosis error caused by quality inheritance in the multi-stand rolling process, and realizes the aims of accurate source tracing of abnormal quality in the multi-stand rolling process and long-term stable production in the rolling process.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic diagram of a PLS based single rack quality monitoring and diagnostics framework in an embodiment of the present invention;
FIG. 2 is a schematic diagram of a PLS algorithm according to an embodiment of the present invention;
FIG. 3 is a block diagram of a finishing mill group for hot rolled strip according to an embodiment of the present invention;
FIG. 4 is a statistical chart of the longitudinal thickness out-of-tolerance of a hot-rolled strip in an embodiment of the invention;
FIG. 5 is a statistical chart of the F1 rack T2 statistics in an embodiment of the present invention;
FIG. 6 is a F1 chassis SPE statistics statistical chart in an embodiment of the invention;
FIG. 7 is a statistical chart of the F2 rack T2 statistics in an embodiment of the present invention;
FIG. 8 is a F2 chassis SPE statistics statistical chart in an embodiment of the invention;
FIG. 9 is a statistical chart of the F3 rack T2 statistics in an embodiment of the present invention;
FIG. 10 is a F3 chassis SPE statistics statistical chart in an embodiment of the invention;
FIG. 11 is a statistical chart of the F4 rack T2 statistics in an embodiment of the present invention;
FIG. 12 is a statistics plot for the F4 chassis SPE statistics in an embodiment of the present invention;
FIG. 13 is a statistical chart of the F5 rack T2 statistics in an embodiment of the present invention;
FIG. 14 is a F5 chassis SPE statistics statistical chart in an embodiment of the invention;
FIG. 15 is a statistical chart of the F6 rack T2 statistics in an embodiment of the present invention;
FIG. 16 is a F6 chassis SPE statistics statistical chart in an embodiment of the invention;
FIG. 17 is a statistical chart of the F7 rack T2 statistics in an embodiment of the present invention;
FIG. 18 is a F7 chassis SPE statistics statistical chart in an embodiment of the invention;
FIG. 19 is a statistical chart of the F1 rack variable contribution values in an embodiment of the present invention;
FIG. 20 is a statistical chart of the contribution of the F2 rack variable in an embodiment of the present invention;
FIG. 21 is a statistical chart of the contribution of the F3 rack variable according to an embodiment of the present invention;
FIG. 22 is a statistical chart of the contribution of the F4 rack variable in an embodiment of the present invention;
FIG. 23 is a statistical chart of the contribution of the F5 rack variable in an embodiment of the present invention;
FIG. 24 is a statistical chart of the contribution of the F6 rack variable in an embodiment of the present invention;
FIG. 25 is a statistical chart of the contribution of the F7 rack variable in an embodiment of the present invention;
FIG. 26 is a diagram illustrating an embodiment of the present invention for overall monitoring T 2 A statistics histogram;
FIG. 27 is a statistical chart of the overall monitoring SPE statistics in an embodiment of the present invention;
FIG. 28 is a statistical chart of the total contribution of quality related variables in an embodiment of the present invention;
FIG. 29 is a genetic topology map of quality-related anomaly variable screening in an embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The strip rolling process is a multi-frame combined rolling process, data transmission exists among frames, information flow and mass flow transmission exists in each pass rolling process, the monitoring and diagnosis of the whole rolling process is mostly realized in the quality control and diagnosis stage of the rolling production process, the information flow and the mass flow of the rolling process are ignored and present a transmission state among the frames, abnormal fluctuation of variables is transmitted in the rolling process and amplification of abnormal signals possibly generated, so that the original abnormal variables are ignored and misjudged in the diagnosis process.
Aiming at the problems, in order to realize the diagnosis of the quality related abnormal root of the multi-frame hot finishing process and consider the genetic characteristics of variables among the processes, the invention provides a Partial-frame combined monitoring and diagnosis method of the plate strip hot finishing process based on a PLS algorithm, which is a novel multivariate statistical data analysis method by utilizing PLS (Partial Least Squares), and not only can overcome the problem of collinearity, but also can emphasize the explanation and prediction functions of independent variables on dependent variables when selecting the characteristic vectors, remove the influence on regression noise and ensure that the model contains the Least variable quantity) to model data subblocks formed by the thickness related variables of each frame plate strip in the hot finishing stage by projecting the high-dimensional data space of the independent variables and the dependent variables to the corresponding low-dimensional space to respectively obtain mutually orthogonal characteristic vectors of the independent variables and the dependent variables, and obtaining the fluctuation condition of the quality-related monitoring signal through monitoring statistics, building a variable contribution histogram of each subblock by adopting a contribution graph, obtaining a key variable generating quality abnormal fluctuation, and finally identifying a quality-related fault source by combining a sub-rack combined monitoring strategy, so that the quality-related abnormal source diagnosis of the multi-rack hot finish rolling process is realized, and the diagnosis accuracy is improved.
As shown in fig. 1, a partial rack joint monitoring and diagnosis method for a plate strip hot finishing process based on a PLS algorithm provided in an embodiment of the present invention includes:
c1 is a data layer, according to the production stage of the finish rolling of the hot rolled plate strip, process variables related to the thickness and quality in the rolling process of each stand are selected, production data subblock samples of each stand are constructed, and the original data are subjected to standardization processing in the initial stage of data;
in this example, a flow chart of a hot-rolled strip finish rolling production stage is shown in fig. 3, the rolling stage is a seven-stand four-roll finish rolling mill, strip temperature measuring devices are respectively arranged at a finish rolling inlet and a finish rolling outlet, and an inlet temperature is used as a finish rolling inlet temperature variable and an outlet temperature of a final stand is used as a finish rolling temperature variable. And extracting relevant data from data storage by adopting ibaAnalyzer offline software in the actual production process, and dividing sub-blocks according to each rack. The actually measured thickness data of the strip in the rolling process is shown in figure 4, so that the strip is known to have head thickness difference exceeding and the strip steel head thickness is thicker.
In this embodiment, the data in the 1580 hot rolling production line finish rolling stage is used as a training set and a test set, the data in the table is divided into rack data subblocks according to the process variables of each rack, and the raw data is standardized by using the mean value and standard deviation of the data to eliminate the influence of the dimension on the monitoring and diagnosis process result, wherein the data variable names are shown in table 1:
TABLE 1
In this embodiment, 44 process variables are collected, including 43 process variables and 1 quality variable.
C2 is a feature extraction layer, verifies the cumulative contribution rate of each rack by using a PRESS (Predicted Error Sum of Squares, Sum of squared prediction errors) cross-check model, and determines the number of principal elements capable of interpreting 85% of feature information in the original data according to the cumulative contribution rate.
In specific implementation, the step of determining the number of pivot elements includes:
s201, giving the number A of main elements of each rack i (i-1 … 7), modeling and checking to obtain a prediction error; divide the sample data into two groups, n 2 One sample for regression modeling, n 1 One sample was used for test.
S202, according to sampling test, obtaining a total prediction error as follows:
in the formula, y i Is n 1 The ith sample point in the test sample points;
and predicting the value of the ith sample point of the test sample.
S203, calculating the prediction error S when the number of the pivot elements is A-1 by the method A-1 And the cross validity of the A-th principal element of the dependent variable is as follows:
s204, when
If the model is not accurate, the added A-th principal component is discarded.
The method needs to give the number of the principal elements of each rack, then checks whether the number of the principal elements is proper one by one, and the number A of the principal elements when the loop meets the checking condition is the number of the principal elements adopted by the final monitoring modeling.
According to steps S201 to S204, the number of the selected pivot elements in this embodiment is shown in table 2:
TABLE 2
C3 is a statistical monitoring layer, a data subblock model of each rack is constructed by using a partial least square method, subblock statistics is calculated, and monitoring data state information in the production process of each rack is obtained;
s301, establishing a data subblock model of each rack by using a partial least square method; as shown in fig. 2.
Taking the process variable of each rack as an independent variable as an input matrix X of the model, taking the thickness quality of a finish rolling outlet as a dependent variable as an output matrix Y of the model, and obtaining an external model;
forming a PLS model based on the variables:
wherein T is a process variable score matrix, and P is a process variable load vector; u is a quality variable score matrix; q is a mass variable load vector; e is a process variable residual error matrix; f is a quality variable residual error matrix; a is the PLS model principal element number; t is t j ,u j Principal component vectors extracted in the process variable and the quality variable, respectively. PLS takes into account not only the data information that maximizes the abstraction of the argument space, but also the argumentsThe interpretive role of pivot on dependent variable changes.
S302, calculating monitoring statistics of each data sub-block according to the score vectors and the residual errors;
the new input sample is x new Training model independent variable projection direction matrix R ═ R 1 ,r 2 ,…r A ]. T-XR is satisfied, and thus a principal component matrix is obtained. The projection direction matrix is then:
R＝W(P T W) -1 (4)
P T R＝R T P＝W T W＝I A (5)
calculating score vectors and residuals:
t new ＝R T x new (6)
calculating T 2 And Q statistics:
in the formula, a score matrix covariance in the training model.
S303, calculating the control limit of each rack monitoring process;
statistical procedure T 2 And Q respectively obeying F distribution and standard normal distribution, and calculating the control limits of two variables:
s304, obtaining the stable condition of monitoring the process variable of each rack according to the monitoring statistics of each data sub-block, and obtaining the state information whether the production process is stable or not.
When the system running state is stable and no fault occurs, the relation between the statistic and the control limit is as follows:
and when the statistic and the control limit do not meet the relationship, the abnormity of the production process is shown.
Fig. 5-18 show the monitoring statistics of the sub-frame partial least squares method in this embodiment, which can result in that most sample points of the strip are below the control limit during the rolling process, and the strip head has different degrees of overrun. Therefore, the abnormal overrun of the head of the strip is gradually reduced along with the rolling process, and the quality overrun of the head of the strip is corrected to a certain extent in the rolling process. Therefore, the thickness of the head of the primary positioning plate belt is out of tolerance in the monitoring process and is consistent with the actual measurement result.
C4 is a diagnosis inference layer, firstly, calculating the contribution value of each rack process variable according to the projection size of each rack data submodule to the variable principal component, obtaining the contribution diagram of each variable to the quality-related fault, and determining the cause of the rack causing the quality abnormal state by combining the contribution diagram and the state information of whether the production process is stable; and then constructing a causal topological structure of the continuous rolling process based on the Glangel causal relationship idea, establishing a genetic topological structure of each rack sub-monitoring module, and reasoning a fault source according to the genetic characteristics of the quality among the racks and the maximum contribution variable of the contribution diagram of each rack.
The contribution graph is composed of the contribution value of each variable of the overrun sample point to the square prediction error. The calculation formula of the contribution value of the jth variable of the overrun sample point i to the square prediction error is as follows:
in the formula (I), the compound is shown in the specification,
is the total contribution value; x is the number of ij A detection value of a jth variable in an ith sample point;
is the predicted value of the jth variable in the ith sample point.
Adding expected factors to make the expected factors more consistent with the diagnosis requirements;
according to the contribution diagram, the larger the contribution value is, the more likely the corresponding variable will cause abnormal fluctuation of the quality-related variable, so that the root cause of the quality-related variable abnormality can be found.
Fig. 19 to 25 are histograms of the contribution values to the variables in the present embodiment, and it can be seen from the contribution graphs that the F1 stand variable 1 (rolling speed), the F2 stand variable 2 (front tension), the F3, F4, and F5 stand variable 4 (entrance thickness), the F6 stand variable 1 (rolling speed), and the F7 stand variable 6 (bending force) are important influencing variables in each sub-block.
And (3) combining the diagnosis results of the subblocks, reasoning the main reasons of the abnormal variables of the racks to obtain the root cause abnormal variable and the passive abnormal variable caused by the genetic characteristic, and reasoning according to the subblocks to realize the root cause diagnosis of the quality-related faults of the multi-rack process production process.
Fig. 26 to 28 are diagrams for overall monitoring and contribution to the rolling stage, and it can be seen from the monitoring that the statistical monitoring of the quality-related variables in this embodiment is mostly below the control limit, and F6 (roll bending force) becomes a main factor of head out-of-tolerance, however, the abnormal out-of-limit of the head thickness between the F1 to F7 stands is caused by the head abnormality of the finish rolling incoming material, and therefore, the diagnosis result of the overall monitoring and diagnosis process obviously ignores the genetic characteristic of the quality of the multi-stand rolling process.
In the above embodiment, the monitoring models of the sub-blocks are constructed, and the monitoring statistics of the production process are calculated as shown in fig. 4 to 18, so that the head of the strip steel in the finish rolling process is out of limit. And (3) constructing a contribution graph of each variable to the abnormal strip steel head, wherein as shown in FIGS. 19-25, the inlet temperature of the F1 stand has the largest contribution value to the abnormal strip steel head, and the inlet thickness variable or the roll gap value of the F2-F7 stand has larger contribution. Due to the fact that the quality of the continuous rolling process has genetic characteristics, analysis of causal topological relations shows that the thickness over-tolerance of the head of the hot rolled strip is caused by low rolling inlet temperature, and the thickness abnormality of the head is transmitted in the rolling process, so that the thickness of the subsequent rack inlet or the roll gap value is detected abnormally, and the main reason influencing the thickness over-tolerance of the head of the strip is the temperature abnormality of the F1 rack inlet, as shown in fig. 29.
The monitoring and diagnosis of each machine frame sub-block are integrated to know that the head out-of-tolerance of the strip steel does not cause the abnormal fluctuation of the thickness related variable of each machine frame in the rolling process, and the sample points of the rear machine frame in the SPE statistics which is irrelevant to the quality are out-of-tolerance, so that the abnormal fluctuation serving as a potential variable is also considered although the final strip quality is not directly influenced. Therefore, the technical scheme provided by the embodiment realizes the quality-related variable monitoring and diagnosis target considering the quality genetic characteristics in the multi-stand hot rolling process.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (6)
1. A strip hot finish rolling process sub-frame joint monitoring and diagnosis method based on a PLS algorithm is characterized by comprising the following steps:
according to the hot rolled plate strip finish rolling production process flow, production data related to thickness quality are collected, the collected data are subjected to standardized processing, and sub-block division is carried out on the data according to each sub-frame;
determining the accumulated contribution rate of each rack variable according to the sum of the square of the forecast errors and a cross test model to obtain the number of principal elements of the monitoring variable of each data subblock;
establishing each rack data subblock model by using a partial least square method PLS based on the principal element number of each data subblock monitoring variable, and extracting a variable principal component vector;
calculating monitoring statistics of each data subblock according to the data subblock model, and determining the control limit of each rack monitoring process;
obtaining the stable condition of monitoring the process variable of each rack according to the monitoring statistic of each data sub-block and the monitoring process control limit of each rack, and obtaining the state information whether the production process is stable;
calculating the contribution value of each rack process variable according to the projection size of each rack data submodule to the variable principal component, obtaining a contribution diagram of each variable to the quality-related fault, and determining the cause of the rack causing the quality abnormal state by combining the contribution diagram and the state information of whether the production process is stable or not;
and constructing a causal topological structure of the continuous rolling process based on the Glange causal relationship idea, establishing a genetic topological structure of each rack sub-monitoring module, and reasoning a fault source according to the genetic characteristics of the quality among the racks and the maximum contribution variable of the contribution diagram of each rack.
2. The PLS algorithm-based plate strip finish hot rolling process sub-rack joint monitoring and diagnosis method of claim 1, wherein the determination of the accumulated contribution rate of each rack variable according to the forecast error square sum and cross-check model to obtain the number of principal elements of each data sub-block monitoring variable comprises:
given the number A of the principal elements of each rack i Modeling and checking to obtain a prediction error; divide the sample data into two groups, n 2 One sample for regression modeling, n 1 The samples are used for testing and checking;
according to the sampling test, the total prediction error is obtained as follows:
in the formula y i Is n 1 The ith sample point of the test sample points;
predicting a value for the ith sample point of the test sample;
calculating n 1 Prediction error S of a sample point pair A-1 components A-1 Then, the cross validity of the dependent variable at the a-th principal element is:
when in use
If the model is not accurate, the added A-th principal component is discarded.
3. The partial-rack combined monitoring and diagnosis method for the plate strip finish hot rolling process based on the PLS algorithm as claimed in claim 1, wherein the partial least square PLS is used for establishing each rack data subblock model based on the number of the principal elements of each data subblock monitoring variable, and the method comprises the following steps:
taking the process variables of each frame of the hot finish rolling as independent variables as an input matrix X of the model, taking the thickness quality of a finish rolling outlet as a dependent variable as an output matrix Y of the model, and obtaining an external model:
wherein T is a process variable score matrix, and P is a process variable load vector; u is a quality variable score matrix; q is a mass variable load vector; e is a process variable residual error matrix; f is a quality variable residual error matrix; a is the PLS model principal element number; t is t j ，u j Principal components extracted in the process variable and the quality variable, respectively.
4. The PLS algorithm-based strip finishing hot rolling process sub-rack joint monitoring and diagnosis method of claim 3, wherein calculating the monitoring statistics of each data sub-block according to the data sub-block model comprises:
calculating score vector t of variable data in the ith frame production process new，i And residual error
t new，i ＝R T x new，i ；
Wherein x is new，i For a new sample of plate strip thickness quality related variable data needing to be monitored, R is a projection direction matrix, and R is W (P) T W) -1 ；P T R＝R T P＝W T W＝I A ；R T Is the transpose of the projection direction matrix, and I is the identity matrix; i is A Is a unit matrix corresponding to the number of the pivot elements; w is a weight matrix of the input matrix;
calculating T according to the score vector and the residual error 2 And Q statistics:
wherein ^ is the covariance of the scoring matrix in the training model,
is a transpose of the score vector,
is the transpose of the new sample.
5. The PLS algorithm-based strip finishing hot rolling process sub-rack joint monitoring and diagnosis method of claim 4, wherein the calculating the control limit of each rack comprises:
statistic T 2 And Q, respectively obeying F distribution and standard normal distribution, and calculating the control limits of the two variables
And
wherein the content of the first and second substances,
as a statistic T 2 The control limit of (a) is set,
is the control limit of the statistic Q, n is the number of sample points, and alpha is the control limit confidence level;
s is the variance of the Q statistic, and mu is the mean of the Q statistic;
F A，n-A；α and
is the distribution mode of the data.
6. The PLS algorithm-based plate strip finish hot rolling process sub-rack combined monitoring and diagnosis method of claim 1, wherein the method comprises the steps of calculating the contribution value of each rack process variable according to the projection size of each rack data submodule to the variable principal component, and obtaining the contribution graph of each variable to the quality-related fault, and comprises the following steps:
calculating the contribution value of the jth variable of the overrun sample point i to the square prediction error;
in the formula (I), the compound is shown in the specification,
is the total contribution value; x is the number of ij The detection value of the jth variable in the ith sample point is obtained;
the predicted value of the jth variable in the ith sample point is obtained;
adding the expected factors:
and forming a contribution graph based on the contribution value of each variable of the overrun sample point to the square prediction error after the expected factor is added.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210495337.9A CN115081798A (en) | 2022-05-07 | 2022-05-07 | PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210495337.9A CN115081798A (en) | 2022-05-07 | 2022-05-07 | PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115081798A true CN115081798A (en) | 2022-09-20 |
Family
ID=83247250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210495337.9A Pending CN115081798A (en) | 2022-05-07 | 2022-05-07 | PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115081798A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117831659A (en) * | 2024-03-04 | 2024-04-05 | 山东钢铁股份有限公司 | Method and device for online detection of quality of wide and thick plates, electronic equipment and storage medium |
-
2022
- 2022-05-07 CN CN202210495337.9A patent/CN115081798A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117831659A (en) * | 2024-03-04 | 2024-04-05 | 山东钢铁股份有限公司 | Method and device for online detection of quality of wide and thick plates, electronic equipment and storage medium |
| CN117831659B (en) * | 2024-03-04 | 2024-05-03 | 山东钢铁股份有限公司 | Method and device for online detection of quality of wide and thick plates, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1416348B1 (en) | Methodology for temporal fault event isolation and identification | |
| US8285513B2 (en) | Method and system of using inferential measurements for abnormal event detection in continuous industrial processes | |
| CN108803520B (en) | Dynamic process monitoring method based on variable nonlinear autocorrelation rejection | |
| EP3764184A1 (en) | Abnormality determination assistance device | |
| CN105955219A (en) | Distributed dynamic process fault detection method based on mutual information | |
| CN109675935A (en) | A kind of IPCA operation of rolling on-line fault diagnosis method becoming control limit | |
| CN111368428B (en) | Sensor precision degradation fault detection method based on monitoring second-order statistics | |
| CN113468479B (en) | Cold continuous rolling industrial process monitoring and abnormality detection method based on data driving | |
| CN108508865A (en) | A kind of fault detection method based on distributing OSC-PLS regression models | |
| CN112000081B (en) | Fault monitoring method and system based on multi-block information extraction and Mahalanobis distance | |
| CN111324110A (en) | Fermentation process fault monitoring method based on multiple shrinkage automatic encoders | |
| CN108830006B (en) | Linear-nonlinear industrial process fault detection method based on linear evaluation factor | |
| CN115081798A (en) | PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method | |
| CN114896735A (en) | Partial least square improved hot-rolled strip steel head wave defect reason identification method | |
| CN114611067A (en) | Chemical process slow-change fault detection method based on typical variable dissimilarity analysis | |
| CN116305733A (en) | Quality-related fault detection method based on global and local feature extraction | |
| CN113376562B (en) | CVT (continuously variable transmission) verification method, device and medium based on rolling time window-FCM (fuzzy c-means) clustering | |
| CN114200914A (en) | MW-OCCA-based quality-related early fault detection method | |
| CN111966966B (en) | Method and system for analyzing feasible domain of sensor measurement error model parameters | |
| CN114970311A (en) | Method for establishing remote module life prediction model and life prediction method | |
| CN113050606A (en) | Quality-related fault detection method for complex non-stationary process | |
| CN108844662B (en) | Method for evaluating state of electrical cabinet of numerical control machine tool | |
| CN113283157A (en) | System, method, terminal and medium for predicting life cycle of intelligent stamping press part | |
| CN113467434B (en) | CCA fermentation process KPI related fault monitoring method based on double-layer k nearest neighbor standardization | |
| CN110928263B (en) | Fault detection method and system for complex process considering dynamic relationship in advance |
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 |