CN114896900B - Target tracking system - Google Patents

Target tracking system Download PDF

Info

Publication number
CN114896900B
CN114896900B CN202210831903.9A CN202210831903A CN114896900B CN 114896900 B CN114896900 B CN 114896900B CN 202210831903 A CN202210831903 A CN 202210831903A CN 114896900 B CN114896900 B CN 114896900B
Authority
CN
China
Prior art keywords
furnace
target
strip steel
strip
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202210831903.9A
Other languages
Chinese (zh)
Other versions
CN114896900A (en
Inventor
曹卫华
张永月
宋文硕
贺江
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China University of Geosciences
Original Assignee
China University of Geosciences
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China University of Geosciences filed Critical China University of Geosciences
Priority to CN202210831903.9A priority Critical patent/CN114896900B/en
Publication of CN114896900A publication Critical patent/CN114896900A/en
Application granted granted Critical
Publication of CN114896900B publication Critical patent/CN114896900B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • C—CHEMISTRY; METALLURGY
    • C21—METALLURGY OF IRON
    • C21D—MODIFYING THE PHYSICAL STRUCTURE OF FERROUS METALS; GENERAL DEVICES FOR HEAT TREATMENT OF FERROUS OR NON-FERROUS METALS OR ALLOYS; MAKING METAL MALLEABLE, e.g. BY DECARBURISATION OR TEMPERING
    • C21D1/00—General methods or devices for heat treatment, e.g. annealing, hardening, quenching or tempering
    • C21D1/26—Methods of annealing
    • C—CHEMISTRY; METALLURGY
    • C21—METALLURGY OF IRON
    • C21D—MODIFYING THE PHYSICAL STRUCTURE OF FERROUS METALS; GENERAL DEVICES FOR HEAT TREATMENT OF FERROUS OR NON-FERROUS METALS OR ALLOYS; MAKING METAL MALLEABLE, e.g. BY DECARBURISATION OR TEMPERING
    • C21D11/00—Process control or regulation for heat treatments
    • C—CHEMISTRY; METALLURGY
    • C21—METALLURGY OF IRON
    • C21D—MODIFYING THE PHYSICAL STRUCTURE OF FERROUS METALS; GENERAL DEVICES FOR HEAT TREATMENT OF FERROUS OR NON-FERROUS METALS OR ALLOYS; MAKING METAL MALLEABLE, e.g. BY DECARBURISATION OR TEMPERING
    • C21D9/00—Heat treatment, e.g. annealing, hardening, quenching or tempering, adapted for particular articles; Furnaces therefor
    • C21D9/52—Heat treatment, e.g. annealing, hardening, quenching or tempering, adapted for particular articles; Furnaces therefor for wires; for strips ; for rods of unlimited length
    • C21D9/54—Furnaces for treating strips or wire
    • C21D9/56—Continuous furnaces for strip or wire
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation
    • 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)
  • Chemical & Material Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Metallurgy (AREA)
  • Thermal Sciences (AREA)
  • Organic Chemistry (AREA)
  • Materials Engineering (AREA)
  • Evolutionary Computation (AREA)
  • Mechanical Engineering (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Control Of Heat Treatment Processes (AREA)

Abstract

The invention provides a target tracking system, comprising: the system comprises a key variable acquisition module, a multivariate characteristic acquisition module, a target running state determination module and a target tracking prediction module, wherein the key variable acquisition module, the multivariate characteristic acquisition module, the target running state determination module and the target tracking prediction module are used for dividing the key variable into a bottom layer equipment layer, a running process layer and a plan index layer; analyzing a plurality of related variables, extracting state characteristics of the related variables, screening existing abnormal characteristics, making a state tracking strategy, fusing the multivariate characteristics to establish a target state detection model, and tracking a dynamically changed target in the furnace body in real time. And establishing a prediction model from the target to the position of each furnace section outlet and the like based on the tracking information, and predicting the time of entering and exiting each furnace section. By fusing sequence characteristics of multiple time-space level parameters, the real-time position of the target strip steel in the continuous annealing process is jointly decided and tracked, a prediction model from the target strip steel to each key position is established, and a foundation is laid for subsequent process modeling and system control.

Description

Target tracking system
Technical Field
The invention relates to the technical field of annealing and heating, in particular to a target tracking system.
Background
The continuous annealing and heating process is an important process on a cold-rolling hot galvanizing production line, and various strip steels entering an annealing furnace need to be heated according to corresponding heating requirements. The heating requirements of different strip steels in different furnace sections are inconsistent, so that different heating targets, namely the strip steels, in the furnace need to be tracked in real time so as to be conveniently and correspondingly regulated and controlled in time.
The general cold rolling hot galvanizing production line is introduced with a relatively mature production line and a system thereof abroad, and different target strip steels in the furnace cannot be accurately tracked in time under the condition. The reason is that: 1. although each roll of strip steel has an initial furnace entering plan before entering the production line, the plan has the initial length of each roll of strip steel, when entering the production line, according to different requirements, the crescent shears trim the two sides and the front and back of the strip steel, for example, the transition strip steel is trimmed in length, the specific length of the trimming is uncertain and unknown, so that the length of the strip steel heated in the furnace in the production line is uncertain and is not equal to the length of the strip steel in the initial furnace entering plan; 2. the strip steel trimmed by the crescent shears can be welded together before entering the annealing furnace, and the specific length of the strip steel is also influenced; 3. the combination modes of different types and specifications of strip steel in the production line are irregular; 4. when the production line runs, the whole process is continuously and uninterruptedly operated at a high speed day and night, so that a great deal of labor energy is consumed only by manual observation; 5. due to the difference of the types and the lengths of the strip steels and the high speed of operation, no sensor capable of detecting and tracking the target strip steel in real time exists in general production line equipment and the existing control system, and only welding seam detection devices are discretely arranged at a plurality of positions of a production line so as to roughly estimate the position of the target strip steel, so that the higher tracking requirement of the specific target strip steel cannot be met.
Disclosure of Invention
In order to solve the problems, the invention provides a target tracking system, which aims to fuse the sequence characteristics of multi-space-time hierarchical parameters and jointly decide and track the real-time position of target strip steel in the continuous annealing process, thereby establishing a prediction model of the target strip steel to each key position and laying a foundation for subsequent process modeling and system control. The target tracking system mainly comprises:
a key variable acquisition module for extracting parameters related to the position of the target strip steel as key variables according to the continuous annealing heating process in the annealing furnace and dividing the key variables into a bottom equipment layer I according to equipment and control types 1 Run Process layer I 2 And a planning index layer I 3 (ii) a The key variables comprise the type specification of the strip steel in a furnace plan, the length of the strip steel in a furnace entering plan, the furnace entering sequence of the strip steel in the furnace entering plan, the welding seam distance at the furnace entering position, the welding seam at the PH-NOF joint, the welding seam distance at the outlet of the NOF furnace section, the welding seam distance at the outlet of the RTF furnace section, the running speed, the outlet plate temperature of the NOF furnace section and the outlet plate temperature of the RTF furnace section;
a multi-element feature acquisition module for the bottom equipment layer I 1 And run Process layer I 2 Analyzing the running state of multiple related variables, extracting the data time sequence distribution characteristics of each variable, and determining multiple abnormal characteristics and bases existing in combination with the production process of the annealing furnaceScreening abnormal features according to the priority strategy and the time sequence distribution features to obtain multivariate features; wherein the plurality of related variables comprise a welding line distance at a furnace entrance, a welding line at a PH-NOF joint, a welding line distance at an outlet of an NOF furnace section, a welding line distance at an outlet of an RTF furnace section, an operation speed, an outlet plate temperature of the NOF furnace section and an outlet plate temperature of the RTF furnace section;
an operation state determination module of the target for determining the operation state of the target according to the planned index layer I in the existing system 3 The variable and multivariate feature acquisition module obtains multivariate features and formulates a gradual detailed state tracking strategy; based on the formulated conditions and constraints of the progressive detail state tracking strategy, a target state detection model is established by fusing the multivariate characteristics, so that the running state of the target is determined;
a target tracking prediction module for predicting the planning index layer I on a time scale 3 Time alignment is carried out on the operation state result of a certain coil steel in an operation state determining module of the variable and the target, wherein the plan index layer I 3 The variables comprise the type and specification of the strip steel in the furnace entering plan, the length of the strip steel in the furnace entering plan and the furnace entering sequence of the strip steel in the furnace entering plan, and equipment information of a furnace body is combined on a spatial scale, so that a dynamically changing target in the furnace body is tracked in real time; and establishing a prediction model from the target to the outlet position of each furnace section based on the tracking information of the target so as to predict the time of the target entering and exiting each furnace section.
Further, the specific process of obtaining the multivariate features in the multivariate feature obtaining module is as follows:
2.1: selecting the bottom device layer I 1 Related variables as sets
Figure 461697DEST_PATH_IMAGE001
And run Process layer I 2 Related variables as sets
Figure 926307DEST_PATH_IMAGE002
Wherein i = the distance of a welding line at a furnace entrance, the welding line at a PH-NOF joint, the distance of a welding line at an outlet of an NOF furnace section or the distance of a welding line at an outlet of an RTF furnace section, and j = the running speed, the temperature of an outlet plate of the NOF furnace section or the temperature of an outlet plate of the RTF furnace section;
2.2: to pair
Figure 33941DEST_PATH_IMAGE003
And
Figure 632412DEST_PATH_IMAGE004
the variables in (1) are screened by the common multiple of the minimum sampling frequency, and the common multiple of the minimum sampling frequency is as follows:
Figure 256904DEST_PATH_IMAGE005
wherein k =1,2, [ alpha ], [
Figure 407263DEST_PATH_IMAGE006
]Representation solution
Figure 752925DEST_PATH_IMAGE007
And
Figure 14142DEST_PATH_IMAGE002
common multiple of sampling frequency of all variables; to obtain
Figure 761649DEST_PATH_IMAGE008
Then, to
Figure 958275DEST_PATH_IMAGE001
And
Figure 40500DEST_PATH_IMAGE002
screening for each variable in (1):
Figure 590562DEST_PATH_IMAGE009
Figure 441843DEST_PATH_IMAGE010
and
Figure 970824DEST_PATH_IMAGE011
is the set after screening;
2.3: for those after screening
Figure 415711DEST_PATH_IMAGE010
And
Figure 753152DEST_PATH_IMAGE011
carrying out exploratory analysis on the time sequence distribution characteristics of each variable to obtain multivariate data;
2.4: preprocessing the multi-element data screened by the same sampling frequency, wherein the preprocessing priority is as follows: the running speed is greater than the variable of the bottom equipment layer and is greater than the temperature of the outlet plates of the two furnace sections; the pretreatment is as follows: when the running speed is 0 or negative, the production line is stopped or an abnormality occurs on the site, and all variables in the time interval are not processed.
Further, in the target operation state determination module, the gradual detailed state tracking strategy is to firstly divide the length of the strip steel in the furnace entering plan by the operation speed to obtain an initial linear time interval, and then indicate a plan index layer I 3 The method comprises the steps of lengthening a time sequence, aligning the time sequence with an initial linear time interval, calculating the length of strip steel through the head and tail positions of certain adjacent strip steel detected by a sensor, and comparing with the length of strip steel in a furnace entering plan, wherein the calculated length of strip steel is more accurate because the length of strip steel in the furnace entering plan is an initial value, the strip steel can be slightly or greatly reduced according to production requirements and process requirements when entering a production line, and the reduction length is unknown, so the length of strip steel in the furnace entering plan is not consistent with the length of strip steel actually entering the furnace, and the more accurate length of strip steel can be calculated through the step, but at the moment, the more accurate length of strip steel still exists
Figure 475251DEST_PATH_IMAGE012
The position of the target strip steel is further refined and tracked by an outlet plate temperature detection model of the two furnace sections finally so as to obtain more accurate strip steel length, wherein,
Figure 138314DEST_PATH_IMAGE013
to represent
Figure 945864DEST_PATH_IMAGE014
The corresponding sampling time interval, v, represents the line speed of travel.
Further, in the target operation state determination module, the establishment process of the state detection model of the bottom device layer variable is as follows:
(1) variables for each underlying equipment layer
Figure 227941DEST_PATH_IMAGE015
Is represented by
Figure 788235DEST_PATH_IMAGE016
Three continuous sliding windows are established: mean value calculation windowW m Instantaneous change detection windowW d Sum variance calculation windowW v The window length is respectivelym，nAndv；
(2) computing
Figure 638511DEST_PATH_IMAGE015
Is/are as followsW m Mean sumW d Mean value, expressed asM m AndM d and calculateW v Mean value ofM v The sum variance V is calculated as:
Figure 651466DEST_PATH_IMAGE017
in the formula, k 0 Defining a cumulative sum of start and end events for the first sample point
Figure 609670DEST_PATH_IMAGE018
And
Figure 24471DEST_PATH_IMAGE019
the expression is
Figure 45648DEST_PATH_IMAGE020
Wherein δ is a weight parameter
Figure 686845DEST_PATH_IMAGE021
For the variance threshold, the larger the δ, the current
Figure 435358DEST_PATH_IMAGE018
And
Figure 986556DEST_PATH_IMAGE019
the larger the ratio of the statistical value is, the stronger the accumulative capacity is, otherwise, the smaller the statistical value is, the judgment is made
Figure 427902DEST_PATH_IMAGE018
And
Figure 166182DEST_PATH_IMAGE019
the positions of the transient feature points can be determined.
Further, the specific implementation process of the target operation state determination module is as follows:
3.1: for plan index
Figure 593752DEST_PATH_IMAGE022
Performing initial time interval linear filling, wherein k =1,2 and 3 respectively represent the type specification of the strip steel in a furnace entering plan, the length of the strip steel in the furnace entering plan and the furnace entering sequence of the strip steel in the furnace entering plan; the linear filling of the initial time interval refers to that the length of the strip steel in the furnace entering plan is taken as a reference, the division calculation is carried out on the length of the strip steel and the running speed to obtain the initial linear time interval Z 0 ；
3.2: utilizing the variable data processed in the step 3.1 and the multivariate characteristic acquisition module to formulate a state tracking strategy of progressive detail;
3.3: utilizing the established multivariate state detection model of the bottom equipment layer to carry out detection on the bottom equipment layer I 1 Detecting the variable state, namely determining a position point of state change by determining a transient characteristic point;
3.4: by means of step 3.3, the position of the strip can be further reduced in the spatial domain from the strip length in the furnace entry plan to the calculated strip length L r In the time domain, from Z 0 Is reduced to
Figure 983145DEST_PATH_IMAGE023
In order to more accurately track the position of the target strip steel, the specific position is determined through a detection model of the temperature state of the outlet plates of the two furnace sections, and the establishing process is as follows:
(1) processed in a selected multivariate feature acquisition module
Figure 343195DEST_PATH_IMAGE002
The time sequence data set of any variable is y 1 ,y 2 ,…,y k Calculating the mean value of the data set as the reference value of the data set as
Figure 83618DEST_PATH_IMAGE024
The deviation is calculated as
Figure 314879DEST_PATH_IMAGE025
，i=1,2,…,k；
(2) Calculating the standard deviation of the time sequence data set of any variable by using the following formula, and obtaining the interval boundary +/-m corresponding to the P according to the occurrence probability P of the data in the stable state in all data and a standard normal distribution table 1 σ；m 1 As positive number, determining the distribution of steady state data at + -m centered on the reference value 1 Within the sigma range;
Figure 309511DEST_PATH_IMAGE026
in the above formula, v i For the ith data y in the historical data sequence i A corresponding deviation;
obtaining real-time data sequencesDetermining the real-time state by taking continuous s data as a group; for each group, the specific determination method is as follows: respectively calculating the set of s real-time data y 1 ,y 2 ,…,y s Obtaining s deviations according to the corresponding deviations; averaging the s deviations to obtain a mean value v; if | v $>m 1 Sigma, showing that the group of real-time data belongs to the strip steel switching time point T r 2 Combined with determination in step 3.3
Figure 827080DEST_PATH_IMAGE027
Or
Figure 805531DEST_PATH_IMAGE028
And taking the intersection between the two, and determining the specific time sequence position of the strip head and the strip tail of the strip steel, wherein the corresponding space position is the position of the outlet of the two furnace sections at the moment.
Further, the process of making the state tracking policy is as follows:
1): detecting the head and tail positions of certain adjacent strip steel based on a multi-element state detection model of a bottom equipment layer, and utilizing the distance of a welding seam at a furnace entering position
Figure 699538DEST_PATH_IMAGE029
Calculating the head end position of a certain roll of strip steel
Figure 814255DEST_PATH_IMAGE030
And tail end position
Figure 378092DEST_PATH_IMAGE031
R represents a coil of strip and its length L is calculated at the same time r ，
Figure 827528DEST_PATH_IMAGE032
；
2): second use of Z 0 Screening is carried out if
Figure 273028DEST_PATH_IMAGE033
Then the result is retained, go to 4) operation, if
Figure 491520DEST_PATH_IMAGE034
If yes, indicating data exception, and turning to 3);
3): for description purpose
Figure 367203DEST_PATH_IMAGE035
Detected and calculated
Figure 38356DEST_PATH_IMAGE030
、
Figure 415111DEST_PATH_IMAGE031
And
Figure 504420DEST_PATH_IMAGE036
exceptions, using the next variable of the underlying equipment layer, i.e.
Figure 534693DEST_PATH_IMAGE037
To perform detection and calculation
Figure 709454DEST_PATH_IMAGE030
、
Figure 748954DEST_PATH_IMAGE031
And
Figure 817404DEST_PATH_IMAGE036
again, 2) until i +1=4, if
Figure 766381DEST_PATH_IMAGE038
If the data is still calculated to be abnormal, the strip steel entering the production line is wrong, the wrong information is returned to a system interface for alarming, and the production plan and the strip steel information in the production line need to be manually re-calibrated at the moment;
4): calculated L r Common multiple of minimum sampling frequency
Figure 677705DEST_PATH_IMAGE008
Constraining to obtain a ratio to Z 0 More precise ranges, but still exist
Figure 6050DEST_PATH_IMAGE039
The error in the interval is determined by the error in the interval,
Figure 319219DEST_PATH_IMAGE013
represent
Figure 442027DEST_PATH_IMAGE008
The corresponding sampling time interval is set to be,
Figure 716014DEST_PATH_IMAGE008
and v represents the common multiple of the minimum sampling frequency, and the operating speed of the production line is used for further determining the accurate position of the target strip steel, and the large-amplitude jump position of the NOF section outlet strip steel and the large-amplitude jump position of the RTF section outlet strip steel are respectively detected based on a two-furnace section outlet plate temperature detection model, so that the position of the target strip steel is further finely tracked.
Further, in the target tracking and predicting module, the specific process of establishing a prediction model from the target to the outlet position of each furnace section is as follows:
4.1: updating the initial linear time interval Z 0 Keeping consistent with the switching time interval of the head and the tail of the certain winding steel strip in the step 3.4, setting the switching time interval as
Figure 97317DEST_PATH_IMAGE040
That is, the specific time of the coiled steel completely passing a certain detection point in the furnace is multiplied by the running speed, and the distance length of the coiled steel is calculated
Figure 15725DEST_PATH_IMAGE041
Meanwhile, the specific position of the tape head of the coiled steel tape can be known from the step 3.4
Figure 824281DEST_PATH_IMAGE042
；
4.2: the furnace body information is a fixed numerical value, the spatial position of each key position is a fixed known value, and the spatial position of a certain key position is set as
Figure 926842DEST_PATH_IMAGE043
And then establishing a prediction model from the target to the outlet position of each furnace section as follows:
Figure 987202DEST_PATH_IMAGE044
wherein n represents the specific position of the strip head
Figure 9384DEST_PATH_IMAGE042
To a critical position
Figure 473995DEST_PATH_IMAGE043
The number of changes in running speed during the period, N represents the specific position of the strip head
Figure 847207DEST_PATH_IMAGE042
To a critical position
Figure 321045DEST_PATH_IMAGE043
The maximum number of times during which the operating speed changes,
Figure 932155DEST_PATH_IMAGE045
the operating speed of the nth change is indicated,
Figure 223459DEST_PATH_IMAGE046
is shown in
Figure 569121DEST_PATH_IMAGE045
The distance of the transfer.
The technical scheme provided by the invention has the beneficial effects that: the method saves the labor cost, integrates the sequence characteristics of multiple time-space level parameters, and jointly decides and tracks the real-time position of the target strip steel in the continuous annealing process, thereby establishing a prediction model from the target strip steel to each key position, gradually improving the tracking precision of the target strip steel, and laying a foundation for the subsequent process modeling and system control.
Drawings
The invention will be further described with reference to the following drawings and examples, in which:
FIG. 1 is a schematic diagram of a target tracking system in an embodiment of the invention.
FIG. 2 is a graph of the time-series distribution of underlying device-level variables in an embodiment of the invention.
FIG. 3 is a timing profile of operating speeds in an embodiment of the present invention.
FIG. 4 is a time-series distribution characteristic diagram of the outlet plate temperature of two furnace sections in the embodiment of the invention.
FIG. 5 is a diagram of non-contact temperature detection and contact temperature detection results in an embodiment of the invention.
FIG. 6 is a graph of the time series profile of the outlet plate temperature in an embodiment of the present invention.
FIG. 7 is a graph of NOF plate temperature results for a certain section after pretreatment in an example of the present invention.
FIG. 8 shows an embodiment of the present invention
Figure 95917DEST_PATH_IMAGE035
The time sequence distribution is shown schematically.
Detailed Description
Referring to fig. 1, fig. 1 is a schematic diagram of a target tracking system in an embodiment of the present invention, taking a continuous annealing heating process of an annealing furnace as an example, collecting key variables related to multiple spatial layers and target tracking, where the key variables have different sampling frequencies and different hysteresis on a time scale. Wherein the multi-spatial layer includes: bottom layer of equipment layer I 1 Run Process layer I 2 And a planning index layer I 3 . The classification of the multiple spatial layers is determined according to variable sources and acquisition frequency differences, for example, parameters of a bottom layer equipment layer are parameters of acquiring time unit rules and acquiring frequency in millisecond or second level, an operation process layer is process data acquired by a sensor deployed in a control system, the acquiring frequency is second level, a plan index layer is plan data in a furnace entering plan, and time unit intervals of the data are irregular and are generally in minute or hour level. The target tracking refers to that one or more rolls of continuous multiple rolls of strip steel enter into an annealing production line for being used asAnd performing real-time positioning and tracking for the target. The key variables refer to variables related to target strip steel tracking in initial production equipment and a control system of an annealing production line, and comprise: the type and specification of the strip steel in the furnace entering plan, the length of the strip steel in the furnace entering plan, the furnace entering sequence of the strip steel in the furnace entering plan, the distance of a welding seam at the furnace entering position, a welding seam at the joint of a preheating-heating section (hereinafter referred to as PH-NOF), the distance of a welding seam at the outlet of a direct-fired furnace section (hereinafter referred to as NOF furnace section), the distance of a welding seam at the outlet of a radiation heating furnace section (hereinafter referred to as RTF furnace section), the running speed, the temperature of an outlet plate of the NOF furnace section and the temperature of an outlet plate of the RTF furnace section. The target tracking system specifically includes:
a key variable acquisition module for extracting parameters related to the position of the target strip steel as key variables according to the continuous annealing heating process in the annealing furnace and dividing the key variables into a bottom equipment layer I according to equipment and control types 1 Run Process layer I 2 And a planning index layer I 3 (ii) a The key variables comprise the type specification of the strip steel in the furnace plan, the length of the strip steel in the furnace entering plan, the furnace entering sequence of the strip steel in the furnace entering plan, the distance of a welding seam at the furnace entering position, the welding seam at a PH-NOF combination position, the distance of a welding seam at an outlet of an NOF furnace section, the distance of a welding seam at an outlet of an RTF furnace section, the running speed, the temperature of an outlet plate of the NOF furnace section and the temperature of an outlet plate of the RTF furnace section;
a multi-element feature acquisition module for the bottom equipment layer I 1 And run Process layer I 2 Analyzing the running state of a plurality of related variables, extracting the state characteristics of the related variables, determining various abnormal characteristics existing in combination with the production process of the annealing furnace, and screening the abnormal characteristics based on priority distribution and distribution type difference to obtain multiple characteristics; wherein the plurality of relevant variables comprise a weld distance at a furnace entrance, a weld at a PH-NOF junction, a weld distance at an outlet of an NOF furnace section, a weld distance at an outlet of an RTF furnace section, an operating speed, an outlet plate temperature of the NOF furnace section and an outlet plate temperature of the RTF furnace section. The specific process for obtaining the multivariate characteristics is as follows:
2.1: selecting the underlying device layer I 1 Related variables as sets
Figure 574916DEST_PATH_IMAGE047
And run Process layer I 2 Related variables as sets
Figure 37121DEST_PATH_IMAGE048
Wherein i = the distance of a welding line at a furnace entrance, the welding line at a PH-NOF joint, the distance of a welding line at an outlet of an NOF furnace section or the distance of a welding line at an outlet of an RTF furnace section, and j = the running speed, the temperature of an outlet plate of the NOF furnace section or the temperature of an outlet plate of the RTF furnace section;
2.2: to pair
Figure 853767DEST_PATH_IMAGE003
And
Figure 669408DEST_PATH_IMAGE004
the variables in (1) are screened by common multiple of minimum sampling frequency. Variable data of the bottom device layer
Figure 130476DEST_PATH_IMAGE003
And running process layer variable data
Figure 638949DEST_PATH_IMAGE004
All present the collection regularity, but the collection frequency is different each other, consequently carries out same sampling frequency data processing to data, for guaranteeing the authenticity of data, carries out minimum sampling frequency common multiple screening to all variables, and minimum sampling frequency common multiple is:
Figure 677312DEST_PATH_IMAGE005
wherein k =1,2, [ alpha ], [
Figure 293714DEST_PATH_IMAGE006
]Representation solving
Figure 999501DEST_PATH_IMAGE007
And
Figure 537930DEST_PATH_IMAGE002
common multiple of sampling frequency of all variables in the system(ii) a To obtain
Figure 345480DEST_PATH_IMAGE008
Then, to
Figure 752191DEST_PATH_IMAGE001
And
Figure 328797DEST_PATH_IMAGE002
screening for each variable in (1):
Figure 38127DEST_PATH_IMAGE009
Figure 316661DEST_PATH_IMAGE010
and
Figure 277795DEST_PATH_IMAGE011
is the set after screening; e.g., if the initial set is
Figure 692596DEST_PATH_IMAGE007
The sampling frequency of the weld distance variable at the furnace inlet is 500ms, and the data acquisition data at the sampling frequency is 1,2,3,4,5,6,7 …
Figure 734281DEST_PATH_IMAGE008
Is 1s, then
Figure 500111DEST_PATH_IMAGE010
The variable of the distance between the welding seams at the furnace inlet is 1,3,5 and 7 ….
2.3: for those after screening
Figure 123991DEST_PATH_IMAGE010
And
Figure 675189DEST_PATH_IMAGE011
and carrying out exploratory analysis on the time sequence distribution characteristics of the variables to know that the time sequence distribution characteristics of the variables are different.
Figure 585376DEST_PATH_IMAGE010
The time sequence distribution characteristics of all variables in the set are shown in fig. 2, and a linear increasing characteristic and a large-amplitude jumping characteristic are presented;
Figure 589235DEST_PATH_IMAGE011
the speed time sequence distribution characteristics in the set are as shown in fig. 3, the linear invariance is kept in a time interval, and if the speed time sequence distribution characteristics are changed, the linear invariance is kept continuously after jumping occurs;
Figure 141439DEST_PATH_IMAGE011
the time sequence distribution characteristics of the outlet variables of the two furnace sections in the set are shown in fig. 4, and show slow fluctuation change characteristics on a time scale, that is, a single change trend and a single value are not kept on the time scale, and data between adjacent time sampling points do not change suddenly but change smoothly relatively.
2.4: preprocessing a plurality of variables (the welding line distance at the furnace entrance, the welding line at the PH-NOF combination position, the welding line distance at the NOF furnace section outlet, the welding line distance at the RTF furnace section outlet, the operation speed, the NOF furnace section outlet plate temperature and the RTF furnace section outlet plate temperature) (namely, preprocessing multivariate data screened by the same sampling frequency) of a bottom equipment layer and an operation process layer, wherein the preprocessing priority is as follows: the running speed is greater than the variable of the bottom equipment layer and is greater than the temperature of the outlet plate of the two furnace sections. Variables with different types of distribution characteristics have different anomaly characteristics or require different data preprocessing methods: when the running speed is 0 or negative, the production line is shut down or the field is abnormal, and all variables in the time interval are not processed;
the judgment directions of the variables of the bottom equipment layer and the abnormal characteristics of the two furnace sections are different, and the following explanation is made one by one:
(1) two kinds of change characteristics can appear in bottom equipment layer variable, need judge whether for unusual characteristic: the first is that when a negative value of a certain variable occurs because of a detection problem, the variable is an abnormal characteristic, and time zone data of the variable in the negative value needs to be discarded; the second is that under the normal condition, the change trend of all the variables in the class is linear incrementAnd the change is increased, and at a certain moment, a certain variable does not completely show a linear incremental change suddenly, so that the situation needs to be calculated and judged. The time point when the change of the data point is detected is used as the starting time point t 0 At t 0 First, a linear increasing slope k of a variable is calculated 0 From t 0 Initially, set the step size to
Figure 671778DEST_PATH_IMAGE049
Wherein n =1,2, 3.,
Figure 34757DEST_PATH_IMAGE050
to represent
Figure 509601DEST_PATH_IMAGE008
Corresponding to the sampling time interval, the 4 variables of the bottom layer equipment layer are changed from t 0 The starting data are formed into a four-dimensional feature vector
Figure 347719DEST_PATH_IMAGE051
Wherein
Figure 857198DEST_PATH_IMAGE052
The table shows the weld distance at the furnace,
Figure 515713DEST_PATH_IMAGE053
represents the welding seam distance at the PH-NOF joint,
Figure 494164DEST_PATH_IMAGE054
the weld seam distance at the outlet of the NOF furnace section is shown,
Figure 122591DEST_PATH_IMAGE055
the welding line distance at the outlet of the RTF furnace section is represented, m step length detection points are arranged, and the maximum detection time interval is
Figure 237309DEST_PATH_IMAGE056
In the maximum detection time interval, taking pairwise adjacent calculation for each vector to calculate the slope change of m step-length detection points of each vector
Figure 191359DEST_PATH_IMAGE057
Wherein l is represented in
Figure 781740DEST_PATH_IMAGE056
Every two adjacent distances in the time interval, i =1,2, …, m-1, because the distances of the variable detection points of the bottom layer equipment layer are fixed and are at the same running speed, if the distances are constant, the device layer is in a state of being parallel to the running speed of the device layer
Figure 699012DEST_PATH_IMAGE058
If yes, the change characteristics belong to normal changes and are not abnormal characteristics; if it is
Figure 183083DEST_PATH_IMAGE059
If the slope change value is not consistent with the other three values, the slope change value of the single variable needs to be corrected and is consistent with the other three values; if two of the variable slope change values are inconsistent and the other two are consistent, the two inconsistent variable slope change values need to be corrected to be consistent with the other two; if the two variables are consistent, taking the average value to endow the slope change of all the variables; if the difference is different, all the variable values in the time zone are discarded.
(2) For the temperature of the outlet plate of the NOF furnace section and the temperature of the outlet plate of the RTF furnace section, only a non-contact infrared thermometer can be adopted for detection due to process requirements. Since the non-contact infrared thermometer is greatly affected by environmental factors (temperature of an object, air medium), as shown in fig. 5, (a) is a schematic diagram of the outlet plate temperature of the NOF furnace section (non-contact type), and (b) is a schematic diagram of the outlet plate temperature of the NOF furnace section (contact type), the non-contact infrared thermometer has more noise and outliers than the contact type temperature sensor. If the plate temperature data is not processed, the accuracy of a data model (such as the overfitting problem) cannot be ensured when the data is subsequently used for modeling, so that the outlet plate temperature data needs to be preprocessed according to the characteristics of the outlet plate temperature data.
By analyzing the time sequence distribution characteristics of the two furnace section outlet plate temperatures, the outliers and occasional 'dirty data' of the sequences, which are caused by the fact that the measured data have sudden outliers due to the pollutants and the like existing in the equipment such as the sensor or the heating environment, can be known, and the data can not reflect the actual change in the production process, such as the data represented by 'diamond' in fig. 6. Based on the characteristics, considering that different working conditions need to be identified in actual production, effective local information cannot be lost, and in order to improve the accuracy and efficiency of subsequent number work, data noise needs to be reduced and the signal to noise ratio needs to be improved, so that the change trend needs to be kept, the effective outlier needs to be strengthened, and meanwhile, invalid 'dirty data' needs to be processed smoothly. Considering that the same production state repeatedly appears in the process industrial data, the data contains repeated information and is continuously distributed on a time scale, and the front data and the rear data have correlation on the time dimension, therefore, the similar information is utilized to strengthen the screening treatment of the abnormal characteristics of the outlet plate temperature of the two furnace sections, and the method comprises the following specific steps:
1) two kinds of sliding windows with fixed time length are set, namely a search window for limiting the range of searching related points and a neighborhood window for determining the size of a noise-removing point and a neighborhood of similar points, which are respectively expressed as
Figure 55836DEST_PATH_IMAGE060
，
Figure 992568DEST_PATH_IMAGE061
Ds and Ds are parameters for determining the size of D and D, and can be set according to production state characteristics and algorithm speed;
2) setting a search window and a neighborhood window, wherein the search window takes a time point i as a center, a first neighborhood window takes i as a center, a second neighborhood window slides in the search window, a point j is taken as a similarity measurement point to be calculated, and the similarity between the two points is represented by a weight factor w (i, j):
Figure 369323DEST_PATH_IMAGE062
in the formula (I), the compound is shown in the specification,
Figure 193053DEST_PATH_IMAGE063
the smaller the value is, the more similar the neighborhood point and the target point are, and the larger the weight factor w (i, j) is. Z (i) is a normalization coefficient, and h is a smoothing parameter;
3) and sliding the second neighborhood window in the search window range, traversing all points, and solving the similarity between all points in the search window and a neighborhood taking the target point i as a center, wherein the denoised data u (i) at the time point i in v (t) is as follows:
Figure 488906DEST_PATH_IMAGE064
t refers to all time points traversed by the second neighborhood window in the search window, and when the two points are more similar, the proportion of the point j in the calculation of u (i) is larger;
4) and traversing each data point of a certain time zone of the outlet plate temperatures of the two furnace sections, and denoising and smoothing the dimension data according to the method. Taking data of certain time interval of the temperature of the NOF section outlet plate as an example, preprocessing the data in FIG. 6, and obtaining a result shown in FIG. 7, it can be seen that the enhancement change trend can be kept and outliers can be effectively eliminated.
The running state determining module of the target is used for extracting the plan index layer I in the existing system 3 The variables (generally, in the existing production line, the variables of the plan index layer include the type and the specification of the strip steel in the furnace entering plan, the length of the strip steel in the furnace entering plan and the furnace entering sequence of the strip steel in the furnace entering plan) and the multivariate characteristics obtained in the multivariate characteristic acquisition module, and a gradual and detailed state tracking strategy is formulated. Based on formulated strategy conditions and constraints, a target state detection model is established by fusing the multivariate characteristics, so that the running state of a target is determined; the gradual detailed state tracking strategy is that the length of the strip steel in the furnace entering plan is taken as a reference, the division calculation is carried out on the length of the strip steel and the running speed to obtain an initial linear time interval, and the plan index layer I is indicated 3 The variable of (a) is elongated in time sequence and aligned with the initial linear time interval, and the length of the strip steel is calculated according to the head and tail positions of certain adjacent strip steel detected by a sensor and compared with the strip in a furnace entering planThe calculated strip steel length is more accurate (the strip steel length in the furnace entering plan is an initial value, the strip steel can be slightly or greatly reduced according to the production requirement and the process requirement when entering the production line, the reduction length is unknown, so the strip steel length in the furnace entering plan is generally inconsistent with the strip steel length actually entering the furnace, and the difference is larger under general conditions), and the more accurate strip steel length can be calculated through the step, and at the moment, the strip steel length still exists
Figure 663666DEST_PATH_IMAGE039
And finally, further refining and tracking the position of the target strip steel by an outlet plate temperature detection model of the two furnace sections, wherein,
Figure 906429DEST_PATH_IMAGE013
represent
Figure 115824DEST_PATH_IMAGE014
The corresponding sampling time interval, v, represents the line speed of travel. The specific implementation process of the target operation state determination module is as follows:
3.1: for plan index
Figure 316998DEST_PATH_IMAGE022
And performing initial time interval linear filling, wherein k =1,2 and 3 respectively represent the type specification of the strip steel in the furnace entering plan, the length of the strip steel in the furnace entering plan and the furnace entering sequence of the strip steel in the furnace entering plan. The linear filling of the initial time interval refers to that the length of the strip steel in the furnace entering plan is taken as a reference, the division calculation is carried out on the length of the strip steel and the running speed to obtain the initial linear time interval Z 0 Note that: for individual variables of a certain type of strip
Figure 976125DEST_PATH_IMAGE022
In other words, having a uniform Z 0 ；
3.2: and (4) utilizing the variable data processed in the step 3.1 and the multivariate feature acquisition module to formulate a state tracking strategy with progressive detail. The method specifically comprises the following steps:
step 1: based on bottom layer equipmentThe multi-element state detection model of the layer detects the head and tail positions of certain adjacent strip steel and utilizes the distance of welding seams at the position of entering a furnace
Figure 429103DEST_PATH_IMAGE029
Calculating the head end position of a certain roll of strip steel
Figure 742273DEST_PATH_IMAGE030
And tail end position
Figure 130660DEST_PATH_IMAGE031
R represents a coil of strip and its length L is calculated at the same time r ，
Figure 263701DEST_PATH_IMAGE032
；
Step 2: second use of Z 0 Screening is carried out if
Figure 395736DEST_PATH_IMAGE033
Then the result is retained and the process goes to Step4, if so
Figure 297833DEST_PATH_IMAGE034
If the data is abnormal, the Step goes to Step 3;
step 3: for description purpose
Figure 981756DEST_PATH_IMAGE029
Detected and calculated
Figure 352825DEST_PATH_IMAGE030
、
Figure 272240DEST_PATH_IMAGE031
And
Figure 42225DEST_PATH_IMAGE036
exceptions, using the next variable of the underlying equipment layer, i.e.
Figure 21682DEST_PATH_IMAGE037
New in detection and calculation
Figure 270261DEST_PATH_IMAGE030
、
Figure 478520DEST_PATH_IMAGE031
And
Figure 355209DEST_PATH_IMAGE036
then Step2 is performed again until i +1=4, if
Figure 521879DEST_PATH_IMAGE038
If the data is still calculated to be abnormal, the band steel entering the production line is wrong, the wrong information is returned to a system interface for alarming, and the production plan and the band steel information in the production line need to be manually re-calibrated at the moment;
step 4: calculated L r Common multiple of minimum sampling frequency
Figure 116808DEST_PATH_IMAGE008
Constraining to obtain a ratio to Z 0 More precise ranges, but still exist
Figure 863179DEST_PATH_IMAGE039
And errors in the interval need to be further determined due to high running speed of a production line, the large-amplitude jump position of the NOF section outlet strip steel and the large-amplitude jump position of the RTF section outlet strip steel are respectively detected based on a two-furnace section outlet plate temperature detection model, and the position of the target strip steel is further finely tracked.
3.3: establishing a multi-element state detection model of the bottom equipment layer for the bottom equipment layer
Figure 735320DEST_PATH_IMAGE065
The detection is to detect the large jump trend, that is, to determine the position point of the state change by determining the transient characteristic point. By analysing the time-series distribution characteristics of the variables of the underlying equipment layer, the characteristics of a coil of strip
Figure 56580DEST_PATH_IMAGE066
The variable has a steady line increasing trend, and when the strip steel is switched, a large-amplitude jump process occurs, as shown in FIG. 8
Figure 621029DEST_PATH_IMAGE066
A time sequence distribution diagram illustrating the coiled steel
Figure 951516DEST_PATH_IMAGE067
During which the strip head enters the detection point, i.e. the strip head position, with the strip at
Figure 287950DEST_PATH_IMAGE068
The tail of the strip enters the detection point, namely the tail position of the strip steel, and the distribution of the strip steel in the time sequence interval is
Figure 780112DEST_PATH_IMAGE023
And therefore of each strip coil
Figure 224999DEST_PATH_IMAGE066
A temporally changing characteristic point in time series. The steps of establishing the multivariate state detection model of the bottom equipment layer are as follows:
is provided with
Figure 578752DEST_PATH_IMAGE015
For establishing three consecutive sliding windows, { i (j) }: mean value calculation windowW m Instantaneous change detection windowW d Sum variance calculation windowW v The window lengths are m, n and v, respectively. Computing
Figure 284539DEST_PATH_IMAGE015
Is/are as followsW m Mean sumW d Mean value, expressed asM m AndM d and calculateW v Mean value ofM v Sum varianceVThe calculation formula is as follows:
Figure 963914DEST_PATH_IMAGE069
in the formula, k 0 The first sample point. Defining a cumulative sum of start and end events
Figure 489573DEST_PATH_IMAGE018
And
Figure 909665DEST_PATH_IMAGE019
the expression is:
Figure 610905DEST_PATH_IMAGE020
wherein, delta is a weight parameter,
Figure 179290DEST_PATH_IMAGE021
for variance threshold, δ is larger, at the current
Figure 474136DEST_PATH_IMAGE018
And
Figure 684537DEST_PATH_IMAGE019
the larger the ratio of the statistical value is, the stronger the accumulative capacity is, otherwise, the smaller the statistical value is, the judgment is made
Figure 115650DEST_PATH_IMAGE018
And
Figure 854936DEST_PATH_IMAGE019
the positions of the transient feature points can be determined.
3.4: and (3) a two-furnace-section outlet plate temperature state detection model. The position of the strip has been reduced in the spatial domain from the strip length in the schedule to L in the above steps r In the time domain, from Z 0 Is reduced to
Figure 105920DEST_PATH_IMAGE023
Wherein, when the strip steel is switched, the strip head is switchedDetecting two state change time points
Figure 995378DEST_PATH_IMAGE070
Detecting two state change time points by switching with tail
Figure 530265DEST_PATH_IMAGE068
The time range of the complete length of the coil of strip steel in the furnace is
Figure 719413DEST_PATH_IMAGE023
At the moment, the deviation still exists, in order to more accurately track the position of the target strip steel, the specific position is determined through the two-furnace-section outlet plate temperature state detection model, and the establishment process of the two-furnace-section outlet plate temperature state detection model is as follows:
(1) after being processed by a selected multi-element characteristic acquisition module
Figure 706961DEST_PATH_IMAGE002
The time sequence data set of any variable is y 1 ,y 2 ,…,y k Calculating the mean value of the data set as the reference value of the data set as
Figure 9897DEST_PATH_IMAGE071
The deviation is calculated as
Figure 540236DEST_PATH_IMAGE025
，i=1,2,…,k；
(2) Calculating the standard deviation of the time sequence data set of any variable by using the following formula, and obtaining a section boundary +/-m corresponding to P according to the occurrence probability P of the data in the stable state in all data and a standard normal distribution table 1 σ；m 1 As a positive number, the steady state data is determined to be distributed in + -m centered on the reference value 1 Within a range of σ (m in this example) 1 >3）；
Figure 152483DEST_PATH_IMAGE026
In the above formula, v i For the ith data y in the historical data sequence i A corresponding deviation;
(3) acquiring a real-time data sequence, and judging a real-time state by taking continuous s data as a group; for each group, the specific determination method is as follows: respectively calculating the set of s real-time data y 1 ,y 2 ,…,y s Obtaining s deviations according to the corresponding deviations; averaging the s deviations to obtain a mean value v; if | v |)>m 1 Sigma, indicating that the group of real-time data belongs to the strip steel switching time point
Figure 378059DEST_PATH_IMAGE072
Determined in conjunction with step S3.3
Figure 468374DEST_PATH_IMAGE027
Or
Figure 728586DEST_PATH_IMAGE028
And taking the intersection between the two to obtain the specific time sequence position of the strip head and the strip tail of the strip steel, wherein the corresponding space position at the moment is the position of the outlet of the two furnace sections.
A target tracking prediction module for layering the planning index on a time scale
Figure 387100DEST_PATH_IMAGE073
The variable information (specific strip steel number, type and sequence) in the system and the running state determining module of the target align the running state result of a certain coil of steel in time, and the equipment information of the furnace body is combined on the spatial scale, so that the target which dynamically changes in the furnace body is tracked in real time. Based on the tracking information of the target, a prediction model from the target to key positions such as the outlet of each furnace section is established, the time of entering and exiting each furnace section is predicted, and a foundation is laid for the subsequent process modeling and the system control. The specific process of establishing a prediction model from a target to the outlet position of each furnace section is as follows:
4.1: updating the initial linear time interval Z 0 Keeping consistent with the switching time interval of the head and the tail of the certain winding steel strip in the step 3.4, setting the switching time interval as
Figure 349240DEST_PATH_IMAGE040
That is, the specific time of the coiled steel completely passing a certain detection point in the furnace is multiplied by the running speed, and the distance length of the coiled steel is calculated
Figure 14487DEST_PATH_IMAGE041
Meanwhile, the specific position of the tape head of the tape steel can be known from the step 3.4
Figure 112893DEST_PATH_IMAGE042
；
4.2: the furnace body information is a fixed numerical value, and the spatial position of each key position is a fixed known value. Let a key position be
Figure 817675DEST_PATH_IMAGE043
Then the specific time to pass the key location is predicted as:
Figure 267111DEST_PATH_IMAGE044
wherein n represents the specific position of the strip head
Figure 840174DEST_PATH_IMAGE042
To a critical position
Figure 74978DEST_PATH_IMAGE043
The number of changes in running speed during the period, N represents the specific position of the strip head
Figure 934349DEST_PATH_IMAGE042
To a critical position
Figure 621814DEST_PATH_IMAGE043
The maximum number of changes in the speed of the run during the run, which can be detected by a speed sensor in the production line,
Figure 592044DEST_PATH_IMAGE045
the operating speed of the nth change is indicated,
Figure 412845DEST_PATH_IMAGE046
is shown in
Figure 708697DEST_PATH_IMAGE045
Distance of transmission wherein the condition is satisfied
Figure 8091DEST_PATH_IMAGE074
。
The technical scheme provided by the invention has the beneficial effects that: and sequence features of multiple time-space level parameters are fused, and the real-time position of the target strip steel in the continuous annealing process is jointly decided and tracked, so that a prediction model from the target strip steel to each key position is established, and a foundation is laid for subsequent process modeling and system control.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (7)

1. A target tracking system, characterized by: the method comprises the following steps:
a key variable acquisition module for extracting parameters related to the position of the target strip steel as key variables according to the continuous annealing heating process in the annealing furnace and dividing the key variables into a bottom equipment layer I according to equipment and control types 1 Run Process layer I 2 And plan index layer I 3 (ii) a The key variables comprise the type specification of the strip steel in the furnace plan, the length of the strip steel in the furnace entering plan, the furnace entering sequence of the strip steel in the furnace entering plan, the distance of a welding seam at the furnace entering position, the welding seam at a PH-NOF combination position, the distance of a welding seam at an outlet of an NOF furnace section, the distance of a welding seam at an outlet of an RTF furnace section, the running speed, the temperature of an outlet plate of the NOF furnace section and the temperature of an outlet plate of the RTF furnace section;
a multi-element feature acquisition module for the bottom equipment layer I 1 And run Process layer I 2 Analyzing and extracting the running state of multiple related variablesDetermining various abnormal characteristics by combining the data time sequence distribution characteristics of all variables and the production process of the annealing furnace, and screening the abnormal characteristics based on a priority strategy and the time sequence distribution characteristics to obtain multivariate characteristics; wherein the plurality of related variables comprise a welding line distance at a furnace entrance, a welding line at a PH-NOF joint, a welding line distance at an outlet of an NOF furnace section, a welding line distance at an outlet of an RTF furnace section, an operation speed, an outlet plate temperature of the NOF furnace section and an outlet plate temperature of the RTF furnace section;
an operation state determination module of the target for determining the operation state of the target according to the planned index layer I in the existing system 3 The variable and multivariate feature acquisition module obtains multivariate features and formulates a gradual detailed state tracking strategy; based on the formulated progressive detailed state tracking strategy conditions and constraints, a target state detection model is established by fusing the multivariate features, so that the running state of the target is determined;
a target tracking prediction module for predicting the planning index layer I on a time scale 3 Time alignment is carried out on the operation state result of a certain coil steel in an operation state determining module of the variable and the target, wherein the plan index layer I 3 The variables comprise the type and specification of the strip steel in the furnace entering plan, the length of the strip steel in the furnace entering plan and the furnace entering sequence of the strip steel in the furnace entering plan, and equipment information of a furnace body is combined on a spatial scale, so that a dynamically-changed target in the furnace body is tracked in real time; and establishing a prediction model from the target to the outlet position of each furnace section based on the tracking information of the target so as to predict the time of the target entering and exiting each furnace section.
2. A target tracking system as claimed in claim 1, wherein: the specific process for obtaining the multivariate characteristics in the multivariate characteristic obtaining module is as follows:
2.1: selecting the bottom device layer I 1 Dependent variables as collections
Figure 119259DEST_PATH_IMAGE001
And run Process layer I 2 Related variables as sets
Figure 301979DEST_PATH_IMAGE002
Wherein i = the distance of a welding line at a furnace entrance, the welding line at a PH-NOF joint, the distance of a welding line at an outlet of an NOF furnace section or the distance of a welding line at an outlet of an RTF furnace section, and j = the running speed, the temperature of an outlet plate of the NOF furnace section or the temperature of an outlet plate of the RTF furnace section;
2.2: for is to
Figure 160344DEST_PATH_IMAGE003
And
Figure 617871DEST_PATH_IMAGE004
the variables in (2) are screened by common multiple of minimum sampling frequency, and the common multiple of the minimum sampling frequency is as follows:
Figure 947090DEST_PATH_IMAGE005
wherein k =1,2, [ solution ]
Figure 97448DEST_PATH_IMAGE006
]Representation solving
Figure 630061DEST_PATH_IMAGE007
And
Figure 907589DEST_PATH_IMAGE002
common multiple of sampling frequency of all variables; to obtain
Figure 638785DEST_PATH_IMAGE008
Then, to
Figure 146996DEST_PATH_IMAGE001
And
Figure 229221DEST_PATH_IMAGE002
screening for each variable of (1):
Figure 231812DEST_PATH_IMAGE009
Figure 833826DEST_PATH_IMAGE010
and
Figure 529250DEST_PATH_IMAGE011
is the set after screening;
2.3: for those after screening
Figure 353898DEST_PATH_IMAGE010
And
Figure 691339DEST_PATH_IMAGE011
carrying out exploratory analysis on the time sequence distribution characteristics of the variables to obtain multivariate data;
2.4: preprocessing the multi-element data screened by the same sampling frequency, wherein the preprocessing priority is as follows: the running speed is greater than the variable of the bottom equipment layer and is greater than the temperature of the outlet plates of the two furnace sections; the pretreatment is as follows: when the running speed is 0 or negative, the production line is stopped or an abnormality occurs on the spot, and all variables in the time interval are not processed.
3. A target tracking system as claimed in claim 1, wherein: in the target running state determining module, the gradual and detailed state tracking strategy is to firstly take the length of the strip steel in the furnace entering plan as a reference and carry out division calculation with the running speed to obtain an initial linear time interval, and then plan index layer I 3 The variable is elongated in time sequence and aligned with the initial linear time interval, and the length of the strip steel is calculated through the head and tail positions of the adjacent strip steel detected by the sensorThe strip length in the furnace plan is not consistent with the strip length actually fed into the furnace because the strip length is not known, so that the more accurate strip length in one step can be calculated through the step, but the strip length still exists at the moment
Figure 600389DEST_PATH_IMAGE012
The position of the target strip steel is further refined and tracked by an outlet plate temperature detection model of the two furnace sections finally so as to obtain more accurate strip steel length, wherein,
Figure 14184DEST_PATH_IMAGE013
to represent
Figure 8684DEST_PATH_IMAGE014
The corresponding sampling time interval is set to be,
Figure 664663DEST_PATH_IMAGE014
represents the common multiple of the minimum sampling frequency, and v represents the running speed of the production line.
4. A target tracking system as claimed in claim 1, wherein: in the target running state determining module, the establishment process of the state detection model of the bottom layer equipment layer variable is as follows:
(1) variables for each underlying equipment layer
Figure 224957DEST_PATH_IMAGE015
Is represented by
Figure 809653DEST_PATH_IMAGE016
Three continuous sliding windows are established: mean value calculation windowW m Instantaneous change detection windowW d Sum variance calculation windowW v The window length is respectivelym，nAndv；
(2) calculating out
Figure 25871DEST_PATH_IMAGE015
IsW m Mean sumW d Mean value, expressed asM m AndM d and calculateW v Mean value ofM v The sum variance V is calculated as:
Figure 970693DEST_PATH_IMAGE017
in the formula, k 0 Defining a cumulative sum of start and end events for the first sample point
Figure 900341DEST_PATH_IMAGE018
And
Figure 108468DEST_PATH_IMAGE019
the expression is
Figure 625031DEST_PATH_IMAGE020
Wherein δ is a weight parameter
Figure 311228DEST_PATH_IMAGE021
For the variance threshold, the larger the δ, the current
Figure 846114DEST_PATH_IMAGE018
And
Figure 536727DEST_PATH_IMAGE019
the larger the ratio of the statistical value is, the stronger the accumulative capacity is, otherwise, the smaller the statistical value is, the judgment is made
Figure 727537DEST_PATH_IMAGE018
And
Figure 30474DEST_PATH_IMAGE019
the positions of the transient feature points can be determined.
5. A target tracking system as claimed in claim 4, wherein: the specific implementation process of the target operation state determination module is as follows:
3.1: for plan index
Figure 357550DEST_PATH_IMAGE022
Performing initial time interval linear filling, wherein k =1,2 and 3 respectively represent the type specification of the strip steel in a furnace entering plan, the length of the strip steel in the furnace entering plan and the furnace entering sequence of the strip steel in the furnace entering plan; the linear filling of the initial time interval refers to that the length of the strip steel in the furnace entering plan is taken as a reference, the division calculation is carried out on the length of the strip steel and the running speed to obtain the initial linear time interval Z 0 ；
3.2: utilizing the variable data processed in the step 3.1 and the multivariate characteristic acquisition module to formulate a state tracking strategy of progressive detail;
3.3: utilizing the established multivariate state detection model of the bottom equipment layer to carry out detection on the bottom equipment layer I 1 Detecting the variable state, namely determining a position point of state change by determining a transient characteristic point;
3.4: by means of step 3.3, the position of the strip can be further reduced in the spatial domain from the strip length in the furnace entry plan to the calculated strip length L r In the time domain, from Z 0 Is reduced to
Figure 969797DEST_PATH_IMAGE023
In order to more accurately track the position of the target strip steel, the specific position is determined through a detection model of the temperature state of the outlet plates of the two furnace sections, and the establishing process is as follows:
(1) selecting processed in a multivariate feature acquisition module
Figure 688049DEST_PATH_IMAGE002
Data of two variables, either variableIs y 1 ,y 2 ,…,y k Calculating the mean value of the data set as the reference value of the data set as
Figure 981627DEST_PATH_IMAGE024
The deviation is calculated as
Figure 241838DEST_PATH_IMAGE025
，i=1,2,…,k；
(2) Calculating the standard deviation of the time sequence data set of any variable by using the following formula, and obtaining the interval boundary +/-m corresponding to the P according to the occurrence probability P of the data in the stable state in all data and a standard normal distribution table 1 σ；m 1 As a positive number, the steady state data is determined to be distributed in + -m centered on the reference value 1 Within the sigma range;
Figure 697090DEST_PATH_IMAGE026
in the above formula, v i For the ith data y in the historical data sequence i A corresponding deviation;
acquiring a real-time data sequence, and judging a real-time state by taking continuous s data as a group; for each group, the specific determination method is as follows: respectively calculating the set of s real-time data y 1 ,y 2 ,…,y s Obtaining s deviations according to the corresponding deviations; averaging the s deviations to obtain a mean value v; if | v $> m 1 Sigma, showing that the group of real-time data belongs to the strip steel switching time point T r 2 Combined with the determination in step 3.3
Figure 659230DEST_PATH_IMAGE027
Or
Figure 802504DEST_PATH_IMAGE028
Taking the intersection between the two, the specific time sequence position of the strip head and the strip tail of the strip steel is obtained, and the corresponding space position at the moment is the position of the outlet of the two furnace sectionsA position.
6. A target tracking system as claimed in claim 5, wherein: the process of making the state tracking strategy is as follows:
1): detecting the head and tail positions of certain adjacent strip steel based on a multi-element state detection model of a bottom equipment layer, and utilizing the distance of a welding seam at a furnace entrance
Figure 369752DEST_PATH_IMAGE029
Calculating the head end position of a certain roll of strip steel
Figure 808954DEST_PATH_IMAGE030
And tail end position
Figure 258390DEST_PATH_IMAGE031
R represents a certain coil of strip steel and its length L is calculated at the same time r ，
Figure 893771DEST_PATH_IMAGE032
；
2): second use of Z 0 Screening is carried out if
Figure 361530DEST_PATH_IMAGE033
Then retain the result, go to 4) operation, if
Figure 424164DEST_PATH_IMAGE034
If yes, indicating that the data is abnormal, and turning to 3);
3): for description purpose
Figure 111629DEST_PATH_IMAGE035
Detected and calculated
Figure 347438DEST_PATH_IMAGE030
、
Figure 623698DEST_PATH_IMAGE031
And
Figure 168818DEST_PATH_IMAGE036
exceptions, using the next variable of the underlying equipment layer, i.e.
Figure 530529DEST_PATH_IMAGE037
New in detection and calculation
Figure 320762DEST_PATH_IMAGE030
、
Figure 451529DEST_PATH_IMAGE031
And
Figure 652703DEST_PATH_IMAGE036
again, 2) until i +1=4, if
Figure 819154DEST_PATH_IMAGE038
If the data is still calculated to be abnormal, the band steel entering the production line is wrong, the wrong information is returned to a system interface for alarming, and the production plan and the band steel information in the production line need to be manually re-calibrated at the moment;
4): calculated L r Common multiple of minimum sampling frequency
Figure 147499DEST_PATH_IMAGE008
Constrain, get a ratio to Z 0 More precise ranges, but still exist
Figure 460668DEST_PATH_IMAGE039
The error in the interval is determined by the error in the interval,
Figure 82011DEST_PATH_IMAGE013
to represent
Figure 418315DEST_PATH_IMAGE008
The corresponding sampling time interval is set to be,
Figure 799618DEST_PATH_IMAGE008
and v represents the common multiple of the minimum sampling frequency, the running speed of the production line, and in order to further determine the accurate position of the target strip steel, the large-amplitude jump position of the NOF section outlet strip steel and the large-amplitude jump position of the RTF section outlet strip steel are respectively detected based on the two furnace section outlet plate temperature detection models, so that the position of the target strip steel is further finely tracked.
7. A target tracking system as claimed in claim 6, wherein: in the target tracking and predicting module, the specific process of establishing a prediction model from a target to the outlet position of each furnace section is as follows:
4.1: updating the initial linear time interval Z 0 Keeping consistent with the switching time interval of the head and the tail of the certain winding steel strip in the step 3.4, setting the switching time interval as
Figure 655709DEST_PATH_IMAGE040
That is, the specific time when the coiled steel completely passes a certain detection point in the furnace is multiplied by the running speed, and the distance length of the coiled steel is calculated
Figure 198686DEST_PATH_IMAGE041
Meanwhile, the specific position of the tape head of the tape steel can be known from the step 3.4
Figure 802712DEST_PATH_IMAGE042
；
4.2: the furnace body information is a fixed numerical value, the spatial position of each key position is a fixed known value, and the spatial position of a certain key position is set as
Figure 987705DEST_PATH_IMAGE043
And then establishing a prediction model from the target to the outlet position of each furnace section as follows:
Figure 947571DEST_PATH_IMAGE044
wherein n represents the specific position of the strip head
Figure 677761DEST_PATH_IMAGE042
To a critical position
Figure 785394DEST_PATH_IMAGE043
The number of changes in running speed during the period, N represents the specific position of the strip head
Figure 554505DEST_PATH_IMAGE042
To a critical position
Figure 119609DEST_PATH_IMAGE043
The maximum number of times during which the operating speed changes,
Figure 269968DEST_PATH_IMAGE045
the operating speed of the nth change is indicated,
Figure 944243DEST_PATH_IMAGE046
is shown in
Figure 205460DEST_PATH_IMAGE045
The distance of the transfer.
CN202210831903.9A 2022-07-15 2022-07-15 Target tracking system Active CN114896900B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210831903.9A CN114896900B (en) 2022-07-15 2022-07-15 Target tracking system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210831903.9A CN114896900B (en) 2022-07-15 2022-07-15 Target tracking system

Publications (2)

Publication Number Publication Date
CN114896900A CN114896900A (en) 2022-08-12
CN114896900B true CN114896900B (en) 2022-09-30

Family

ID=82730151

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210831903.9A Active CN114896900B (en) 2022-07-15 2022-07-15 Target tracking system

Country Status (1)

Country Link
CN (1) CN114896900B (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111625934A (en) * 2020-04-30 2020-09-04 中国地质大学（武汉） Multi-mode identification method for annealing heating process based on D-S evidence theory

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4231088A (en) * 1978-10-23 1980-10-28 International Business Machines Corporation Allocating and resolving next virtual pages for input/output
CN100372950C (en) * 2004-05-28 2008-03-05 上海宝信软件股份有限公司 Process for controlling temperature of strip steel
CN102886383B (en) * 2011-07-22 2015-03-04 宝山钢铁股份有限公司 Method for controlling mechanical properties of cold-rolling strip steel in online manner
DE102014017273A1 (en) * 2014-11-18 2016-05-19 Salzgitter Flachstahl Gmbh High strength air hardening multiphase steel with excellent processing properties and method of making a strip of this steel

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111625934A (en) * 2020-04-30 2020-09-04 中国地质大学（武汉） Multi-mode identification method for annealing heating process based on D-S evidence theory

Also Published As

Publication number Publication date
CN114896900A (en) 2022-08-12

Similar Documents

Publication Publication Date Title
CN104809333B (en) Capacity prediction methods and system based on Kalman filter
CN116013087B (en) Traffic flow statistical method based on urban moving vehicle detection
Kuo et al. Multi-sensor integration for on-line tool wear estimation through radial basis function networks and fuzzy neural network
CN107092582A (en) One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence
CN108817103B (en) Steel rolling model steel family layer classification optimization method
CN103722022B (en) Friction coefficient model optimizing system and method in rolling process
CN102081846A (en) Expressway charge data track matching based traffic state recognition method
Zhang et al. A study on the method for cleaning and repairing the probe vehicle data
CN116628616B (en) Data processing method and system for high-power charging energy
CN114329810A (en) Real-time shield tunneling machine working attitude prediction method based on big data
CN105821170A (en) Soft measuring system and method for quality indexes of multielement molten iron of blast furnace
CN104267610B (en) High-precision blast furnace smelting process abnormal data detection and repair method
CN103834758B (en) Blast furnace materiel level real-time detection method with continuous high precision
CN110310491B (en) Short-distance double-node geomagnetic vehicle speed detection system and detection method
CN110956155B (en) Fuzzy clustering identification method for fully mechanized mining face operation procedure based on CO data
CN116957120A (en) Device state history trend anomaly prediction method based on data analysis
CN114896900B (en) Target tracking system
Liu et al. Intelligent Prediction and Real-time Monitoring System for Gas Flow Distribution at the Top of Blast Furnace
CN117943892A (en) Cutter monitoring data prediction optimization method based on time sequence data analysis
CN101780488B (en) Internal thread cold-extrusion processing quality on-line forecasting method
CN106203526A (en) Target behavior pattern online classification method based on multidimensional characteristic
CN115017818B (en) Intelligent prediction method for oxygen content of power plant flue gas based on attention mechanism and multilayer LSTM
CN112418522A (en) Industrial heating furnace steel temperature prediction method based on three-branch integrated prediction model
CN105631231A (en) Method for correcting temperature measured value for hot rolling process
CN112765219B (en) Stream data abnormity detection method for skipping steady region

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant