FIELD OF THE INVENTION:
The present invention relates to facilitate hot strip rolling of Cold Rolled Non Oriented
(CNRO) grade steel in hot strip rolling mill. In particular the present invention is directed to
develop a hybrid mill set-up model and an integrated mill automation system adapted to
automatically and efficiently calculate draft and speed schedule during hot rolling for CNRO
grade steel in the hot strip rolling mill. The present invention is particularly useful to
facilitate and optimize the operation of hot strip rolling mill.
BACKGROUND OF THE INVENTION:
Strips of different grades, thickness and width are rolled from slab in a hot strip mill. The
conventional Hot Strip Mill comprises 3 Roughing stands and 6 Finishing stands. The first
roughing stand (R0/V0) is a combination horizontal stand and a vertical stand. The other two
roughing stands (R1 and R2) are 4 high horizontal stands. There is a delay table after R2
stand, one coil box and a crop shear at the end of the delay table. There are six numbers of
4 high finishing stands (F1 to F6) and two hydraulic down coilers. A schematic diagram of
such mill is provided in the accompanying figure 1. These mills are manual operated and the
draft and speed schedules for different stands are set by the operator. The values of roll gap
for draft schedule and the values of speed for speed schedule at different stands are
calculated by the operators based on their experience and skill. Often erroneous calculation
disturbs the mass balance of material flow between different stands.
Now the rolling of CRNO grade of steel in finishing stands of HSM often creates difficulty to
operators! While it is easier for the operators to predict a suitable range of values for the
parameters like roll force, torque, power, finishing temperature before entry of the material
to stands, it is very difficult to predict these parameters for CRNO grade of steel as the flow
stress characteristics of the CRNO grade of steel is completely different from other material.
Since parameters are important for draft and speed scheduling and wrong prediction may
lead to violation of mill constraints, the mill operators has operate the mill with very
caution. They operate the mill with slower speed than desired and roll gap values are often
set at higher side than desired. This leads to loss of productivity of mill due to low speed
and also high finishing thickness (more than target finishing thickness value of 2.3 mm).
Higher finishing thickness leads to increase in number of reversing passes during
subsequent cold rolling thereby decreasing cold mill productivity.

Thus there has been always need for a system or method which would be adapted to
automatically and correctly determine parameters like roll force, torque, power, finishing
temperature before entry of the material to stands, so that hot strip rolling of the CRNO
grade of steel can done without slowing the speed and without increasing the roll gap
values.
OBJECTS OF THE INVENTION:
It is thus the basic object of the present invention is to develop a model which would be
adapted to automatically determine optimize calibration parameters of a hot rolling mill and
facilitate the automatic operation of the hot rolling mill based on the said calibration
parameter.
Another object of the present invention is to develop a model which would be adapted to
adapted to automatically determine optimize calibration parameters a hot rolling mill for hot
rolling of CRNO grade of steel.
Yet another object of the present invention is to develop an rolling mil automation system
which would be adapted to operated integrate with the model for determining optimize
calibration parameters hot rolling mill and accordingly operate different hardware of the
rolling mill.
SUMMARY OF THE INVENTION:
Thus according to the basic aspect of the present invention, there is provided a system for
facilitating automatic rolling or coiling of steel strips in a hot rolling mill comprising
PLC system for acquiring process data of the rolling including roll gap, entry temperature of
the steel strips, roll diameter, rolling speed;
VAX system for acquiring primary data including diameter of work rolls, chemical
components of the steel and size of the strips;
process works station operatively connected with the PLC system and the VAX system for
receiving the process data and the primary data and executing a hybrid mill set-up model

by involving the process and the primary data and thereby predicting draft and speed
schedule for rolling;
said draft and speed schedule is operatively transmitted to the rolling mill hardware through
the PLC system for said automatic rolling or coiling of steel strips in a hot rolling mill in
accordance with the generated schedule.
In the present system for facilitating automatic rolling or coiling of steel strips in a hot
rolling mill, the said process work station comprises an operator work station providing an
interface to the operator for displaying the draft and speed schedule and allowing the
operator to acknowledge the forwarding said draft and speed schedule to the rolling mill
hardware. The said process works station comprises MS access database for storing the
process and the primary data and providing the same to the hybrid mill set-up model for
determining the draft and speed schedule.
According to another aspect in the present system, the said hybrid mill set-up model
calculates the strip temperature, roll force, torque and power for determining the draft and
speed schedule.
According to yet another aspect in the present system, the said hybrid mill set-up model
involves artificial neural network for calculating the roll force, wherein the said artificial
neural network executes prediction of roll force by involving Sims' theory and Tselikov's
theory alongwith the PLC system acquired process data and the VAX system acquired
primary data as input parameter.
According to further aspect in the present system, provided for automatic scheduling draft
and speed during hot strip rolling of CRNO grade steel, the said artificial neural network
involves temperature dependent Flow stress as an input parameter for calculating the roll
force.
According to another aspect in the present system provided for automatic scheduling draft
and speed during hot strip rolling of CRNO grade steel, the temperature dependent Flow
stress variations of the CRNO grade steel are temperature range wise isolated, wherein each
of the isolated temperature range provides single state of the Flow stress variation which is
anyone of the increment of the Flow stress variation and decrement of the Flow stress
variation with respect to the temperature.

According to yet another aspect in the present system, the said hybrid mill set-up model
comprises Auto Adaptation module adapted to acquired data from the PLC system after
rolling is completed and trains the artificial network with the measured data of roll force,
temperature and predicted final strip thickness for predicting error free mill calibration
coefficients.
In accordance with another aspect in the present invention there is provided a method for
automatically determining draft and speed schedule in hot strip rolling mill during rolling of
steel strips involving the present system comprising
acquiring the primary data from the VAX system and thereby calculating temperature of the
CNRO grade steel strip, roll force, torque, power, looper tension and motor current before
entry of the CNRO grade steel to different stands of the hot strip rolling mill;
measuring values of roll force, temperature and thickness of strip after the rolling and
training the ANN model of the hybrid mill set-up model for minimizing error between the
calculated and the measured data and thereby generating the modified values;
predicting optimized mill calibration coefficients by using the modified values for the next
coil.
According to another aspect in the said method provided for automatically determining draft
and speed schedule during hot rolling for CNRO grade steel strips comprising
isolating temperature dependent flow stress variations of the CNRO grade steel in separate
temperature ranging zones, wherein each of the temperature ranging zone provides single
state of variation i.e. either increment or decrement in the flow stress with respect to the
variation in the temperature;
differently determine the calibration coefficients separately for each zone.
BRIEF DESCRIPTION OF THE ACCOMPANYING FIGURES:
Figure 1 shows a schematic illustration of hardware associated with a conventional hot strip
rolling mill.

Figure 2 and 3 shows the flow stress characteristics of CRNO grade of steel obtained from
experimentation in dynamic thermo-mechanical simulators (Gleeble-3500) at strain rate of
10 s-1 and 100 s-1 respectively.
Figure 4 the hardware arrangement of a mill automation system adapted to integrate with
hybrid mill setup model in accordance with the present system.
Figure 5 shows the flow chart for the determination of the draft and speed schedule during
hot rolling by involving the Hybrid mill set-up model of the present invention.
Figure 6 shows a schematic diagram showing conceptual representation of hybrid roll force
module which is an essential component of the present Hybrid mill setup model.
Figure 7 shows the hierarchical data groups used in the AAM of the present Hybrid mill
setup model.
Figure 8 shows screenshot of web-portal based hybrid model output installed at the finishing
stand of Hot Strip rolling mill.
Figure 9 and 10 shows results of the Industrial trials carried out in the mill with hot rolling
of CRNO grade of steel from the newly developed hybrid model based system.
DETAILED DESCRIPTION OF THE INVENTION WITH REFERENCE TO THE
ACCOMPANYING FIGURES:
Reference is first invited from the accompanying figure 2 and 3 which shows the flow stress
characteristics of CRNO grade of steel obtained from experimentation in dynamic thermo-
mechanical simulators (Gleeble-3500) at strain rate of 10 s-1 and 100 s-1 respectively. As
stated earlier the CRNO grade steel shows different flow stress characteristics compare to
other material, the same can be verified with the results as shown in the figure 2 and 3. It
is a common perception that if a material is heated to higher temperature, the stress
required to deform it is lower. So, generally flow stress of material decreases with increase
in temperature. However, this is not the case of the CRNO grade of steel. The flow stress
characteristic shows that it varies with change in temperature. It is not only temperature
dependent, but also strain rate dependent.

Reference is next invited from the accompanying figure 4, which shows the hardware
arrangement of the mill automation system for integrating the hybrid mill setup model with
existing hot strip rolling mill. The hardware arrangement comprises Windows based process
work station (PWS) in which the Hybrid mill setup model is loaded and works on continuous
basis. An Operator Work Station (OWS) has also been installed at the operator pulpit in
which web-based portal has been installed for operator interface. This process work station
is connected with two PLC systems using an OPC network. Through this PLC system the
input process data like roll gap settings, entry temperature, roll diameter, rolling speed
settings comes to the PWS. The hardware arrangement also comprises a VAX system for
providing primary data such as diameter of work rolls, chemical composition of steel and
size of strips to the PWS. In PWS, all these input parameters are stored in a MS Access
database. The hybrid mill setup model application reads all these input data from the
database and calculates speed and draft schedules. The output data passes through the PLC
system in similar fashion and the rolling mill is set automatically.
Reference is now invited from the accompanying figure 5, which shows the flow chart for
the determination of the draft and speed schedule during hot rolling by involving the
present Hybrid mill set-up model. The determination of the draft and speed schedule
includes calculation of different deformation parameters like roll force, torque and power
before entry of material to finishing stands.
As shown in the said figure the scheduling process starts with acquiring the primary and
process data such as Input steel grade i.e. chemical composition of steel, strip temperature
at roughing stands of the hot rolling mill, the roll gap, coil box status, target coil size. After
acquiring this primary data the speed schedule is equally reduced in all stands. Next the
strip temperature at the first high finishing stand is calculated by involving Hybrid delay
table thermal module associated with the present Hybrid mill set-up model and then strip
temperature after each stand is calculated by involving Hybrid mill thermal module of the
said mill set-up model. After this Roll force torque and power at each stand is continuously
calculated by using the Hybrid roll force module of the said mill set-up model.
The accompanying figure 6 shows a schematic diagram having conceptual representation of
hybrid roll force module which is an essential component of the present mill setup model.
Two well known theories i.e., Sims' theory and Tselikov's theory taken from literature have
been used in the hybrid model to calculate roll force using mathematical equations. RF1

represents roll force predicted from Sims' theory and RF2 represents roll force predicted
from Tselikov's theory. An Artificial Neural Network (ANN) model algorithm selected as feed-
forward back propagation algorithm with variable learning rate and conjugate gradient
technique of error minimization. These two calculated parameters are considered as inputs
to the ANN model along with 6 other input parameters like carbon component in the steel,
silicon equivalent, temperature of the steel strip, roll gap, roll diameter which affect roll
force as shown in the figure. All the 8 input variables and output roll force have been
normalized to values 0 to 1. There is one hidden layer with 2 nodes in the ANN model. The
model transfer function has been chosen as tansig function (which is mathematically
equivalent to tanh function).
Normalized values of RF1 and RF2 have been functionalized (with a function tanh-1) as
shown in the figure. The ANN output has been again functionalized with same tanh-1
function to predict normalized value of roll force. Generally, initial weights & biases are
chosen as random numbers during training of ANN models. But, in this hybrid model, the
weight between first input node and first hidden node has been chosen as Unity. Similarly,
the weight between second input node and second hidden node has also been chosen as
Unity. Values of two weights connecting hidden nodes to output node is chosen as 0.5. All
other weights and biases of the network have been chosen as zero. The functionalization
and initial weight selection have been made in such a way that when input data would be
passed to the untrained network, then the predicted roll force would be arithmetic average
of roll forces predicted by Sims' theory and that predicted by Tselikov's theory.
In the present scheduling of the draft and speed, a multivariable optimization technique is
used to reduce the target thickness and predicts the roll gap and speed for different stands
considering all the mill constraints within the specified limits.
The factor Si Equivalent (Sieq) used in the model, obtained from literature, is given by the
following equation:

where, [%Si], [%AI], [%Mn] and [%P] are percentage of Si, Al, Mn and P in steel
composition respectively.

The hybrid mill set-up model has an Auto Adaptation module (AAM). It takes input data
from the system and calculates temperature of strip, roll force, torque, power, looper
tension and motor current before entry of material to different stands using Mathematical-
Reg ression-ANN model of the hybrid mill set-up model. The measured values of roll force,
temperature and thickness of strip are recorded after the rolling and the ANN model of the
hybrid mill set-up model is trained with the measured data. The error between the predicted
and measured data are minimized to determine modified values calibration coefficients,
weights and bias of the hybrid model using a self-activated multivariate optimization
algorithm. The modified values are used for prediction of parameters for the next coil. For
this purpose the calibration data has been divided into a hierarchical data groups. The
accompanying figure 7 shows the hierarchical data groups used in the AAM. This shows that
separate ANN trainings are executed for each group consists of a grade, a Silicon
Equivalent, coil width and segment number.
In the present hybrid mill set-up model the typical flow stress characteristics of the CRNO
grade steel is also used as the input parameter to the present roll force module for
optimizing the rolling procedure. In the present scheduling procedure the temperature
dependent flow stress variations are divided into three separate temperature ranging zones,
wherein each of the temperature ranging zone provides single state of variation i.e. either
increment or decrement in the flow stress with respect to the variation in the temperature.
Different flow stress equations have been developed for each of the three zones and the
calibration coefficients have been determined separately for each zone.
Figure-8 shows screenshot of web-portal based hybrid model output installed at the
finishing stand of Hot Strip rolling mill. When the operator presses "Send Model Setting to
PLC", the mill is automatically set with the model predicted Roll Gap (Draft) and Speed
Settings.
Industrial Validation of Model
Industrial trials have been carried out in the mill with hot rolling of CRNO grade of steel
from the newly developed hybrid model based system and the results of the trials are
provided in the accompanying figure 9 and 10.
Figure-9 shows validation of model predicted roll force with actual measured roll force for
2409 coil samples. It has been found that there is a close matching between the prediction

and measured values with Root Mean Square Error (RMSE) of 1.2 MN only. Figure 10 shows
that there is substantial reduction of finished thickness of CRNO grade coils rolled in hybrid
model based system.

We claim:
1. A system for facilitating automatic rolling or coiling of steel strips in a hot rolling mill
comprising
PLC system for acquiring process data of the rolling including roll gap, entry
temperature of the steel strips, roll diameter, rolling speed;
VAX system for acquiring primary data including diameter of work rolls, chemical
components of the steel and size of the strips;
process works station operatively connected with the PLC system and the VAX
system for receiving the process data and. the primary data and executing a hybrid
mill set-up model by involving the process and the primary data and thereby
predicting draft and speed schedule for rolling;
said draft and speed schedule is operatively transmitted to the rolling mill hardware
through the PLC system for said automatic rolling or coiling of steel strips in a hot
rolling mill in accordance with the generated schedule.
2. The system as claimed in claim 1, wherein the said process work station comprises
an operator work station providing an interface to the operator for displaying the
draft and speed schedule and allowing the operator to acknowledge the forwarding
said draft and speed schedule to the rolling mill hardware.
3. The system a claimed in anyone of the claims 1 or 2, wherein the said process works
station comprises MS access database for storing the process and the primary data
and providing the same to the hybrid mill set-up model for determining the draft and
speed schedule.
4. The system as claimed in anyone of the claims 1 to 3, wherein the said hybrid mill
set-up model calculates the strip temperature, roll force, torque and power for
determining the draft and speed schedule.

5. The system as claimed in anyone of the claims 1 to 4, wherein the said hybrid mill
set-up model involves artificial neural network for calculating the roll force, wherein
the said artificial neural network executes prediction of roll force by involving Sims'
theory and Tselikov's theory alongwith the PLC system acquired process data and the
VAX system acquired primary data as input parameter.
6. The system as claimed in anyone of the claims 1 to 5, provided for automatic
scheduling draft and speed during hot strip rolling of CRNO grade steel, the said
artificial neural network involving temperature dependent Flow stress as an input
parameter for calculating the roll force.
7. The system as claimed in claim 6, the temperature dependent Flow stress variations
of the CRNO grade steel are temperature range wise isolated, wherein each of the
isolated temperature range provides single state of the Flow stress variation, which is
anyone of the increment of the Flow stress variation and decrement of the Flow
stress variation with respect to the temperature.
8. The system as claimed in anyone of the claims 1 to 8, wherein the said hybrid mill
set-up model comprises Auto Adaptation module adapted to acquired data from the
PLC system after rolling is completed and trains the artificial network with the
measured data of roll force, temperature and predicted final strip thickness for
predicting error free mill calibration coefficients.
9. A method for automatically determining draft and speed schedule in hot strip rolling
mill during rolling of steel strips involving the system as claimed in anyone of the
claim 1 to 8 comprising
acquiring the primary data from the VAX system and thereby calculating temperature
of the CNRO grade steel strip, roll force, torque, power, looper tension and motor
current before entry of the CNRO grade steel to different stands of the hot strip
rolling mill;

measuring values of roll force, temperature and thickness of strip after the rolling
and training the ANN model of the hybrid mill set-up model for minimizing error
between the calculated and the measured data and thereby generating the modified
values;
predicting optimized mill calibration coefficients by using the modified values for the
next coil.
10. The method as claimed in claim 9 for automatically determining draft and speed
schedule during hot rolling for CNRO grade steel strips comprising
isolating temperature dependent flow stress variations of the CNRO grade steel in
separate temperature ranging zones, wherein each of the temperature ranging zone
provides single state of variation i.e. either increment or decrement in the flow stress
with respect to the variation in the temperature;
differently determine the calibration coefficients separately for each zone.

ABSTRACT

A system and a method provided for facilitating hot strip rolling of steel including Cold
Rolled Non Oriented (CNRO) grade steel in hot strip rolling mill. The present system and
method is adapted to automatically and efficiently calculate draft and speed schedule for hot
rolling of CNRO grade steel in the hot strip rolling mill in accordance with flow stress
characteristics of the steel. The system and the method automatically and correctly
determine the mill calibration parameters like roll force, torque, power, finishing stand
temperature before entry of the steel to the stands and drive the rolling operation without
slowing the speed and without increasing the roll gap values.