Summary of the invention:
Technical problem to be solved by this invention provides a kind of method of utilizing data redundancy to improve rolling mill thickness control accuracy.
It is as follows to address the above problem the technical scheme that is adopted:
A kind of method of utilizing data redundancy to improve rolling mill thickness control accuracy, it as THICKNESS CONTROL equipment, and adopts following steps with computer and Thickness measuring instrument:
A. set up the milling train multifunctional tester, milling train multifunctional tester and Thickness measuring instrument, both are installed in the outlet of last milling train, and Thickness measuring instrument and multifunctional tester are installed at interval, and Thickness measuring instrument is near the milling train outlet;
B. described Thickness measuring instrument and multifunctional tester are gathered thickness data jointly, and the gained data are all delivered to basic automatization PLC and handled, and then deliver to process computer machine, and two groups of data are judged;
C. judge the valid data point value of Thickness measuring instrument and multifunctional tester,, illustrate that then Thickness measuring instrument is in abnormal condition if the valid data point value of Thickness measuring instrument is bigger than multifunctional tester significant figure strong point count value;
D. the one group of data that multifunctional tester is recorded are sent into computer, try to achieve the data that mean value and milling train setting model calculate generation in advance and compare, try to achieve thickness difference, and according to learning strategy weighting draws best learning coefficient, utilize this coefficient that setting model is revised;
E. with the control model of amended setting model as next step.
The above-mentioned method of utilizing data redundancy to improve rolling mill thickness control accuracy, between described Thickness measuring instrument and the multifunctional tester at a distance of 1.0~1.5 meters.
Technological progress effect of the present invention is:
The present invention utilizes data redundancy, guaranteed that the function of milling train setting model under the situation that Thickness measuring instrument breaks down normally move, thereby the model adaptation ability is strengthened, improved the self study performance of milling train THICKNESS CONTROL, realized the raising of milling train setting accuracy.Compared with prior art, it can make the milling train self-learning function improve, and has solved because the problem that the thickness self-learning function that single Thickness measuring instrument error in data produces lost efficacy.
The specific embodiment
Production line of rolling mill among the present invention comprises heating furnace TF, vertical miller, horizontal mill R1, R2, F1~F5, intercooling device ICS, laminar flow cooling device CTCS, coiling machine DC1, DC2, Thickness measuring instrument TG, multifunctional tester MG.
In milling train was produced, milling train outlet multifunctional tester was the equipment that is used to measure finished product strip convexity and glacing flatness generally speaking, but it has the function of measuring thickness concurrently.Utilize this characteristics, break down or the real data deviation measured when too big at Thickness measuring instrument, the actual plate tape thickness data that multifunctional tester is measured are as the standby input data of self-learning function.
Thickness measuring instrument and multifunctional tester all are installed to the outlet of last rolling mill, and at a distance of 1.2 meters, and Thickness measuring instrument is in the front of multifunctional tester.Such equipment configuration has determined the measured valid data starting point of Thickness measuring instrument to shift to an earlier date than multi-functional.Because the various data of multifunctional tester have been sent in the process computer, only needing in program increases the statement of judging Thickness measuring instrument data and multifunctional tester size of data, and it is correct just can to detect which kind of data.The method that detects is to judge the valid data point value of Thickness measuring instrument and multifunctional tester, if the valid data point value of Thickness measuring instrument is bigger than multifunctional tester significant figure strong point count value, illustrates that then Thickness measuring instrument is in abnormal condition.
In the control program of computer, priority is set.Under all correct available situation of two kinds of data, preferentially adopt the Thickness measuring instrument data.
Selecting to adopt the data of Thickness measuring instrument according to judged result still is the data of multifunctional tester.
So-called data redundancy is exactly the used data of repeated configuration control system among the present invention.When one group of data generation problem wherein, the data of redundant configuration get involved the work that former data support of also bearing, and reduce the fault time of milling train thus, improve the reliability of milling train control.
Below operating procedure of the present invention is described in further detail:
A. utilize milling train multifunctional tester thickness measuring function to measure.Multifunctional tester is the measurement device that metallurgical SPHC station-service is measured strip convexity and glacing flatness.Main measuring principle is to utilize gamma-rays that steel plate is shone in the section direction from the top down, because the thickness difference of steel plate, therefore the roentgen dose X that passes steel plate is also different, and the dosage that the recipient below steel plate obtains is also different, thereby measures the data of each point on the section of steel plate.According to these data, carry out the data processing by the processor on the multi-functional instrument, thereby obtain strip convexity, the glacing flatness of needs.The RM312 multi-functional instrument of model commonly used such as Britain Radiometrie company, or the like.By multifunctional tester is furtherd investigate discovery, it also implies the data of strip thickness when measuring strip convexity, glacing flatness, therefore, can utilize the redundancy of this thickness data as the strip thickness data of thickness gauge measurement.As for the Thickness measuring instrument in milling train exit, it is the measurement device that is used for measuring strip thickness specially.Main measuring principle is to utilize the x ray that steel plate is shone from bottom to top, because the thickness difference of steel plate, the roentgen dose X that therefore passes steel plate is also different, and the dosage that the recipient on steel plate obtains is also different, thereby measures each dot thickness on the section of steel plate.According to these data, carry out the data processing by the processor on the thickness gauge, thereby obtain the strip thickness of needs.The RM215 thickness gauge of model commonly used such as Britain Radiometrie company, or the like.Both are installed in the outlet of last rolling mill, and Thickness measuring instrument and multifunctional tester are installed at interval, and Thickness measuring instrument is near the milling train outlet.
B. above-mentioned Thickness measuring instrument and the common collection plate tape thickness of multifunctional tester data, the data that each instrument collects at first are stored in separately the processor, by optical fiber two groups of data are sent to basic automatization PLC then.It is MELPLAC that basic automatization PLC adopts model.It mainly acts on is that limit is carried out electrical control according to the setting value that the process automation computer provides, also receive the various data that collection in worksite arrives simultaneously, these data parts are used for showing that a part is sent to process computer machine by the TCP/IP network that twisted-pair feeder constitutes.Process computer machine is the core of hot rolling line automation control, it can calculate the required setting value of basic automatization control appliance according to process modeling, also can learn model simultaneously according to the actual value that basic automatization equipment returns, it is more accurate that model is calculated, and more meets on-site actual situations.Process computer machine adopts ALPHA SERVER, and model is DS20E, certainly, also can adopt other models.
C. the data that Thickness measuring instrument and multi-functional examination instrument are recorded are sent into process computer, judge the valid data point value of Thickness measuring instrument and multifunctional tester, if the significant figure strong point count value of Thickness measuring instrument is bigger than multifunctional tester significant figure strong point count value, illustrate that then Thickness measuring instrument is in abnormal condition.
D. when Thickness measuring instrument was in abnormal condition, milling train setting model self-learning function read in the data that multi-functional instrument records according to this information.The data of reading in are one group of data, and these group data are handled through mathematical method, obtain effective mean value.The data of this value and milling train setting model being calculated in advance generation compare, and obtain a thickness difference, and this difference is exactly that model calculates and the actual rolling deviation that produces.According to learning strategy, when the computation model learning coefficient, not only to consider the deviation of model in this generation, therefore the deviation that is produced before also needing to consider needs basis weight separately to be weighted calculating, draws the optimum thickness learning coefficient.Utilize this coefficient that setting model is revised;
E. with the control model of amended setting model as next step.
Below provide again one with side of the present invention go as control foundation main program sequence.
1 at first issues the data of process computer machine from basic automatization PLC, extracts the thickness data of Thickness measuring instrument generation and the thickness data that multifunctional tester produces.The real data that Thickness measuring instrument and multifunctional tester collect all is kept in the ADH actual database, and (buffer retc) is read into respectively among core buffer th_meter and the mu_meter for dbname, recno to read in function read with data.
Realize in the program:
……….
read(tsmpdb,recno,th_meter,retcl);
read(msmpdb,recno,mu_meter,retc2);
………..
Wherein tsmpdb is a Thickness measuring instrument real data table among the ADH,
Msmpdb is a multifunctional tester real data table among the ADH,
2 data volumes that Thickness measuring instrument is collected send variable i m (11) to, and the data volume that multifunctional tester collects sends variable i m (12) to.
Realize in the program
………….
im(11)=th_meter.data_counter;
im(12)=mu_meter.data_counter;
3 with two kinds of instrument to collect to data volume compare, determine that Thickness measuring instrument has precedence over multifunctional tester, then in the data volume variable that the data assignment of needs is used to the model self study.
Realize in the program
…………
stim=im(11);
th_buffer=th_meter;
if(im(11)>im(12))
{
stim=im(12);
th_buffer=mu_meter;
}
………….
Then, the original program of the milling train production process that continues can realize the redundancy utilization of multifunctional tester data.
By above step, can guarantee that data switch automatically, realized data redundancy milling train setting model self-learning method (punctuate in the statement " ... ... " represent identical or omission) with original program.