CN110705785A - Method and device for monitoring thermal state of crystallizer of continuous casting machine - Google Patents
Method and device for monitoring thermal state of crystallizer of continuous casting machine Download PDFInfo
- Publication number
- CN110705785A CN110705785A CN201910933908.0A CN201910933908A CN110705785A CN 110705785 A CN110705785 A CN 110705785A CN 201910933908 A CN201910933908 A CN 201910933908A CN 110705785 A CN110705785 A CN 110705785A
- Authority
- CN
- China
- Prior art keywords
- sample data
- value
- determining
- crystallizer
- current
- 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 90
- 238000009749 continuous casting Methods 0.000 title claims abstract description 61
- 238000012544 monitoring process Methods 0.000 title claims abstract description 20
- 230000002159 abnormal effect Effects 0.000 claims abstract description 68
- 238000004519 manufacturing process Methods 0.000 claims abstract description 55
- 239000011159 matrix material Substances 0.000 claims description 71
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 66
- 238000012545 processing Methods 0.000 claims description 30
- 239000000498 cooling water Substances 0.000 claims description 22
- 238000005266 casting Methods 0.000 claims description 17
- 230000001186 cumulative effect Effects 0.000 claims description 17
- 239000007788 liquid Substances 0.000 claims description 14
- 229910000831 Steel Inorganic materials 0.000 claims description 13
- 239000010959 steel Substances 0.000 claims description 13
- 238000012216 screening Methods 0.000 claims description 11
- 239000013598 vector Substances 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 238000009826 distribution Methods 0.000 claims description 5
- 238000012546 transfer Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000005856 abnormality Effects 0.000 description 4
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 2
- 229910052802 copper Inorganic materials 0.000 description 2
- 239000010949 copper Substances 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 239000013643 reference control Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000009628 steelmaking 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B22—CASTING; POWDER METALLURGY
- B22D—CASTING OF METALS; CASTING OF OTHER SUBSTANCES BY THE SAME PROCESSES OR DEVICES
- B22D11/00—Continuous casting of metals, i.e. casting in indefinite lengths
- B22D11/16—Controlling or regulating processes or operations
-
- 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
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C3/00—Registering or indicating the condition or the working of machines or other apparatus, other than vehicles
- G07C3/005—Registering or indicating the condition or the working of machines or other apparatus, other than vehicles during manufacturing process
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Manufacturing & Machinery (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Health & Medical Sciences (AREA)
- Mechanical Engineering (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The embodiment of the invention provides a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine, wherein the method comprises the following steps: acquiring multiple groups of historical sample data in the continuous casting production process; determining a principal component model of historical sample data; determining a control limit of a first square prediction error statistic of the pivot model based on historical sample data; determining a second mean prediction error statistic of the pivot model based on each set of online data; judging whether the second square prediction error statistic value at each moment is larger than the control limit value, if so, acquiring the current group of online data corresponding to the second square prediction error statistic value at the current moment, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determining the current principal element generating an abnormal condition according to the first contribution value; and respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining the current sample data causing the abnormal production according to the second contribution value.
Description
Technical Field
The invention relates to the technical field of steelmaking continuous casting, in particular to a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine.
Background
The heat transfer state of the continuous casting crystallizer determines the working state of the continuous casting crystallizer, and finally influences the surface quality of the continuous casting slab and the production efficiency of the continuous casting machine.
However, the heat transfer process in the crystallizer is very complex and is limited by continuous casting high-temperature detection equipment, and the heat state and the heat transfer effect of a casting blank in the crystallizer are still lack of effective direct observation means, so that the heat transfer state of the continuous casting crystallizer cannot be monitored in real time, the quality index of the continuous casting blank cannot be controlled in the production process, and the quality of the continuous casting blank cannot be ensured.
Disclosure of Invention
Aiming at the problems in the prior art, the embodiment of the invention provides a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine, which are used for solving the technical problems that the heat transfer state of the continuous casting crystallizer cannot be monitored in real time, the quality index of a continuous casting billet cannot be controlled in the production process, and the quality of the continuous casting billet cannot be ensured in the prior art.
The embodiment of the invention provides a method for monitoring the thermal state of a crystallizer of a continuous casting machine, which comprises the following steps:
obtaining multiple groups of historical sample data in the continuous casting production process, wherein each group of historical sample data comprises: at least two of the crystallizer water inlet temperature, the pulling speed, the molten steel casting temperature and the crystallizer liquid level fluctuation value, the crystallizer cooling water flow and the crystallizer inlet and outlet water temperature rise;
standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix;
determining a principal component model of the historical sample data according to the standardized matrix;
determining a control limit value of a first square prediction error statistic value of the principal component model at any moment based on the historical sample data;
when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data;
judging whether the second square prediction error statistic value at each moment is larger than the control limit value, if so, acquiring a current group of online data corresponding to the second square prediction error statistic value at the current moment, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determining the current principal element generating an abnormal condition according to the first contribution value;
and respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining the current sample data causing abnormal production according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value.
In the foregoing solution, the determining the principal component model of the historical sample data according to the normalization matrix includes:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors and corresponding eigenvalues of the covariance matrix, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to each feature vector and the corresponding time.
In the foregoing solution, the determining a control limit value of a first mean prediction error statistic of the principal component model at any time based on the historical sample data includes:
according to the formula
Determining a first intermediate variable θ1Second intermediate variable theta2And a third intermediate variable theta3A value of (d); i is any time, j is any process variable, and
the eigenvalue of the covariance matrix is, n is the number of all principal elements, and k is the number of principal elements reserved in the principal element model;
according to the formula
Determining a fourth intermediate variable h0；
According to the formula
Determining a control limit value Q for a first mean prediction error statistic of the pivot modela(ii) a Wherein, the CaIs the critical value for a normal distribution at a significance level of a, which is 0.95.
In the foregoing solution, the determining the second contribution value of each sample data in the current set of online data to the current pivot respectively includes:
according to the formulaDetermining a second contribution value Q of each sample data of the current group of online data to the current pivotij(ii) a Wherein, the i is the time corresponding to the current group of online data, and the X isijIs the actual value of the jth sample data at the moment i, the
The predicted value of the jth sample data at the time point i is obtained.
In the foregoing scheme, before the performing the standardization processing on the historical sample data according to the preset processing policy to obtain the standardized matrix, the method includes:
and eliminating abnormal data in the historical sample data according to preset square prediction error SPE statistic.
The embodiment of the invention also provides a device for monitoring the thermal state of the crystallizer of the continuous casting machine, which comprises:
the acquisition unit is used for acquiring multiple groups of historical sample data in the continuous casting production process, and each group of historical sample data comprises: at least two of the crystallizer water inlet temperature, the pulling speed, the molten steel casting temperature and the crystallizer liquid level fluctuation value, the crystallizer cooling water flow and the crystallizer inlet and outlet water temperature rise;
the processing unit is used for carrying out standardization processing on the historical sample data according to a preset processing strategy to obtain a standardized matrix;
a first determining unit, configured to determine a principal component model of the historical sample data according to the normalization matrix;
a second determining unit, configured to determine, based on the historical sample data, a control limit value of a first square prediction error statistic of the principal component model at any time; when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data;
a judging unit, configured to judge whether the second square prediction error statistic at each time is greater than the control limit, and if so, obtain a current set of online data corresponding to the second square prediction error statistic at the current time, determine a first contribution value of each principal element in the principal element model to the second square prediction error statistic, and determine a current principal element generating an abnormal condition according to the first contribution value;
and a third determining unit, configured to determine a second contribution value of each online sample data in the current set of online data to the current pivot, and determine, according to the second contribution value, current sample data causing the abnormal production, where the current sample data is the online sample data corresponding to the largest second contribution value.
In the foregoing solution, the first determining unit is specifically configured to:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors and corresponding eigenvalues of the covariance matrix, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to each feature vector and the corresponding time.
In the foregoing solution, the second determining unit is specifically configured to:
according to the formulaDetermining a first intermediate variable θ1Second intermediate variable theta2And a third intermediate variable theta3A value of (d); i is any time, j is any process variable, and
is a characteristic of the covariance matrixThe n is the number of the historical sample data, and the k is the number of the principal elements reserved in the principal element model;
according to the formula
Determining a fourth intermediate variable h0；
According to the formula
Determining a control limit value Q for a squared prediction error statistic for the pivot modela(ii) a Wherein, the CaIs the critical value for a normal distribution at a significance level of a, which is 0.95.
In the foregoing solution, the third determining unit is specifically configured to:
according to the formula
Determining a second contribution value Q of each sample data of the current group of online data to the current pivotij(ii) a Wherein, the i is the time corresponding to each sample data, and the X isijIs the actual value of the jth sample data at the moment i, the
The predicted value of the jth sample data at the time point i is obtained.
In the above scheme, the apparatus further comprises: and the rejecting unit is used for carrying out standardized processing on the historical sample data according to a preset processing strategy, and rejecting abnormal data in the historical sample data according to a preset rejecting index before a standardized matrix is obtained.
The embodiment of the invention provides a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine, wherein the method comprises the following steps: obtaining multiple groups of historical sample data in the continuous casting production process, wherein each group of historical sample data comprises: cooling water flow of the crystallizer, water temperature rise of a water inlet and a water outlet of the crystallizer, water inlet temperature of the crystallizer and pulling speed; standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix; determining a principal component model of the historical sample data according to the standardized matrix; determining a control limit value of a first square prediction error statistic value of the principal component model at any moment based on the historical sample data; when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data; judging whether the second square prediction error statistic value at each moment is larger than the control limit value, if so, acquiring a current group of online data corresponding to the second square prediction error statistic value at the current moment, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determining the current principal element generating an abnormal condition according to the first contribution value; respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining current sample data causing abnormal production according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value; therefore, the principal component model is determined according to historical data in the continuous casting production process, the thermal state of the crystallizer is monitored on line by using the principal component model, sample data causing production abnormity is predicted, the quality index of the continuous casting billet can be effectively controlled in the production process, and the quality of the continuous casting billet is ensured.
Drawings
FIG. 1 is a schematic flow chart of a method for monitoring the thermal state of a crystallizer of a continuous casting machine according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a device for monitoring the thermal state of a crystallizer of a continuous casting machine according to an embodiment of the present invention;
fig. 3 is an SPE statistical diagram of 20 sets of historical sample data according to the third embodiment of the present invention;
fig. 4 is an SPE statistical graph after a group of abnormal history sample data is removed according to the third embodiment of the present invention;
fig. 5 is an SPE statistical graph after two groups of abnormal historical sample data are removed according to the third embodiment of the present invention;
fig. 6 is an SPE statistical diagram with control limits of historical sample data according to a third embodiment of the present invention;
fig. 7 is a schematic diagram illustrating a first contribution value of each principal element to the second square prediction error statistic according to the third embodiment of the present invention;
FIG. 8 is a second contribution value of the current set of online data to the first pivot element according to a third embodiment of the present invention;
FIG. 9 is a second contribution value of the current set of online data to the second pivot according to the third embodiment of the present invention;
FIG. 10 is a second contribution value of the current set of online data to the third pivot element according to the third embodiment of the present invention;
fig. 11 is an SPE statistical graph of 100 sets of historical sample data according to the fourth embodiment of the present invention;
fig. 12 is an SPE statistical graph after a group of abnormal history sample data is removed according to the fourth embodiment of the present invention;
fig. 13 is an SPE statistical graph after two groups of abnormal history sample data are removed according to the fourth embodiment of the present invention;
fig. 14 is an SPE statistical chart obtained after five groups of abnormal historical sample data are removed according to the fourth embodiment of the present invention;
FIG. 15 is a SPE statistical graph with control limits for historical sample data according to the fourth embodiment of the present invention; (ii) a
Fig. 16 is a schematic diagram illustrating a first contribution value of each principal element to the second square prediction error statistic according to the fourth embodiment of the present invention;
FIG. 17 is a second contribution value of the current set of online data to the second pivot according to the fourth embodiment of the present invention.
Detailed Description
The technical problems that in the prior art, the heat transfer state of a continuous casting crystallizer cannot be monitored in real time, the quality index of a continuous casting billet cannot be controlled in the production process, and the quality of the continuous casting billet cannot be guaranteed are solved. The invention provides a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine, wherein the method comprises the following steps: obtaining multiple groups of historical sample data in the continuous casting production process, wherein each group of historical sample data comprises: cooling water flow of the crystallizer, water temperature rise of a water inlet and a water outlet of the crystallizer, water inlet temperature of the crystallizer and pulling speed; standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix; determining a principal component model of the historical sample data according to the standardized matrix; determining a control limit value of a first square prediction error statistic value of the principal component model at any moment based on the historical sample data; when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data; judging whether the second square prediction error statistic value at each moment is larger than the control limit value, if so, acquiring a current group of online data corresponding to the second square prediction error statistic value at the current moment, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determining the current principal element generating an abnormal condition according to the first contribution value; and respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining the current sample data causing abnormal production according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value.
The technical solution of the present invention is further described in detail by the accompanying drawings and the specific embodiments.
Example one
The embodiment provides a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine, as shown in fig. 1, the method comprises the following steps:
s110, obtaining multiple groups of historical sample data in the continuous casting production process, wherein each group of historical sample data comprises: at least two of the crystallizer water inlet temperature, the pulling speed, the molten steel casting temperature and the crystallizer liquid level fluctuation value, the crystallizer cooling water flow and the crystallizer inlet and outlet water temperature rise;
here, when acquiring multiple sets of historical sample data in the continuous casting production process, the sample data may generally be index data that easily affects the production status, and therefore the method includes: cooling water flow of the crystallizer, water temperature rise of the inlet and the outlet of the crystallizer, water inlet temperature of the crystallizer, pulling speed, molten steel casting temperature, liquid level fluctuation value of the crystallizer and other parameters. The cooling water flow rate of the crystallizer and the water temperature rise of the inlet and the outlet of the crystallizer are required to be included, and parameters such as the pulling speed, the molten steel casting temperature, the fluctuation value of the liquid level of the crystallizer and the like are selected according to requirements. Then each set of historical sample data includes: at least two of the water inlet temperature of the crystallizer, the pulling speed, the molten steel casting temperature and the fluctuation value of the liquid level of the crystallizer, the cooling water flow of the crystallizer and the water temperature rise of the water inlet and the water outlet of the crystallizer.
When historical sample data in the continuous casting production process is acquired, a group of historical sample data is acquired at preset acquisition time intervals according to data acquisition time, and the acquisition time and the acquisition group number can be set according to actual conditions.
Generally, in order to ensure the prediction accuracy of the pivot model, the sample data is preferably normal data under a normal working condition, and after the historical sample data is acquired, if it is not determined whether the historical sample data is data under the normal working condition, the abnormal data in the historical sample data needs to be removed according to a preset square prediction error SPE, which is specifically implemented as follows:
determining the SPE value of each group of historical sample data, and eliminating the historical sample data corresponding to the maximum SPE value, wherein the historical sample data corresponding to the maximum SPE value is abnormal sample data;
and calculating according to the method, and removing abnormal sample data until the calculated SPE value meets a preset SPE value. Thus, normal historical sample data can be acquired.
S111, standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix; determining a principal component model of the historical sample data according to the standardized matrix;
after normal historical sample data is obtained, standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix; the preset processing strategy is a standard deviation standardized processing strategy. The concrete implementation is as follows:
determining the mean value and the standard deviation of each sample data in all historical sample data;
and for each sample data, subtracting the corresponding mean value from each numerical value, and dividing by the corresponding standard deviation to finally obtain a standardized matrix of the sample data. If the sample data includes 4 types and the number of sets of historical sample data is 20, the normalization matrix can be recorded as
After the standardized matrix is determined, as an optional embodiment, the determining the principal component model of the historical sample data according to the standardized matrix includes:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors and corresponding eigenvalues of the covariance matrix, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining a principal component model of the historical sample data according to the feature vectors after screening and the corresponding moments.
For example, the number of the feature values is m, and when the current cumulative contribution rate is greater than the cumulative contribution rate target value, the number of the feature values is k, then the number of the principal elements included in the principal element model is k, and k is smaller than m. The target value of the cumulative contribution rate is generally 0.85-0.95.
The final determined pivot model can be as shown in equation (1):
in the formula (1), the first and second groups,load vectors for the first k principal elements in the principal element model (it is also possible to think ofSolution as principal component eigenvectors), tkThe scoring phasor for the first k principal elements.
S112, determining a control limit value of a square prediction error statistic value of the pivot model at any moment based on the historical sample data;
after the principal component model is determined, the control limit is calculated on the basis of a certain assumption, and when the test level is a, a control limit (generally referred to as a control limit) of a square prediction error statistic of the principal component model at any time i can be determined according to formula (2) based on historical sample data, wherein the determined control limit corresponds to a reference control limit.
Said C isaIs a critical value for a normal distribution at a significance level of a, said a being 0.95; theta1Is a first intermediate variable, θ2Is a second intermediate variable, θ3Is a third intermediate variable, h0Is the fourth intermediate variable.
Wherein the content of the first and second substances,
in formula (3), i is any time, j is any process variable, and
and determining a fourth intermediate variable when the value of i is 1, 2 and 3 under a certain assumed condition, wherein n is the number of all principal elements, k is the number of principal elements reserved in the principal element model.
S113, when the heat state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data;
after the principal component model is determined, when the principal component model is used for online monitoring of the thermal state of the crystallizer, the second square prediction error SPE statistical value of the principal component model at any moment can be determined according to each group of online data. Where each time corresponds to a set of online data.
If there are 20 sets of online data, then finally 20 second SPE statistics may be obtained.
S114, judging whether the second square prediction error statistic value is larger than the control limit value, and if so, acquiring the current group of online data corresponding to the second square prediction error statistic value; determining a first contribution value of each principal element in the principal element model to the second mean prediction error statistic value, and determining a current principal element generating an abnormal condition according to the first contribution value;
and when obtaining each second SPE statistical value, judging whether the second square prediction error statistical value is larger than the determined control limit value, if so, obtaining the current group of online data corresponding to the second square prediction error statistical value, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistical value, and determining the current principal element generating an abnormal condition according to the first contribution value.
Wherein, the online data is consistent with the historical sample data, that is, the online data also includes: four parameters (which can be understood as four process variables) of crystallizer cooling water flow, crystallizer inlet and outlet water temperature rise, crystallizer inlet water temperature and pulling speed.
Such as: if the second square prediction error statistic corresponding to the 15 th group of online data is greater than the control limit value, it indicates that the 15 th group of online data is abnormal, which only indicates that the production condition is abnormal, but cannot determine a specific fault source, and it is necessary to determine a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determine a current principal element generating an abnormal condition according to the first contribution value, where the principal element corresponding to the largest first contribution value is the current principal element.
Here, for example, if the principal component model includes 3 principal components, the first contribution value of the first principal component is 4, the first contribution value of the second principal component is 3, and the first contribution value of the third principal component is 5, then it may be determined that the main cause of the abnormal condition is the third principal component. Then the third principal is the current principal.
S115, respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining the current online sample data causing abnormal production according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value;
because the current group of online data comprises 4 process variables, the first process variable is the pulling speed, the second process variable is the cooling water flow of the crystallizer, the third process variable is the water inlet temperature of the crystallizer, and the fourth process variable is the water inlet and outlet temperature rise of the crystallizer. The abnormal occurrence of each process variable can cause abnormal production conditions, so that it is further determined which process variable has the abnormal occurrence. Then, a second contribution value of each online sample data in the current set of online data to the current pivot needs to be determined, and the current online sample data causing the abnormal production is determined according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value.
Here, a second contribution Q of each sample data of the first current set of online data to the current pivot may be determined according to equation (5)ij：
In formula (5), i is the time corresponding to the current set of online data, and XijIs the actual value of the jth sample data at the moment i, theThe predicted value of the jth sample data at the time point i is obtained.
For example, if the second contribution value of the first process variable is 5, the second contribution value of the second process variable is 6, the second contribution value of the third process variable is 7, and the second contribution value of the fourth process variable is 8, it is determined that the fourth process variable is abnormal, that is, the temperature rise of the water at the inlet and outlet of the crystallizer is abnormal.
Therefore, the thermal state of the crystallizer of the continuous casting machine can be accurately monitored on line, the process variable causing production abnormity is predicted, the quality index of the continuous casting billet can be effectively controlled in the production process, and the quality of the continuous casting billet is ensured.
Based on the same inventive concept, the embodiment further provides a device for monitoring the thermal state of the crystallizer of the continuous casting machine, which is detailed in the second embodiment.
Example two
The embodiment provides a monitoring device for the thermal state of a crystallizer of a continuous casting machine, as shown in fig. 2, the device comprises: an acquisition unit 21, a processing unit 22, a first determination unit 23, a second determination unit 24, a judgment unit 25, a third determination unit 26 and a rejection unit 27;
here, when the obtaining unit 21 obtains a plurality of sets of history sample data in the continuous casting production process, since the sample data is generally index data that easily affects the production status, the obtaining unit may include: cooling water flow of the crystallizer, water temperature rise of the inlet and the outlet of the crystallizer, water inlet temperature of the crystallizer, pulling speed, molten steel casting temperature, liquid level fluctuation value of the crystallizer and other parameters. The cooling water flow rate of the crystallizer and the water temperature rise of the inlet and the outlet of the crystallizer are required to be included, and parameters such as the pulling speed, the molten steel casting temperature, the fluctuation value of the liquid level of the crystallizer and the like are selected according to requirements. Each set of historical sample data includes: at least two of crystallizer water inlet temperature, pulling speed, molten steel casting temperature and crystallizer liquid level fluctuation value, crystallizer cooling water flow and crystallizer inlet and outlet water temperature
When historical sample data in the continuous casting production process is acquired, a group of historical sample data is acquired at preset acquisition time intervals according to data acquisition time, and the acquisition time and the acquisition group number can be set according to actual conditions.
Generally, in order to ensure the prediction accuracy of the pivot model, the sample data is preferably normal data under a normal working condition, and after the historical sample data is acquired, if it is not determined whether the historical sample data is data under the normal working condition, the removing unit 27 further removes abnormal data in the historical sample data according to a preset square prediction error SPE, which is specifically implemented as follows:
determining the SPE value of each group of historical sample data, and eliminating the historical sample data corresponding to the maximum SPE value, wherein the historical sample data corresponding to the maximum SPE value is abnormal sample data;
and calculating according to the method, and removing abnormal sample data until the calculated SPE value meets a preset SPE value. Thus, normal historical sample data can be acquired.
After the obtaining unit 21 obtains normal historical sample data, the processing unit 22 is configured to perform standardized processing on the historical sample data according to a preset processing strategy to obtain a standardized matrix; the preset processing strategy is a standard deviation standardized processing strategy. The concrete implementation is as follows:
determining the mean value and the standard deviation of each sample data in all historical sample data;
and for each sample data, subtracting the corresponding mean value from each numerical value, and dividing by the corresponding standard deviation to finally obtain a standardized matrix of the sample data. If the sample data includes 4 types and the number of sets of historical sample data is 20, the normalization matrix can be recorded as
After the normalization matrix is determined, as an alternative embodiment, the first determining unit 23 is specifically configured to:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors and corresponding eigenvalues of the covariance matrix, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to each feature vector and the corresponding time.
For example, the number of the feature values is m, and when the current cumulative contribution rate is greater than the cumulative contribution rate target value, the number of the feature values is k, then the number of the principal elements included in the principal element model is k, and k is smaller than m.
The final determined pivot model can be as shown in equation (1):
in the formula (1), the first and second groups,
a transpose of a matrix of load vectors (also understood as principal component eigenvectors) of the first k principal components in the principal component model, tkThe scoring phasor for the first k principal elements.
After the principal component model is determined, the calculation of the control limit is based on a certain assumption, and when the test level is a, the second determining unit 24 may determine a control limit (generally referred to as a control limit) of the square prediction error statistic of the principal component model at any time i according to formula (2) based on the historical sample data, where the determined control limit corresponds to a reference control limit.
Said C isaIs a cutoff value for normal distribution at a significance level of a, said a being 0.05; theta1Is a first intermediate variable, θ2Is a second intermediate variable, θ3Is a third intermediate variable, h0Is the fourth intermediate variable.
Wherein the content of the first and second substances,
in formula (3), i is any time, j is any process variable, and
and determining a fourth intermediate variable when the value of i is 1, 2 and 3 under a certain assumed condition, wherein n is the number of all principal elements, k is the number of principal elements reserved in the principal element model.
After the principal component model is determined, when the principal component model is used to perform online monitoring on the thermal state of the crystallizer, the second determining unit 24 may further determine a second square prediction error SPE statistical value of the principal component model at any time according to each group of online data. Where each time corresponds to a set of online data.
If there are 20 sets of online data, then finally 20 second SPE statistics may be obtained.
When obtaining each second SPE statistic, the determining unit 25 is configured to determine whether the second square prediction error statistic is greater than the determined control limit, and if so, obtain the current group of online data corresponding to the second square prediction error statistic. And determining a first contribution value of each principal element in the principal element model to the second mean prediction error statistic value, and determining a current principal element generating an abnormal condition according to the first contribution value. Wherein, the online data is consistent with the historical sample data, that is, the online data also includes: four parameters (which can be understood as four process variables) of crystallizer cooling water flow, crystallizer inlet and outlet water temperature rise, crystallizer inlet water temperature and pulling speed.
Such as: if the second square prediction error statistic corresponding to the 15 th group of online data is greater than the control limit value, it indicates that the 15 th group of online data is abnormal, which only indicates that the production condition is abnormal, but cannot determine a specific fault source, the determining unit 25 further needs to determine a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, determine a current principal element generating an abnormal condition according to the first contribution value, and determine the principal element corresponding to the largest first contribution value as the current principal element.
Here, for example, if the principal component model includes 3 principal components, the first contribution value of the first principal component is 4, the first contribution value of the second principal component is 3, and the first contribution value of the third principal component is 5, then it may be determined that the main cause of the abnormal condition is the third principal component. Then the third principal is the current principal.
Because the current group of online data comprises 4 process variables, the first process variable is the pulling speed, the second process variable is the cooling water flow of the crystallizer, the third process variable is the water inlet temperature of the crystallizer, and the fourth process variable is the water inlet and outlet temperature rise of the crystallizer. Since each process variable abnormality causes an abnormality in the production condition, the third determination unit 26 needs to further determine which process variable has the abnormality. Then, a second contribution value of each online sample data in the current set of online data to the current pivot needs to be determined, and current sample data causing abnormal production is determined according to the second contribution value, where the current sample data is the online sample data corresponding to the maximum second contribution value.
Here, the third determining unit 26 may determine a second contribution Q of each sample data of the first current set of online data to the square prediction error statistic according to equation (5)ij：
In formula (5), i is the time corresponding to the current set of online data, and XijIs the actual value of the jth sample data at the moment i, the
The predicted value of the jth sample data at the time point i is obtained.
For example, if the second contribution value of the first process variable is 5, the second contribution value of the second process variable is 6, the second contribution value of the third process variable is 7, and the second contribution value of the fourth process variable is 8, it is determined that the fourth process variable is abnormal, that is, the temperature rise of the water at the inlet and outlet of the crystallizer is abnormal.
Therefore, the thermal state of the crystallizer of the continuous casting machine can be accurately monitored on line, the process variable causing production abnormity is predicted, the quality index of the continuous casting billet can be effectively controlled in the production process, and the quality of the continuous casting billet is ensured.
EXAMPLE III
In practical applications, if the method provided in the first embodiment and the apparatus provided in the second embodiment are used to monitor the thermal state of the crystallizer of the continuous casting machine on line, the following steps are specifically implemented:
because the conventional square billet crystallizer is designed as an integral copper pipe and only has one crystallizer water quantity and corresponding crystallizer inlet and outlet water temperature rise, historical sample data comprise the crystallizer cooling water flow quantity, the crystallizer inlet and outlet water temperature rise, the crystallizer inlet water temperature and the pulling speed. For example, if the production steel grade is 45 steel and the casting section is 180 × 220mm, when historical sample data in the continuous casting production process is acquired, a group of historical sample data needs to be acquired at preset acquisition time intervals according to data acquisition time, the acquisition time is 5s, the number of the acquired groups is 20, and the historical sample data can be as shown in table 1:
TABLE 1
Generally, in order to ensure the prediction accuracy of the pivot model, the sample data is preferably normal data under a normal working condition, and after the historical sample data is acquired, if it is not determined whether the historical sample data is data under the normal working condition, the abnormal data in the historical sample data needs to be removed according to a preset square prediction error SPE, which is implemented as follows:
and determining the SPE value of each group of historical sample data, and eliminating the historical sample data corresponding to the maximum SPE value, wherein the historical sample data corresponding to the maximum SPE value is abnormal sample data. Referring to fig. 3, fig. 3 shows the SPE values of 20 determined sets of history sample data. As can be seen from fig. 3, the SPE statistic of the 14 th group of sample data is the largest, at which time, the first eigenvalue λ of the covariance matrix1,01.8526, the 14 th set of sample data is culled.
And continuing to determine the SPE value of each group of the history sample data after being removed, referring to fig. 4, where fig. 4 is the determined SPE value of 19 groups of history sample data. As can be seen from fig. 4, the SPE statistic of the 12 th group of sample data is the maximum, and at this time, the first eigenvalue λ of the covariance matrix established after deleting the 12 th group of sample data1,11.8697, the 12 th set of sample data is culled.
At this time | λ1,1-λ1,0|＝0.0171>ξ, then it needs to be eliminated. Where ξ is 0.01, which is a preset reference value.
And continuing to determine the SPE value of each group of the history sample data after being removed, referring to fig. 5, where fig. 5 is the determined SPE value of 18 groups of history sample data. As can be seen from fig. 5, the SPE statistic of the 9 th group of sample data is the maximum, and at this time, the first eigenvalue λ of the covariance matrix established after deleting the 9 th group of sample data1,2＝1.8757，|λ1,2-λ1,1|＝0.006<ξ. The remaining 18 sets of historical sample data are illustrated as normal data.
After normal historical sample data is obtained, standardizing the historical sample data according to the processing strategy provided by the first embodiment to obtain a standardized matrix; and then determining a principal component model of the historical sample data according to the standardized matrix, wherein the method is specifically realized as follows:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors (corresponding to all principal elements) of the covariance matrix and corresponding eigenvalues, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to the characteristic values after screening and the corresponding moments.
The eigenvalues of the covariance matrix, the contribution rate of each eigenvalue, and the cumulative contribution rate are shown in table 2:
TABLE 2
In this embodiment, the cumulative contribution rate target value is 0.85, and as can be seen from table 2, 0.9779 is greater than 0.85, so that 3 pivot elements can be reserved, and 4 pivot elements can be replaced by 3 pivot elements to reduce the dimension of the pivot model.
After the principal component model is determined, the control limit determined by the method provided in the first embodiment is 0.33119.
When the principal component model is used for carrying out online monitoring on the thermal state of the crystallizer, the second square prediction error SPE statistical value of the principal component model at any moment can be determined according to each group of online data. Where each time corresponds to a set of online data. The resulting second square prediction error SPE statistics may refer to fig. 6.
As can be seen from FIG. 6, if the second SPE statistic for the 58 th online data exceeds the control limit, indicating that the production condition is abnormal at 23:56:28, then the process variables in the 58 th online data are obtained. And determining a first contribution value of each principal element in the principal element model to the second mean prediction error statistic, the first contribution value of each principal element being referred to in fig. 7.
As can be seen in FIG. 7, the first principal has the largest first contribution value, and it is the first principal that produces the exception condition.
Because the current group of online data comprises 4 process variables, the first process variable is the pulling speed, the second process variable is the cooling water flow of the crystallizer, the third process variable is the water inlet temperature of the crystallizer, and the fourth process variable is the water inlet and outlet temperature rise of the crystallizer. The abnormal occurrence of each process variable can cause abnormal production conditions, so that it is further determined which process variable has the abnormal occurrence. Then, a second contribution value of each sample data in the current set of online data to the current pivot needs to be determined, and the current sample data causing the abnormal production is determined according to the second contribution value, where the current sample is the sample data corresponding to the maximum second contribution value.
Referring to fig. 8 to 10, fig. 8 shows second contribution values of 4 variables to a first pivot, fig. 9 shows second contribution values of 4 variables to a second pivot, and fig. 10 shows third contribution values of 4 variables to the first pivot. As can be seen in fig. 8, the second contribution value of the 2 nd variable is the largest, so that it can be determined that the 2 nd variable is abnormal, that is, at the time 23:56:28, the flow rate of the mold cooling water may cause the casting slab to have a quality defect, so that the flow rate of the mold cooling water can be controlled in the production process to ensure the quality of the casting slab.
The online data of this embodiment is 100 groups, which is detailed in table 3.
TABLE 3
Example four
In practical applications, if the method provided in the first embodiment and the apparatus provided in the second embodiment are used to monitor the thermal state of the crystallizer of the continuous casting machine on line, the following steps are specifically implemented:
the slab crystallizer is a combined copper plate, and the detectable process variables are obviously more than those of the square slab crystallizer in number. Selecting 10 variables, including: the water flow of the left side of the crystallizer, the water temperature rise of the left side inlet and outlet of the crystallizer, the water flow of the right side of the crystallizer, the water temperature rise of the right side inlet and outlet of the crystallizer, the water flow of the inner arc side of the crystallizer, the water temperature rise of the inner arc side inlet and outlet of the crystallizer, the water flow of the outer arc side of the crystallizer, the water temperature rise of the side inlet and outlet of the outer arc side of the crystallizer.
When historical sample data in the continuous casting production process is acquired, a group of historical sample data is acquired at preset acquisition time intervals according to data acquisition time, the acquisition time is 5s, the number of the acquired groups is 20, and the historical sample data can be shown in a table 4:
TABLE 4
Generally, in order to ensure the prediction accuracy of the pivot model, the sample data is preferably normal data under a normal working condition, and after the historical sample data is acquired, if it is not determined whether the historical sample data is data under the normal working condition, the abnormal data in the historical sample data needs to be removed according to a preset square prediction error SPE, which is implemented as follows:
and determining the SPE value of each group of historical sample data, and eliminating the historical sample data corresponding to the maximum SPE value, wherein the historical sample data corresponding to the maximum SPE value is abnormal sample data. Referring to fig. 11, fig. 11 illustrates SPE values of 100 determined sets of history sample data. As can be seen from fig. 11, the SPE statistic of the 85 th group of sample data is the maximum, at which the first eigenvalue λ of the covariance matrix1,04.1714, then the 85 th set of sample data is culled.
And continuing to determine the SPE value of each group of the removed historical sample data, referring to fig. 12, where fig. 12 is the determined SPE value of 99 groups of the historical sample data. As can be seen from fig. 12, the SPE statistic of the 83 th group of sample data is the maximum, and at this time, the agreement created by the 83 th group of sample data is deletedFirst eigenvalue λ of the variance matrix1,14.1524, then the 83 th set of sample data is culled.
At this time | λ1,1-λ1,0|＝0.019>ξ, then it needs to be eliminated. Where ξ is 0.01, which is a preset reference value.
And continuing to determine the SPE values of each group of the history sample data after being removed, referring to fig. 13, where fig. 13 is the determined SPE values of 98 groups of history sample data. Referring to fig. 13, fig. 13 illustrates the determined SPE values of 98 sets of history sample data. As can be seen from fig. 13, the SPE statistic of the 5 th sample data is the maximum, and at this time, the second eigenvalue λ of the covariance matrix created by the 98 th sample data is deleted1,24.1418, then the 5 th set of sample data is culled.
After 5 sets of historical sample data are removed in total according to the above method, normal data are obtained, and an SPE statistical graph after five sets of abnormal historical sample data are removed is shown in fig. 14.
After normal historical sample data is obtained, standardizing the historical sample data according to the processing strategy provided by the first embodiment to obtain a standardized matrix; and then determining a principal component model of the historical sample data according to the standardized matrix, wherein the method is specifically realized as follows:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors (corresponding to all principal elements) of the covariance matrix and corresponding eigenvalues, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to the characteristic values after screening and the corresponding moments.
The eigenvalues of the covariance matrix, the contribution rate of each eigenvalue, and the cumulative contribution rate are shown in table 5:
TABLE 5
In this embodiment, the cumulative contribution rate target value is 0.85, and as can be seen from table 5, 0.8928 is greater than 0.85, so that 5 pivot elements can be reserved, and 10 pivot elements can be replaced by 5 pivot elements to reduce the dimension of the pivot model.
After the principal component model is determined, the control limit determined by the method provided in the first embodiment is 12.9727.
When the principal component model is used for carrying out online monitoring on the thermal state of the crystallizer, the second square prediction error SPE statistical value of the principal component model at any moment can be determined according to each group of online data. Where each time corresponds to a set of online data. The resulting second square prediction error SPE statistic can be referred to fig. 15.
As can be seen from fig. 11, if the second SPE statistic of the 237 th online data exceeds the control limit, which indicates that the production condition is abnormal at the time of 00:56:06, the process variables in the 237 th online data are acquired. And determining a first contribution value of each principal element in the principal element model to the second mean prediction error statistic, the first contribution value of each principal element being referred to in fig. 16.
As can be seen in fig. 16, the first contribution of the second principal is the largest, and it is the second principal that generated the exception condition.
Because the current set of online data includes 10 process variables, see table 4, the first process variable is the pulling rate, the second process variable is the liquid level, the third process variable is the left water quantity of the crystallizer, the fourth process variable is the left water temperature rise of the crystallizer, the fifth process variable is the right water quantity, the sixth process variable is the right water temperature rise of the crystallizer, the seventh process variable is the inner arc water quantity of the crystallizer, the eighth process variable is the inner arc water temperature rise of the crystallizer, the ninth process variable is the outer arc water quantity of the crystallizer, and the tenth process variable is the outer arc water temperature rise of the crystallizer. The abnormal occurrence of each process variable can cause abnormal production conditions, so that it is further determined which process variable has the abnormal occurrence. Then, a second contribution value of each sample data in the current set of online data to the current pivot needs to be determined, and the current sample data causing the abnormal production is determined according to the second contribution value, where the current sample is the sample data corresponding to the maximum second contribution value.
Referring to fig. 12, fig. 17 shows the second contribution values of the 10 variables to the second principal element, and it can be seen from fig. 17 that the second contribution value of the 2 nd variable is the largest, so that it can be determined that the abnormality occurs in the 2 nd variable, that is, the crystallizer liquid level may cause the quality defect of the casting slab at the time 00:56:06, so that the crystallizer liquid level can be controlled in the production process to ensure the quality of the casting slab.
The online data of this embodiment is 400 groups, and since the first 200 groups of data are normal data, only the 400 groups of data 201 and 201 are shown here, which is detailed in table 6.
TABLE 6
The embodiment of the invention provides a method and a device for monitoring the thermal state of a crystallizer of a continuous casting machine, which have the following beneficial effects that:
obtaining multiple groups of historical sample data in the continuous casting production process, wherein each group of historical sample data comprises: cooling water flow of the crystallizer, water temperature rise of a water inlet and a water outlet of the crystallizer, water inlet temperature of the crystallizer and pulling speed; standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix; determining a principal component model of the historical sample data according to the standardized matrix; determining a control limit value of a first square prediction error statistic value of the principal component model at any moment based on the historical sample data; when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data; judging whether the second square prediction error statistic value at each moment is larger than the control limit value, if so, acquiring a current group of online data corresponding to the second square prediction error statistic value at the current moment, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determining the current principal element generating an abnormal condition according to the first contribution value; respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining current sample data causing abnormal production according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value; therefore, the principal component model is determined according to historical data in the continuous casting production process, the thermal state of the crystallizer is monitored on line by using the principal component model, sample data causing production abnormity is predicted, the quality index of the continuous casting billet can be effectively controlled in the production process, and the quality of the continuous casting billet is ensured.
The above description is only exemplary of the present invention and should not be taken as limiting the scope of the present invention, and any modifications, equivalents, improvements, etc. that are within the spirit and principle of the present invention should be included in the present invention.
Claims (10)
1. A method for monitoring the thermal condition of a crystallizer of a continuous casting machine, comprising:
obtaining multiple groups of historical sample data in the continuous casting production process, wherein each group of historical sample data comprises: at least two of the crystallizer water inlet temperature, the pulling speed, the molten steel casting temperature and the crystallizer liquid level fluctuation value, the crystallizer cooling water flow and the crystallizer inlet and outlet water temperature rise;
standardizing the historical sample data according to a preset processing strategy to obtain a standardized matrix;
determining a principal component model of the historical sample data according to the standardized matrix;
determining a control limit value of a first square prediction error statistic value of the principal component model at any moment based on the historical sample data;
when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data;
judging whether the second square prediction error statistic value at each moment is larger than the control limit value, if so, acquiring a current group of online data corresponding to the second square prediction error statistic value at the current moment, determining a first contribution value of each principal element in the principal element model to the second square prediction error statistic value, and determining the current principal element generating an abnormal condition according to the first contribution value;
and respectively determining a second contribution value of each online sample data in the current group of online data to the current pivot, and determining the current sample data causing abnormal production according to the second contribution value, wherein the current sample data is the online sample data corresponding to the maximum second contribution value.
2. The method of claim 1, wherein said determining a principal component model of said historical sample data from said normalization matrix comprises:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors and corresponding eigenvalues of the covariance matrix, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to each feature vector and the corresponding time.
3. The method of claim 1, wherein said determining a control limit for a first mean prediction error statistic for said principal component model at any one time based on said historical sample data comprises:
according to the formula
Determining a first intermediate variable θ1Second intermediate variable theta2And a third intermediate variable theta3A value of (d); i is any time, j is any process variable, and
the eigenvalue of the covariance matrix is, n is the number of all principal elements, and k is the number of principal elements reserved in the principal element model;
according to the formula
Determining a fourth intermediate variable h0；
According to the formula
Determining a control limit value Q for a first mean prediction error statistic of the pivot modela(ii) a Wherein, the CaIs normally distributed in the displayThe critical value for a is the severity level, which is 0.95.
4. The method of claim 1, wherein said separately determining a second contribution value of each sample data in the current set of online data to the current pivot comprises:
according to the formula
Determining a second contribution value Q of each sample data of the current group of online data to the current pivotij(ii) a Wherein, the i is the time corresponding to the current group of online data, and the X isijIs the actual value of the jth sample data at the moment i, the
The predicted value of the jth sample data at the time point i is obtained.
5. The method of claim 1, wherein before the normalizing the historical sample data according to the preset processing strategy to obtain the normalized matrix, the method comprises:
and eliminating abnormal data in the historical sample data according to preset square prediction error SPE statistic.
6. A device for monitoring the thermal state of the crystallizer of a continuous casting machine, characterized in that it comprises:
the acquisition unit is used for acquiring multiple groups of historical sample data in the continuous casting production process, and each group of historical sample data comprises: at least two of the crystallizer water inlet temperature, the pulling speed, the molten steel casting temperature and the crystallizer liquid level fluctuation value, the crystallizer cooling water flow and the crystallizer inlet and outlet water temperature rise;
the processing unit is used for carrying out standardization processing on the historical sample data according to a preset processing strategy to obtain a standardized matrix;
a first determining unit, configured to determine a principal component model of the historical sample data according to the normalization matrix;
a second determining unit, configured to determine, based on the historical sample data, a control limit value of a first square prediction error statistic of the principal component model at any time; when the thermal state of the crystallizer is monitored on line, determining a second square prediction error statistic value of the principal component model at any moment based on each group of online data;
a judging unit, configured to judge whether the second square prediction error statistic at each time is greater than the control limit, and if so, obtain a current set of online data corresponding to the second square prediction error statistic at the current time, determine a first contribution value of each principal element in the principal element model to the second square prediction error statistic, and determine a current principal element generating an abnormal condition according to the first contribution value;
and a third determining unit, configured to determine a second contribution value of each online sample data in the current set of online data to the current pivot, and determine, according to the second contribution value, current sample data causing the abnormal production, where the current sample data is the online sample data corresponding to the largest second contribution value.
7. The apparatus of claim 6, wherein the first determining unit is specifically configured to:
determining a covariance matrix of the historical sample data according to the standardized matrix;
determining eigenvectors and corresponding eigenvalues of the covariance matrix, wherein the eigenvalues correspond to all principal elements in the principal element model;
determining the cumulative contribution rate of each eigenvalue;
screening according to the characteristic value based on a preset accumulated contribution rate target value, and determining the screened characteristic value, wherein the screened characteristic value corresponds to a principal element reserved in the principal element model;
and determining the principal component model of the historical sample data according to each feature vector and the corresponding time.
8. The apparatus of claim 6, wherein the second determining unit is specifically configured to:
according to the formula
Determining a first intermediate variable θ1Second intermediate variable theta2And a third intermediate variable theta3A value of (d); i is any time, j is any process variable, and
the eigenvalue of the covariance matrix is, the n is the number of the historical sample data, and the k is the number of the principal elements reserved in the principal element model;
according to the formula
Determining a fourth intermediate variable h0；
According to the formulaDetermining a control limit value Q for a squared prediction error statistic for the pivot modela(ii) a Wherein, the CaIs the critical value for a normal distribution at a significance level of a, which is 0.95.
9. The apparatus of claim 6, wherein the third determining unit is specifically configured to:
according to the formula
Determining a second contribution value Q of each sample data of the current group of online data to the current pivotij(ii) a Wherein, the i is the time corresponding to each sample data, and the X isijIs the actual value of the jth sample data at the moment i, the
The predicted value of the jth sample data at the time point i is obtained.
10. The apparatus of claim 6, wherein the apparatus further comprises: and the rejecting unit is used for carrying out standardized processing on the historical sample data according to a preset processing strategy, and rejecting abnormal data in the historical sample data according to a preset rejecting index before a standardized matrix is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910933908.0A CN110705785A (en) | 2019-09-29 | 2019-09-29 | Method and device for monitoring thermal state of crystallizer of continuous casting machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910933908.0A CN110705785A (en) | 2019-09-29 | 2019-09-29 | Method and device for monitoring thermal state of crystallizer of continuous casting machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110705785A true CN110705785A (en) | 2020-01-17 |
Family
ID=69196325
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910933908.0A Pending CN110705785A (en) | 2019-09-29 | 2019-09-29 | Method and device for monitoring thermal state of crystallizer of continuous casting machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110705785A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112199409A (en) * | 2020-08-17 | 2021-01-08 | 浙江中控技术股份有限公司 | Method and device for monitoring real-time working condition of catalytic reforming device |
| CN113510234A (en) * | 2021-09-14 | 2021-10-19 | 深圳市信润富联数字科技有限公司 | Quality monitoring method and device for low-pressure casting of hub and electronic equipment |
| CN114819391A (en) * | 2022-05-19 | 2022-07-29 | 中山大学 | Photovoltaic power generation power prediction method based on historical data set time span optimization |
| CN116049658A (en) * | 2023-03-30 | 2023-05-02 | 西安热工研究院有限公司 | Wind turbine generator abnormal data identification method, system, equipment and medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040172153A1 (en) * | 2002-12-12 | 2004-09-02 | Yale Zhang | Method and online system for monitoring continuous caster start-up operation and predicting start cast breakouts |
| CN104933012A (en) * | 2015-05-08 | 2015-09-23 | 西南交通大学 | Method for online identifying measurement deviation fault for traction substation instrument transformer |
-
2019
- 2019-09-29 CN CN201910933908.0A patent/CN110705785A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040172153A1 (en) * | 2002-12-12 | 2004-09-02 | Yale Zhang | Method and online system for monitoring continuous caster start-up operation and predicting start cast breakouts |
| CN104933012A (en) * | 2015-05-08 | 2015-09-23 | 西南交通大学 | Method for online identifying measurement deviation fault for traction substation instrument transformer |
Non-Patent Citations (3)
| Title |
|---|
| 徐金梧桐 等: "《冶金生产过程质量监控理论与方法》", 31 May 2015 * |
| 朱苗勇 主编: "《现代冶金学 钢铁冶金卷》", 30 September 2005 * |
| 陈勇: "基于多元统计分析的生产过程故障诊断研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112199409A (en) * | 2020-08-17 | 2021-01-08 | 浙江中控技术股份有限公司 | Method and device for monitoring real-time working condition of catalytic reforming device |
| CN112199409B (en) * | 2020-08-17 | 2024-06-11 | 浙江中控技术股份有限公司 | Method and device for monitoring real-time working condition of catalytic reforming device |
| CN113510234A (en) * | 2021-09-14 | 2021-10-19 | 深圳市信润富联数字科技有限公司 | Quality monitoring method and device for low-pressure casting of hub and electronic equipment |
| CN113510234B (en) * | 2021-09-14 | 2022-01-07 | 深圳市信润富联数字科技有限公司 | Quality monitoring method and device for low-pressure casting of hub and electronic equipment |
| CN114819391A (en) * | 2022-05-19 | 2022-07-29 | 中山大学 | Photovoltaic power generation power prediction method based on historical data set time span optimization |
| CN116049658A (en) * | 2023-03-30 | 2023-05-02 | 西安热工研究院有限公司 | Wind turbine generator abnormal data identification method, system, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110705785A (en) | Method and device for monitoring thermal state of crystallizer of continuous casting machine | |
| CN109766334B (en) | Processing method and system for online monitoring abnormal data of power equipment | |
| CN110765619B (en) | Failure-free canned motor pump fault short-term multi-step prediction method based on multi-state parameters | |
| JP5091604B2 (en) | Distribution evaluation method, product manufacturing method, distribution evaluation program, and distribution evaluation system | |
| CN106054840B (en) | A kind of online managing and control system of whole process product quality | |
| CA2497489A1 (en) | Real-time system and method of monitoring transient operations in continuous casting process for breakout prevention | |
| CN106649755B (en) | Threshold value self-adaptive setting abnormity detection method for multi-dimensional real-time power transformation equipment data | |
| CN110989510A (en) | Hot galvanizing product full-process quality control and grade automatic judgment system | |
| WO2015021751A1 (en) | Data-driven exception warning technical method for integrated circuit technology device | |
| CN109675935A (en) | A kind of IPCA operation of rolling on-line fault diagnosis method becoming control limit | |
| Adamy | Breakout prediction for continuous casting by fuzzy mealy automata | |
| CN105302123B (en) | The monitoring method of on-line measurement data | |
| CN108416386A (en) | A kind of method and system judged extremely for Hydropower Unit bearing temperature | |
| KR101882847B1 (en) | Method and apparatus for searching performance degradation source of power facilities | |
| CN113673600B (en) | Industrial signal abnormality early warning method, system, storage medium and computing device | |
| CN109240253B (en) | Online equipment diagnosis and preventive maintenance method and system | |
| CN110751217A (en) | Equipment energy consumption ratio early warning analysis method based on principal component analysis | |
| CN106204324A (en) | A kind of method determining that power plant's complex device key monitoring parameter and each parameters weighting distribute | |
| Kim et al. | Die-casting defect prediction and diagnosis system using process condition data | |
| CN116809653A (en) | Rolling stability monitoring and early warning method and system | |
| Shigemori | Quality and operation management system for steel products through multivariate statistical process control | |
| KR20200106791A (en) | Early warning device and method of power plant using switching technology of multiple-predictive models | |
| CN115081798A (en) | PLS algorithm-based plate and strip hot finish rolling process sub-frame combined monitoring and diagnosis method | |
| JP6358199B2 (en) | Method and apparatus for determining surface defects of continuous cast slab, and method for producing steel slab using the surface defect determination method | |
| WO2020103761A1 (en) | Bearing maintenance timing acquisition method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200117 |
|
| RJ01 | Rejection of invention patent application after publication |