You are on page 1of 6

16th IFAC Symposium on Automation in Mining, Mineral

and Metal Processing


August 25-28, 2013. San Diego, California, USA

Concept of Shield-Data-Based Horizon Control


for Longwall Coal Mining Automation
Mirjam Holm*, Stephan Beitler*, Thorsten Arndt*,
Armin Mozar**, Martin Junker**, Christian Bohn*

*Clausthal University of Technology; Institute of Electrical Information Technology,


Leibnizstraße 28, 38678 Clausthal Zellerfeld, Germany
(Tel: +49 5323-722342; e-mail: {holm, beitler, arndt, bohn}@iei.tu-clausthal.de)

** RAG Deutsche Steinkohle


Shamrockring 1, 44623 Herne, Germany
(e-mail:{armin.mozar, martin.junker}@rag.de)

Abstract: Coal is an important fossil fuel and will also be for the next decades. Thus, there have been
several approaches and researches in longwall coal mining automation which are mainly focused on the
shearer as the executing machine. In addition approaches of coal interface detection have to be taken into
account for horizon control to reach autonomous shearer guidance. The shield-data-based horizon control
is an alternative control approach which integrates data from the roof support as input and feedback to
reach an automatic cutting drum height adjustment with the aim to follow the seam trend, generate a
defined face opening and keep it. This article contains the basics and difficulties of the control concept
including a short introduction to the simulation model as well as the scheme and the description of the
shield-data-based horizon control. Some simulation results will be shown, as well as results of the
experimental tests.
Keywords: longwall mining automation, shearer horizon control

1. MOTIVATION 2. LONGWALL MINING


There have been several attempts of coal mining automation There are two longwall mining methods, one using the
during the last decades to improve safety and productivity. shearer the other one using the plow. Today the shearer is the
One important method in underground coal mining is the more important exploitation machine because the majority of
longwall mining, where the main equipment includes the present coal is mined from thick seams. Thus, the work
shields for roof support, the armored face conveyer (AFC) to presented in this paper in focused on the shearer.
remove the cut coal and the shearer or plow as exploitation
machine. The horizon control, to keep the longwall workface 2.1 Longwall workface and commonly used automatics
in the seam, and the shearer positioning are still critical points
in automation and also focus of the researches during the last
Fig. 1 shows the assembly of a longwall workface.
years. This paper presents a new developed control concept
for horizon control, the shield-data-based horizon control
(SDHC). The aim of the SDHC is to achieve an automatic
cutting drum height adjustment for the shearer on the basis of
the seam data delivered by the roof supports to provide an
autonomous mode even if the seam trend changes.
In section 2 a short introduction to the controlled system and
the state of the art in longwall automation is given. Section 3
introduces the workface simulator, the system characteristic,
the difficulties in the system behavior and a description of the
control objective is given. The principle concept of the
SDHC and its structure are presented in section 4, followed Fig. 1. Principle of a longwall workface
by an introduction to the test system as well as some test
results in section 5. This article closes with a conclusion and In the front of the workface is the shearer which is connected
an outlook on future work. to the AFC. The AFC is the runway of the shearer and has
also the task to transport the cut coal out of the workface.
Behind the AFC the shields for the roof support are placed.

978-3-902823-42-7/2013 © IFAC 98 10.3182/20130825-4-US-2038.00074


IFAC MMM 2013
August 25-28, 2013. San Diego, USA

They are connected with a bar to the AFC. Because of the network approaches as described in Fan et al. (2010) and Su
connection of the particular components, the system has to be et al. (2010) or another genetic algorithm as described in Mi
high automated to be efficient. Thus underground longwall et al. (2010) and an Artificial Immune as in Wang et al.
mining already uses high automated equipment for a safe and (2009). Also the outcome of the ARCAP landmark project,
productive work process. the enhanced horizon control (EHC), which is based on a cut
model and a seam model and furthermore uses an inertial
Researches and automation projects often are focused on only navigation system (INS) for 3D shearer positioning
one part of the system, e.g. the shearer. So the subsystems measurement, is a memory cut based method as described in
themselves have already reached a high degree of Reid et al. (2005).
automation. Thus, today it is common that the movement of
the roof supports and the AFC are done automatically by the The developed algorithm of shield-data-based horizon control
shield control system, whereas other processes in the take several work face components into account, the above
workface are still the task of an operator. This causes the mentioned CID methods could be included but are not
effect, that the manual control is only based on the miner’s necessary. The shield-data-based horizon control is not an
knowledge and experiences. Today a monitoring system is extension of the memory cut like most other approaches. That
often used to store and provide information to the operator as means an autonomous cut can be started already in the first
basis for specific decisions and analyses. Thus, the process run without the assistance of an operator, and changes of the
control system at the surface is a commonly used equipment seam trend will be taken into account automatically.
which has the monitoring task of the whole mine. All signals
and information are collected in an integrated data base. 3. CONTROL SYSTEM
Though there are several automation approaches for the The main parts of the control system are the shearer, the AFC
particular components but a completely autonomous longwall and the roof support. In addition to the named workface
workface is not reached up to now. components a monitoring system is essential, which has the
function to store, edit and provide necessary data. The
2.2 State of the art in longwall mining automation workface components are equipped with different sensors to
monitor the considered process.
As mentioned above high automated equipment is already
used in longwall mining but there are still a lot of tasks to be 3.1 Simulation model
managed. Two important topics on the way to an autonomous
longwall workface are the horizon control and the therefore Longwall mining is a complex process. Mathematically
required coal interface detection (CID). Therefore, these two descriptions of the system behavior possibly exist for single
points were also the focus of research in recent years. The sub systems e.g. the adjustment of the cutting drum height,
task to guide the shearer and with that the linked equipment is but a consideration of all components to describe the whole
the focus of the horizon control, because the movement mining process is difficult. Nevertheless the whole system
through the seam is mainly affected by the cutting path of the with the main parts has to be considered completely to reach
shearer. Thus, the cuttings of the shearer need to be in an advanced degree of automation. Thus, the main workface
horizon to follow the seam direction, which is only possible components of the system and their special tasks are modeled
if the seam direction can be identified. This is commonly still in a two dimensional simulation model, corresponding to
the task of an operator who is partially supported by CID Fig. 2. The cutting drum height can be given manual or by
technologies. The researches in the field of coal layer calculation of a control algorithm. The executed cut and the
detection are based on several approaches as vibration resulting positions of the equipment are determined by the
detection e.g. used by Yang et al. (2010) or Liu et al. (2011), simulation model. Therefore several force and torque
motor current monitoring of the cutting drum as in Mowrey equations have to be calculated to achieve an even force
(1992), infrared thermography e.g. as in Ralston et al. (2012), distribution.
camera detection of marker bands e.g. as in Reid et al. (2010)
as well as natural gamma radiation and radar e.g. as in Roof host rock
Bausov et al. (2007). An overview of these approaches is also
given in Beitler et al. (2013a). Not all of these approaches are
commonly used or available, those which are available can
only give an optimal output in specific conditions. Therefore
modern shearers often have a multi sensor set.
Seam
There is only one main topic for the horizon control which
includes the cut automation in recent researches, the memory
cut. The memory cut is a commonly used automation modus
where the operator performs a cut manually. During the run
of the exploitation machine all important parameters are Floor host rock
stored and repeated in the following runs by the shearer. Fig. 2. Structure of the simulation model
Several approaches which are based on neural networks or
other biologically inspired methods to achieve an adaptive The simulation model was build up by employees of the Otto
cut require this taught path. These are for example neural von Guericke University of Magdeburg in cooperation with

99
IFAC MMM 2013
August 25-28, 2013. San Diego, USA

the German mining company RAG to receive an adequate change of the AFC position and thereby a new initial
system behavior. Therefore, the experience of miners is used. situation. Thus a compensation of this trend has to be taken
For an easier consideration of the simulation and test results into account in automation algorithms to avoid a permanent
only the defined seam boundaries and the cut contour of the overdrive in the resulting heights.
lower cutting drum at the ground are displayed.
3.3 Control Objective
3.2 Non controlled system behavior
The main target of the SDHC is to follow the seam trend with
The simulation model can be used to test the control a predefined face opening. The seam trend is given by the
algorithms but also shows the difficulties of the non roof respectively by the inclination of the canopy. Therefore
controlled system in order to the reaction towards a defined the controller attempts to align the AFC and the roof support
cutting contour of the shearer. The resulting floor and roof base in a predefined distance to the roof with an inclination
data as well as equipment specific measurement values of the difference as small as possible oriented to the roof.
different components are calculated by the simulator. Fig. 3
shows a result for a given straight seam where the equipment The height adjustment of the cutting drums is the only way to
was fit in the displayed boundaries. At approximately 9 m the influence the face opening and the inclination of the AFC and
shearer executed one undercut of 0.05 m. After this the lower thus the movement of the longwall workface. Therefore the
cutting drum of the shearer is adjusted at a neutral position height adjustment is the main task for the control algorithm.
which corresponds to the ground line of the armored face Fig 4 shows the fundamental strategy of reaching a defined
conveyer. The straight trend of the roof layer is predefined face opening. How it can be seen in Fig. 4 beside the heights
for this simulation and the upper cutting drum is adjusted in a of the roof support the inclination of the components is
small distance under the roof, so this given boundary did not important to calculate the face opening at the shearer position
change during the simulation because it is assumed that the and to align the equipment parallel to the roof in the
left thin coal layer detaches itself from the rock roof layer. predefined distance.
Additionally it is expected that the roof is not influenced by
disturbances e.g. roof fall.

1
height (m)

-1

-2

5 10 15
face advance (m)
Fig. 4. Required cut trend for an oversized or undersized face
opening
Fig. 3. Roof layer and cut contour at the ground for a
simulation run of the non controlled system with one small
undercut at 9 m The seam gives the environmental conditions for the longwall
workface. For efficiency reasons, the shearer should only cut
coal exactly in between the seam boundaries, so that only
For the above mentioned situation Fig. 3 shows that the cut
coal and no host rock is cut. Thus, the thickness of the seam
contour at the floor attains a downward trend despite of the
is an important information and schedules the predefined face
neutral cut position after a small undercut. This behavior
opening which has to be given to the control algorithm.
illustrates that not only the cut contour of the shearer but also
Beside the seam thickness the exploitation machine in
the actual position of the armored face conveyer is
combination with the other workface components need a
responsible for the further cut trend.
defined minimal face opening to pass the other equipment
Each cut, respectively the workface movement and the without damage. This has also to be taken into account by
resulting equipment position lead to changing initial setting the face opening parameter. Especially in seams
conditions for the next step. The different reaction of the where the minimum cutting height of the shearer is almost
system for upward cutting or downward cutting shows the the same as the thickness of the seam, or bigger, it could be
nonlinear system behavior. Considering the cut contour of the necessary to cut into the floor boundary of the seam to reach
floor as comparable to a step response it is obvious that a the needed headroom.
stable rest position is not reached by the system without a
A cut into the roof boundary should be prevented to avoid
recurring override by an operator or another control
possibly resulting roof fall. Beside this safety aspect cutting
algorithm. The attempt of a height correction always causes a

100
IFAC MMM 2013
August 25-28, 2013. San Diego, USA

into the roof perhaps leads to the loss of the command Another advantage is that no higher intelligence is required
variable of the SDHC. By injuring the roof strata by the on the exploitation machine and thus the SDHC is also
cutting drums the roof would no longer present the natural applicable for the thin seam mining utilizing the plow
trend of the seam. Therefore the avoidance of injuring the technology as described in Beitler et al. (2013b).
roof can be identified as another control objective or
otherwise an intact roof can be declared as a requirement for First the information of the environmental conditions is
the SDHC. acquired. The important data are the inclination of the
equipment components which are, beside the desired value of
Another requirement for the shield-data-based horizon the face opening, the main input of the controller. The
control and therefore it is always assumed that the coal measured data therefore are stored to a data base, and
detaches itself from the ambient rock, if the cut is executed in provided to further analysis and preparation. For calculating
a small distance under the roof boundary. Thus, the shields the seam trend, the inclination of the roof supports but also
are directly connected to the original roof and so the roof the angles of the exploitation machine or the AFC as well as
support gives the feedback for the control process which the face advance are considered. Also the geometrical
indicates if the desired face opening is reached or not. Due to relations of the workface equipment are required for a
the fact, that the roof support is the “detecting sensor”, the working SDHC control loop.
controlled system contains a time delay. This delay occurs
from the fact, that the cut in front of the shield produce a face The SDHC controller is the central component and its output
opening which is reached by the roof support some steps is the specification for the height adjustment of the cutting
later. drums of the shearer. Its inputs are an estimated exploitation
machine position, the parameter settings and the edited shield
4. MODE OF OPERATION data. Furthermore the SDHC controller could be supported by
information from a CID system or other seam information
As mentioned above coal interface detection, horizon control (e.g. offline seam model), but like indicated by the dashed
and thereby an automatic calculated target for the cutting blocks, these information are optional and not necessary for a
drum height are the main objectives of longwall automation. working SDHC. The position of the AFC and thus that one of
Almost all of the above mentioned researches in longwall the exploitation machine is provided by the exploitation
mining automation are focused on the shearer or the roof machine positioning estimator (EMPE). The EMPE
supports separately. The SDHC is a new control approach in calculates the position for the next run based on shield data,
longwall coal mining which utilizes the roof supports as well previous position, cut work and geometrical relations. One
as the mining machine. Therefore the roof supports are used possible principle of a shearer positioning estimation is given
as a source to carry data about the seam as in Junker et al. in Xu et al. (2010).
(2009). These data are the input of the controller and are used
to calculate the new cutting adjustments for the exploitation The SDHC itself is a cascade controller with n subordinate
machine to reach the aim of automatic horizon control. The controllers, which are getting their desired values from the
basic structure of the SDHC control loop is shown in Fig. 5. parameter setting input vector and the previous controller in
the cascade. The actual values are provided by shield data
and the data from EMPE. The number of subordinated
controllers differs regarding to the utilized exploitation
machine (plow or shearer). Each one of these n subordinated
controllers handles one determined intermediate value, takes
all relevant actual values into account and calculates a desired
value as an input for the following controller. In this manner,
one output is calculated successive from numerous input
values. The feedback is given in the next AFC advance by the
new shield and EMPE data.
5. SIMULATION AND TEST RESULTS
Under the rough mining conditions it is necessary to receive a
Fig. 5. Block diagram of the SDHC control loop safe and productive system. To examine the developed
control algorithm several simulation tests were made.
The SDHC requires several input data which have to be Because of the numerous components of the system
stored, analyzed and prepared. Those data have to be additionally several experimental tests were accomplished in
collected from different sources. Also for safety reasons the cooperation with the German mining company RAG to
SDHC has to be accessible for different operators and other examine the controller in interaction with real mining
systems (e.g. for an emergency stop). Therefor the SDHC has equipment. Therefore an AFC and a shearer were part of the
to be implemented in a superordinate control level of the test assembly. The simulation model substituted the seam as
mine monitoring system. By implementing the SDHC in the well as the roof support in these tests. To make the test
monitoring system all stored data could be used to provide all scenario as realistic as possible, the SDHC-controller was not
necessary information. Also the computational performance connected to the simulation model. Controller inputs were
for analyzing and data preparation is not a limiting factor. provided by the monitoring system and the controller outputs

101
IFAC MMM 2013
August 25-28, 2013. San Diego, USA

were only given to the shearer and the monitoring system, not comparison of these two figures shows the condition
to the simulation model. A more detailed description of the dependent system behavior even of the controlled system.
test assembly and the communication structure is given in While the longwall workface follows the seam trend
Holm et al. (2013). Beside the controller functionality also immediately for the downward tending seam in Fig. 7, the
the communication between the different sub systems, the controlled longwall workface shows an obvious time delay
correct data storage as well as possible safety critical aspects for the upwards tending ramp in Fig. 8. For the upward
were regarded. The tests were realized with several seam tending ramp the system reaction is only visible 5 m after the
scenarios. Some results with different conditions are shown seam trend changes. At this point the approximation to the
in the following. The black line at the ground shows the desired value of the face opening starts.
resulting cut contour from the executed cut calculated by the
SDHC in each of the following diagrams. The grey dashed 3
lines indicate the original seam boundaries.
In Fig. 6 a straight seam is illustrated where the height of the 2
longwall workface at the beginning is less than the desired

height (m)
height and a cut series with at least one undercut is needed to
reach the desired face opening. 1

0
2
5 10 15 20
face advance (m)
height (m)

1
Fig. 8. Experimental control result for a ramp tending
upwards
0
A similar behavior can also be seen in Fig. 9 for an anticline
seam situation. While the time delay is visible during the
5 10 15 20 25 30 35 ascending part of the seam, in the falling region there is no
face advance (m) delay. This is probably caused by the inertia of the roof
support which reaction time seems to be longer if the contact
Fig. 6. Experimental control result for a seam condition to the host rock is lost in front of the shield while a falling
where the height is less than the desired face opening seam provokes a contact and a guidance of the roof support
which gives the feedback in the control loop. Nevertheless
The controlled system in Fig. 6 reaches the desired face the principle function of the control algorithm is given and a
opening after about 5 m face advance, so after five cuts by desired behavior can be effected.
assuming a face advance of 1 m per cut. In comparison to the
non controlled system behavior shown in Fig. 3 the 3
controlled system reaches a stable state with a small
overshoot in the transient effect. In this experimental test the
controlled system reaches and keeps the desired face opening. 2
height (m)

1
2

0
height (m)

5 10 15 20 25
face advance (m)
0

Fig. 9. Experimental result for an anticline seam situation


-1
5 10 15 20 25 There was a fixed distance between the upper and the lower
face advance (m) cutting drum of the shearer a bit smaller than the desired face
opening for all shown test scenarios. So the upper cutting
Fig. 7. Experimental control result for a ramp tending drum is led in a small distance to the roof layer. In all these
downwards tests the upper cutting drum did not cross the roof boundary;
in this regard one control aim is reached. The tests were also
Fig. 7 and Fig. 8 illustrate ramp scenarios, one tending focused on the generation of a defined face opening and the
upwards and one tending downwards respectively. The

102
IFAC MMM 2013
August 25-28, 2013. San Diego, USA

stabilization of the controlled system. Also the response of Holm, M., S. Beitler, T. Arndt, A. Mozar, M. Junker and C.
the controlled system to changing seam trends for different Bohn. 2013. Shield-data-based horizon control for
situations is tested. The attained outcomes of the tests show longwall shearer guidange. Accepted for publication.
that all mentioned control objectives are satisfied. 23rd World Mining Congress. Montreal, August 2013.
Junker, M. and A. Mozar. 2009. Verfahren zu einer auto-
6. CONCLUSION matischen Herstellung einer definierten Streböffnung bei
A longwall coal mining workface is a complex system. Strebbetrieben im untertägigen Steinkohlenbergbau
Fundamental analyses based on a simulation model make the (Method for automatically creating a defined face
system behavior of a longwall workface visible. Therefor the opening in longwall coal mining opreations). Deutsches
cut contour and the seam trend are considered. Regarding the Patent- und Markenamt WO002009103307A1.
simulation results for the non-controlled system it is obvious Liu, W., Y. Yan and R. Wang. 2011. Application of Hilbert-
that the system does not reach a stable rest position without a Huang transform and SVM to coal gangue interface
recurring override by an operator or another control concept. detection. Journal of computers Vol. 6, No. 6. 1262-69.
Therefore the SDHC is introduced as an approach for Mi, J., C. Tan, L. Zhang and D. Sun. 2010. Research on the
longwall coal mining horizon control. The control objectives memory cutting path of shearer based on genetic
are the generation of a desired face opening as well as the algorithm. Journal of coal science and engineering Vol.
tracking of the seam trend. Thus the SDHC approach 16, No. 3. 333-336.
integrates the roof support for data acquisition but did not Mowrey G. L. 1992. Horizon control holds key to automation
need the memory cut as almost all other horizon control (part 2). Coal Magazine. January 1992. 47-51.
methods. The schematic of the SDHC control algorithm and Ralston J. C. and A. D. Strange. 2012. Thermal infrared-
test results for different conditions are shown. The based seam tracking for intelligent longwall shearer
experiments have shown promising results referring to the horizon control. Proceedings of the 12th Underground
control aims. The defined face opening is generated and kept, Coal Operators’ Conference. Wollongong, Australia,
the controlled system follows changing seam trends and the Feb. 2012. 78-85.
system behavior is stabilized. The next steps in the Reid, D., P. Henderson, D. Hainsworth, B. Roberts, A.
development of an autonomous horizon control by using Chacko, C. McLeod and R. Mehl. 2005. Interconnection
SDHC is the application in an underground mine. Future of Landmark Compliant Longwall Mining Equipment –
work can be focused on the combination with other Shearer communication and functional specification for
automation algorithms e.g. for coal interface detection. enhanced horizon control. CSIRO Exploration and
Mining Report P2004/6, Version 2.36.
7. ACKNOWLEDGEMENT Reid P. B., M. T. Dunn, D. C. Reid and J. C. Ralston. 2010.
The authors thank the employees of the Otto von Guericke Real-world automation: New capabilities for
University of Magdeburg, Mr. C. Daniel and Mr. E. Woschke underground longwall mining. Proceedings of the
for implementing the workface behavior and constructing the Australasian Conference on Robotics and Automation.
simulation model. Brisbane, Australia, December 2010.
Su, X., W. Li, L. Zhang, Y. Wang, Q. Fan, C. Sun and L. Yu.
REFERENCES 2010. Application of RBF hierarchical neural network in
automatic horizon control system of memory-cutting
Bausov I. Y., G. L. Stolarczyk, L. G. Stolarczyk and S.
shearer. Proceedings of the Chinese Control and
Koppenjan. 2007. Look-ahead radar and horizon sensing
Decision Conference. Xuzhou, China, May 2010. 2015–
for coal cutting drums. Proceedings of the 4th
18.
International Workshop on Advanced Ground
Wang Z. and Z. Xu. 2009. Research on the technology of
Penetrating Radar. Naples, Italy, June 2007. 208-211.
Shearer self-adaptive memory cutting. Proceedings of
Beitler, S., M. Holm, A. Mozar, M. Junker and C. Bohn.
the 2nd IEEE International Conference on Intelligent
2013a. State of the art in underground coal mining
Computation Technology and Automation. Changsha,
automation and introduction of a new Shield-Data-Based
Hunan, China, October 2009. 920-923.
Horizon Control approach. Accepted for publication. 13th
Xu Z. and Z. Wang. 2010. Research on key technology of
International Multidisciplinary Scientific Geo
shearer position. Proceedings of the IEEE International
Conference. Albena, Bulgaria, June 2013.
Conference of Intelligent System Design an Engineering
Beitler, S., M. Holm, T. Arndt, A. Mozar, M. Junker and C.
Application. Changsha, Hunan, China, October 2010.
Bohn. 2013b. A shield-data-based horizon control
280-283.
approach for thin seam coal mining utilizing plow
Yang W. and S. Wang. 2010. Coal rock interface sensing
technology. Accepted for publication. 30th International
technology applied for horizon control of mining
Symposium on Automation and Robotics in Construction
shearers. Proceedings of the IEEE International
and Mining. Montreal, August 2013.
Conference on Electrical and Control Engineering.
Fan, Q., W. Li, Y. Wang, L. Zou, X. Su, X. Yang and G. Ye.
Dhaka, Bangladesh, December 2010. 587-590.
2010. Shearer memory cutting strategy research basing
on GRNN. Proceedings of International Conference on
Computer Application and System Modeling. Taiyuan,
China, October 2010. V1-96 - V1-99.

103

You might also like