Designing multi agent based linked state machine

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
__________________________________________________________________________________________
Volume: 02 Issue: 07 | Jul-2013, Available @ http://www.ijret.org 90
DESIGNING MULTI-AGENT BASED LINKED STATE MACHINE
Hyacinth C. Inyiama1
, Ifeyinwa Obiora-Dimson2
, Christiana C. Okezie3
1, 2, 3
Department of Electronic and Computer Engineering Nnamdi Azikiwe University, Awka
Anambra State, Nigeria. Phone: +2348032638315
hcinyiama@gmail.com, ifeyinwa29@yahoo.com, christianaobioma@yahoo.com
Abstract
Every industrial process control algorithm can be represented as an ASM chart and every ASM chart can be implemented using the
state agent based approach. This paper describes a simplified approach to implementing complex process control systems using the
agent-based paradigm. It shows that when the system is complex, it can be broken up into a series of ASM charts, and each of those
ASM charts can be implemented using agent-based approach comprising of one process agent and a number of state agents. These
agent based systems together perform the complex industrial process control. This structure can be interconnected to form a linked
state machine in order to realize a multi-agent based linked state structure. This is a ‘divide and conquer’ approach that is much
simpler than tackling the control system implementation as one very complex effort.
Index Terms: Process control, ASM chart, State Transition Table (STT), Multi-agent, Link state machines, Beverage
blending
---------------------------------------------------------------------***---------------------------------------------------------------------
1. INTRODUCTION
There are several ways of tackling the complexity associated
with complex process control systems. One of such methods is
through the use of linked state machines as showcased by
Akpado et al (2010) and also by Inyiama et al (2011). The
former suggested a way to handle the automation of three
individual state machines of an industrial process control
separately and then linked them together to function as a
complex system. The later in their approach showed how an
ASM chart of a complex control system can be simplified by
breaking up the entire control logic into smaller subsystems or
state machines. The individual state machines were
implemented using a ROM and then linked up together to
perform the process control. The use of ASM chart and the
corresponding State Transition Table (STT) approach
enforced a disciplined approach to the realization of the
system.
Inyiama et al (2012) also described the agent based
architecture for process control system. The research showed
how each state in an ASM chart is assigned to an agent. Hence
for an ASM chart with n number of states there also exists n
numbers of state agents. A process agent was also defined to
handle the complexities that may arise in carrying out the
process control by the multi-agents. The agent classes where
defined to comprise of classes 0, 1, 2, 3 based on the number
of input qualifiers between the states. Furthermore the agent
software was defined and the complex system was made to fit
properly into a standard microcontroller. Where the
input/output line is more than 1 byte i.e. 8, line reduction is
achieved through input multiplexing and output decoding as
demonstrated by Inyiama et al (in print)
This paper extends the technique of linked state machines; by
making each machine in the link an agent based system. I/O
line reduction technique was used where necessary to fit the
system into a microcontroller for individual agent based state
machines. Then the agent based state machines are linked up
as one complex process controller with easily implemented
components.
2. INDUSTRIAL PROCESS EXAMPLE
A certain industrial beverage blender shall be used to
showcase the linked state agent based control system design.
The structure of the beverage blender comprises of three main
parts namely the upper tanks, the lower tanks and the mixers.
The upper tanks (fig 1) are where the raw beverages needed
for the blending process are poured into otherwise referred to
as the reservoir.

__________________________________________________________________________________________
Fig-1: Upper tanks of an industrial beverage blender
Fig-2: Column of lower tanks of an industrial beverage blender
Fig 3: The mixer of the beverage blender
Here four different raw beverage types is required hence the
use of four upper tanks. Each of these tanks has an outlet
which allows flow to a group of 4 lower tanks arranged in
columns as shown in figure 2. Each lower tank has 2 valves
which control inlet from the upper tank and outlet to the mixer
(fig 3). Hence the required ratio of beverage type 1, 2, 3, and 4
are let down through their respective valves automatically
from each of the tanks to the mixer. Then this quantity of
beverage is blended at the mixing chamber and then extruded
through the tap as shown in fig 3.
A composite structure of this industrial blender is as shown in
fig 4. The four boxes above in the diagram represent four
upper tanks whose functions have been described in fig 1.
Each column of four boxes represents the lower beverage
tanks used for the blending processes. They serve the purpose
of taking a known quantity of each beverage type before they
are used in blending processes. The four boxes in a column are
fed from the four upper tanks respectively. Thus they are used
to keep track of the quantity of each beverage type used at that
column’s blending station. The four columns of lower tanks
are identical. In each column, the first tank is connected to
upper tank 1, the second tank to upper tank 2, the third tank to
upper tank 3, the fourth tank to upper tank 4.
Each of the lower tanks has two valves labeled V1 and V2
respectively. V1 in each case controls the refill of the lower
tank from the corresponding upper tank, while V2 controls the
quantity of each lower tank that is involved in blending at the
mixer. Thus, the main difference between one column of
beverage box and another is the blending ratios. The blending
V2 V2 V2
Mixing chamber
V2
Tap
Upper
tank for
beverage
type 1
SM1
SL1
Upper
tank for
beverage
type 2
SM2
SL2
Upper
tank for
beverage
type 3
SM3
SL3
Upper
tank for
beverage
type 4
SM4
SL4
Pipe leading to
lower tanks
V1
Pipe from upper tank 4
Pipe leading to the mixer
Column of 4 lower tanks
beverage
1
beverage
2
beverage
3
beverage
4
V1 V1 V1
Beverage1
Beverage2
Beverage3
Beverage4

__________________________________________________________________________________________
ratio used for column 1 differ from that of column 2 and so on.
Therefore mixers 1, 2, 3, 4 produce different blends of the
beverage in upper tanks 1, 2, 3, 4. The number of times each
of the four lower tanks in a column is refilled is tracked so as
to quantify the volume of each beverage type used at mixer 1.
Each of the lower tanks has a fixed volume, 10 litres say.
Therefore by multiplying the number of lower tank refills for
each of the beverage types with the volume for that, the
volume of each beverage type used at mixer 1 is determined.
The same is true of volume of lower tank leading to 2, 3, 4)
respectively. The difference in the outputs of mixers 1 through
4 is dependent on the blending ratio used for each column of
lower tanks.
Note that the output from each mixer stations can also be
determined using the number of runs and blending ratios for
each mixer station. Thus, the system is able to track both the
volume of input and volume of output. This makes it possible
for management to determine any pilfering of input (raw
material) or output (finished product) done by the work force
during the shift.
Fig 4: Automated blender/mixer for beverages and fruit juices
3. SYSTEM DESIGN
Realization of the beverage blending operation described in
figure 4 involves three processes namely:
a. Upper tank control: co-ordinates the filling of the
upper tank to the adequate level and indicating to the
operators when it is low and requires refilling.
b. The beverage level sensing and refill control: ensures
proper functioning of the lower tanks, the valves
controlling inflow and outflow, and proper
Upper
tank for
beverage
type 1
SM1
SL1
Upper
tank for
beverage
type 2
SM2
SL2
Upper
tank for
beverage
type 3
SM3
SL3
Upper
tank for
beverage
type 4
SM4
SL4
beverage
type 1
S11
S12
beverage
type 2
S21
S22
beverage
type 3
S31
S32
beverage
type 4
S41
S42
beverage
1
color
beverage
2
beverage
3
beverage
4
beverage
1
beverage
2
beverage
3
beverage
4
beverage
1
beverage
2
beverage
3
beverage
4
Mixer
Extruder
V2
V2
V2
V2 V2
V2
V2
V2 V2
V2
V2
V2
V2
V2
V2
V2
V1
V1
V1
V1V1
V1
V1
V1V1V1
V1
V1
V1 V1
V1
V1
1 2 3 4
Flav-
our

__________________________________________________________________________________________
dispensing of desired volumes of beverages into the
mixing chamber.
c. The beverage blending operation: cause appropriate
ratios of pure beverages to flow from the lower tanks
into its mixing chambers, blends and extrudes same.
These three processes each can be handled as one agent based
state machine which would then be linked up to function as a
composite machine using the following agent based linked
state design steps:
• Develop an ASM chart from a given industrial process
problem
• Break down the ASM into simpler state machines to
manage the complexity
• Develop the STT for each state machine
• Modify their state transition table to accommodate
input multiplexing where appropriate and to produce a
fully expanded STT that can be indexed with the state
code
• Assign a process agent for each state machine and a
state agent for each of its states.
• Define agent software
• Fit the design of each state machine into a
microprocessor or microcontroller
• Apply further output line reduction where necessary
using output decoding if necessary
• Then link the state machines together as multi-agent
based linked state machines.
4. UPPER TANK CONTROL
The control algorithm for the 4 upper beverage tanks is shown
in the ASM chart of figure 5 together with the state map and
state assignment. It is designed that a signal STOPRUN be
sent to the beverage blending positions whenever any of the
four upper tanks is low on stock. A STOPALARM is also
issued when this happens. The workforce responsible for
feeding the upper tanks with pure beverages is alerted when
any of the tanks is midway full to enable them process more
fruits for that tank before it runs out of stock.
The description of the ASM is as follows:
• At state ST0, the control system checks if beverage tank
1 is low (BVTNK1L). If so, it triggers a STOPALARM
and generates a STOPRUN signal at state ST1. It
remains in that loop containing ST1 until the situation
is remedied.
Again if at ST0, the beverage tank 1 is not low, the system
checks if the stock has run down to half tank (BVTNK1M). If
BVTNK1M is not yet at the middle position the control
system transits to state ST2. If however the beverage tank 1 is
at the middle position, the control signal triggers an alarm to
alert those processing the fruit so that they know that the tank
would run out of stock if more fruit juice is not put there soon
enough. The control system then transits to state ST2.
2 is low on stock (BVTNK2L) if so, it triggers a
STOPALRM and generates a STOPRUN at state ST3.
It remains in that loop containing ST3 until the situation
is remedied.
Again if at ST2 the beverage tank 2 is not low, the system
checks if the stock has rundown to half/middle tank
(BVTNK2M). If BVTNK2M is not yet at the middle position,
the control system transits to state ST3. If however the
beverage tank 2 is at the middle position, the control signal
triggers an alarm to alert those processing the fruit so that they
know that the tank would run out stock. The control system
then transits to the state ST4.
• At ST4 the control system checks if beverage tank 3 is
low on stock (BVTNK3L). If so, it triggers a
STOPALARM and generates a STOPRUN at state ST5.
It remains in that loop containing ST5 until that
situation is remedied.
Again, if at ST4, the beverage tank 3 is not low, the system
checks if the stock has rundown to middle of tank
(BVTNK3M). If BVTNK3M is not yet at the middle position,
the control system transits to state ST6. If however the
beverage tank 3 is at the middle position, the control signal
triggers an alarm to alert those processing the fruit so they
know the tank would soon run out of stock. The control
system then transits to state ST6.
4 is low on stock (BVTNK4L) if so, it triggers a
STOPALARM and generates a STOPRUN at state ST7.
It remains in that loop containing ST7 until the situation
is remedied. If at ST6, the beverage tank 4 is not low,
the system checks if the stock has rundown to middle of
tank (BVTNK4M). If BVTNK4M is not yet at the
middle position, the control system transits to state ST0.
If however the beverage tank 4 is at the middle
position, the control signal triggers an alarm to alert
those processing the fruit so that they know the tank
would run out of stock soon. The control system then
transits to state ST0 and the entire cycle is repeated.

__________________________________________________________________________________________
Fig-5: ASM Chart for upper tank control
The state transition table corresponding to the ASM chart of
figure 5 is shown in table 1. Here there are 11 input lines
comprised of eight qualifiers (BVTNK1L, BVTNK1M,
BVTNK2L, BVTNK2M, BVTNK3L, BVTNK3M,
BVTNK4L, BVTNK4M), and three present state codes bits
(C, B, A). For a fully expanded state transition table, there
would be 211=2048 address lines to decode. The diagram of
fig 6 shows how input multiplexing can be used to reduce the
direct qualifier input to two. In this implementation, the four
low level sensor signals are multiplexed such that one is
selected at a time. This permits the presentation to the control
system of one low level sensor signal and corresponding
midpoint sensor signal at one and the same time is required by
many of the link paths of figure 5 (the ASM chart). Another
advantage of fig 6 is that the number of address lines is now 5
instead of 11 and the control pattern will now comprise 25=32
rows. Thus, the clever application of multiplexing has
significantly reduced the ROM memory demand of the
application.
1
1
0
0
0
0
0
1
1
0
1
0
0
1
1
1
ST5
ST7
BVTNK1M
BVTNK1L
BVTNK2M
BVTNK2L
BVTNK3M
BVTNK3L
BVTNK4M
BVTNK4L
000
001
010
011
101
100
110
111
STO
STOPALARM
STOP RUN
ST1
ST2
STOPALARM
MDALARMSTOP RUN
ST3
ST4
STOPALARM
STOP RUN
MDALARM
ST6
STOP RUN
STOPALARM
MDALARM
ST0
ST2
0
ST3
ST1
1
00
01
CB
A
State Map
ST4
ST6 ST7
ST510
11
STO 000
State State
ST1 001
ST2 010
ST3 011
State assignment
ST4 100
ST5 101
ST6 110
ST7 111
MDALARM
CBA

__________________________________________________________________________________________
Table-1: STT for upper tank control
Link
Path
Qualifiers Present state Next state State
output
Conditiona
l output
BVTNK1L
BVTNK1M
BVTNK2L
BVTNK2M
BVTNK3L
BVTNK3M
BVTNK4L
BVTNK4M
Name
Code
Name
Code
STOPRUN
STOPALARM
MDALARM
L1 0 0 - - - - - - ST0 000 ST2 010 0 0 0
L2 0 1 - - - - - - ST0 000 ST2 010 0 0 1
L3 1 - - - - - - - ST0 000 ST1 001 0 1 0
L4 0 0 - - - - - - ST1 001 ST2 010 1 0 0
L5 0 1 - - - - - - ST1 001 ST2 010 1 0 1
L6 1 - - - - - - - ST1 001 ST1 001 1 1 0
L7 - - 0 0 - - - - ST2 010 ST4 100 0 0 0
L8 - - 0 1 - - - - ST2 010 ST4 100 0 0 1
L9 - - 1 - - - - - ST2 010 ST3 011 0 1 0
L10 - - 0 0 - - - - ST3 011 ST4 100 1 0 0
L11 - - 0 1 - - - - ST3 011 ST4 100 1 0 1
L12 - - 1 - - - - - ST3 011 ST3 011 1 1 0
L13 - - - - 0 0 - - ST4 100 ST6 110 0 0 0
L14 - - - - 0 1 - - ST4 100 ST6 110 0 0 1
L15 - - - - 1 - - - ST4 100 ST5 101 0 1 0
L16 - - - - 0 0 - - ST5 101 ST6 110 1 0 0
L17 - - - - 0 1 - - ST5 101 ST6 110 1 0 1
L18 - - - - 1 - - - ST5 101 ST5 101 1 1 0
L19 - - - - - - 0 0 ST6 110 ST0 000 0 0 0
L20 - - - - - - 0 1 ST6 110 ST0 000 0 0 1
L21 - - - - - - 1 - ST6 110 ST7 111 0 1 0
L22 - - - - - - 0 0 ST7 111 ST0 000 1 0 0
L23 - - - - - - 0 1 ST7 111 ST0 000 1 0 1
L24 - - - - - - 1 - ST7 111 ST7 111 1 1 0
The fully expanded state transition table corresponding to
figure 6 is shown in table 2. The foregoing shows that input
multiplexing and output decoding can be applied to reduce the
complexity of any sequential control system but the exact
approach would differ depending upon the exact demand of
each application. Two of such approaches have been
demonstrated namely:
Using only one multiplexer at the input if only one qualifier is
required at a time. (fig 8) And using more than one
multiplexer (fig 6).
This makes the approach quite universal as it can be adapted
to match the needs of any particular control application.

__________________________________________________________________________________________
Fig-6: ROM based upper tank control using more than one input multiplexer
Table 2: STT for the optimized ROM solution
Link Path Qualifiers Present state Next state State
output
Conditional
output
BVTNKXL
BVTNKXM
Name
Code
Name
Code
STOPRUN
STOPALAR
M
MDALARM
L1 0 0 ST0 000 ST2 010 0 0 0
L2 0 1 ST0 000 ST2 010 0 0 1
L3
L3
1
1
0
1
ST0
ST0
000
000
ST1
ST1
001
001
0
0
1
1
0
0
L4 0 0 ST1 001 ST2 010 1 0 0
L5 0 1 ST1 001 ST2 010 1 0 1
L6
L6
1
1
0
1
ST1
ST1
001
001
ST1
ST1
001
001
1
1
1
1
0
0
L7 0 0 ST2 010 ST4 100 0 0 0
L8 0 1 ST2 010 ST4 100 0 0 1
L9
L9
1
1
0
1
ST2
ST2
010
010
ST3
ST3
011
011
0
0
1
1
0
0
L10 0 0 ST3 011 ST4 100 1 0 0
L11 0 1 ST3 011 ST4 110 1 0 1
L12
L12
1
1
0
1
ST3
ST3
011
011
ST3
ST3
011
011
1
1
1
1
0
0
L13 0 0 ST4 100 ST6 101 0 0 0
L14 0 1 ST4 100 ST6 110 0 0 1
L15
L15
1
1
0
1
ST4
ST4
100
100
ST5
ST5
101
101
0
0
1
1
0
0
L16 0 0 ST5 101 ST6 110 1 0 0
L17 0 1 ST5 101 ST6 110 1 0 1
L18 1 0 ST5 101 ST5 101 1 1 0
C B
STOPALARM
FFC
FFB
FFA
R CLK
Square
Wave
generator
Mono-
stable
00 1
01 out of
10 4
11 muxBVTNK4M
BVTNK3M
BVTNK2M
BVTNK1M
EPROM
STOPRUN
MDALARM
BVTNXM
BVTNXL
C B A
BVTNK3L
00 1
01 out of
10 4
11 mux
BVTNK1L
BVTNK2L
BVTNK4L

__________________________________________________________________________________________
5. BEVERAGE LEVEL SENSING AND REFILL
CONTROL SYSTEM
The ASM chart for the lower tank beverage level sensing and
refill control is shown in fig 7 along with its state map and
state assignment. Before any production run, the control
system of fig. 7 must check to see that the upper tank is not
low. If low i.e. STOPRUN=1 it stops run and waits (i.e.
WAITING) for upper tank refill. After the refill i.e.
STOPRUN=0 the lower beverage tank levels (i.e. BVL) are
sensed by the signal BVL1, BVL2, BVL3, BVL4 for beverage
tanks 1 through 4 respectively. The tank is low on beverage
when any of the signals BVL1 through BVL4 is logic 1 else
the tank beverage level is high when logic 0. When any of the
signals BVL1 BVL2, BVL3, BVL4 is low (meaning low on
stock), the corresponding valve for that tank (HBV1, HBV2,
HBV3, or HBV4) is turned on to let down the beverage from
their corresponding upper tanks. This takes place at states
ST1, ST3, ST5 and ST7 respectively. Meanwhile any time any
of the valves (HBV1, HBV2, HBV3, HBV4) is turned on to
let down beverage, from the upper tank, the level of the tank is
monitored by the signals BVH1, BVH2, BVH3, BVH4
respectively to ascertain when it is filled. When BVHX=0
(where X= 1, 2, 3, 4) it means the beverage tank is not yet full
hence it remains in the FBVX. If BVHX=1, it means that the
lower tank in question is full
L18 1 1 ST5 101 ST5 101 1 1 0
L19 0 0 ST6 110 ST0 000 0 0 0
L20 0 1 ST6 110 ST0 000 0 0 1
L21
L21
1
1
0
1
ST6
ST6
110
110
ST7
ST7
111
111
0
0
1
1
0
0
L22 0 0 ST7 100 ST0 000 1 0 0
L23 0 1 ST7 100 ST0 000 1 0 1
L24
L24
1
1
0
1
ST7
ST7
100
100
ST7
ST7
100
100
1
1
1
1
0
0

__________________________________________________________________________________________
Fig -7: ASM chart for the lower tank level sensing and refill control
Note that this refill control system is applicable to any of the
four columns of lower beverage tanks. Thus when one
column’s beverage tank sensing and refill control system is
realized, it is replicated for use in the other columns. Table 3
shows the state agent oriented state transition table
corresponding to the ASM chart of fig 7. The optimized ROM
based implementation is shown in fig 8 through the use of a 1-
out-of-8 multiplexer. The number of qualifier lines that reach
the control system has been reduced to 1. The state agent
based fully expanded transition table corresponding to this
optimized ROM architecture is shown in table 4. Using the
multi-agent (process agent plus state agents) based approach
the full multi-agent based implementation of the beverage
sensing and refill control system is shown in fig 9.
WAITING FOR UPPER TANKS REFILL
STOPRUN
BVL
BVH1
HBV1
BVL2
BVH2
HBV2
BVL3
BVH3
HBV3
BVL4
BVH4
HBV4
ST0
ST1
ST2
ST3
ST4
ST5
ST6
000
001
010
100
011
101
110
111
1
0
0
1
10
1
10
1
0
1
1
10
ST7
0
0
0
ST0 ST1
ST3ST2
ST4 ST5
ST6 ST7
0 1
00
01
10
11
ACB
State map
ST0 000
001ST1
ST2 010
ST3 011
100ST4
ST5
ST6
ST7
101
110
111
State
Name
State
Code
State assignment

__________________________________________________________________________________________
Table 3: state transition table corresponding to ASM chart of fig 7
Fig-8: ROM-based implementation of sensing and refill control system
State Present
State
Qualifiers Next
state
State outputs
C B A
STOPRUN
BVL1
BVH1
BVL2
BVH2
BVL3
BVH3
BVL4
BVH4
C′ B′ A′
WAITING
HBV1
HBV2
HBV3
HBV4
ST0 0 0 0 1 - - - - - - - - 0 0 0 1 0 0 0 0
ST0 0 0 0 0 0 - - - - - - - 0 1 0 1 0 0 0 0
ST0 0 0 0 0 1 - - - - - - - 0 0 1 1 0 0 0 0
ST1 0 0 1 - - 0 - - - - - - 0 0 1 0 1 0 0 0
ST1 0 0 0 - - 1 - - - - - - 0 1 0 0 1 0 0 0
ST2 0 1 0 - - - 0 - - - - - 1 0 0 0 0 0 0 0
ST2 0 1 0 - - - 1 - - - - - 0 1 1 0 0 0 0 0
ST3 0 1 1 - - - - 0 - - - - 0 1 1 0 0 1 0 0
ST3 1 0 0 - - - - 1 - - - - 1 0 0 0 0 1 0 0
ST4 1 0 0 - - - - - 0 - - - 1 1 0 0 0 0 0 0
ST4 1 0 0 - - - - - 1 - - - 1 0 1 0 0 0 0 0
ST5 1 0 1 - - - - - - 0 - - 1 0 1 0 0 0 1 0
ST5 1 0 1 - - - - - - 1 - - 1 1 0 0 0 0 1 0
ST6 1 1 0 - - - - - - - 0 - 0 0 0 0 0 0 0 0
ST6 1 1 0 - - - - - - - 1 - 1 1 1 0 0 0 0 0
ST7 1 1 1 - - - - - - - - 0 1 1 1 0 0 0 0 1
ST7 1 1 1 - - - - - - - - 1 0 0 0 0 0 0 0 1
Fully expanded state transition table
HBV3
HBV2
HBV4
BVH/L
BVH2
BVL1 000
001
010
011
100
101
110
111
BVL2
C B A Q C′ B′ A′
FFC
FFB
FFA
R CLK
Square
wave
generator
Power-
up one
shot
HBV1
WAITING
BVL4
BVH3
BVH4
STOPRUN
BVH1
C B A
1 OUT
OF 8
MUXBVL3

__________________________________________________________________________________________
Table -4: fully expanded agent based table corresponding fig 8
State Present State Qualifiers Next State State outputs HEX
OUT-PUT
C B A
STOPRUN
Q C′ B′ A′
WAITING
HBV1
HBV2
HBV3
HBV4
ST0 0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
1
0
1
0
0
0
0
1
0
0
0
0
1
0
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
50
30
10
10
ST1 0
0
0
0
0
0
0
0
1
1
1
1
0
0
1
1
0
1
0
1
0
0
0
0
0
1
0
1
1
0
1
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
28
48
28
48
ST2 0
0
0
0
1
1
1
1
0
0
0
0
0
1
0
1
0
0
1
1
1
0
1
0
0
1
0
1
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
40
60
40
60
ST3 0
0
0
0
1
1
1
1
1
1
1
1
0
0
1
1
0
1
0
1
0
1
0
1
1
0
1
0
1
0
1
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
0
64
84
64
84
ST4 1
1
1
1
0
0
0
0
0
0
0
0
0
0
1
1
0
1
0
1
1
1
1
1
1
0
1
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
C0
C0
A0
A0
ST5 1
1
1
1
0
0
0
0
1
1
1
1
0
0
1
1
0
1
0
1
1
1
1
1
0
1
0
1
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
A2
A2
C2
C2
ST6 1
1
1
1
1
1
1
1
0
0
0
0
0
0
1
1
0
1
0
1
0
1
0
1
0
1
0
1
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
00
00
E0
E0
ST7 1
1
1
1
1
1
1
1
1
1
1
1
0
0
1
1
0
1
0
1
1
0
1
0
1
0
1
0
1
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
E1
E1
01
01

__________________________________________________________________________________________
Fig- 9: Multi-Agent based implementation of sensing and refill control system
6. THE INDUSTRIAL BLENDING PROCESS
The ASM chart of fig 10 shows the algorithm for the blending
process that takes place at each of the mixers 1, 2, 3 & 4 along
with its state map and state assignment.
The ASM chart of fig 4.10 has 8 states. The first state with the
signal ‘BWAIT’ indicates that the beverage blender is waiting
for any of the tanks that is low to be refilled. It should be
noted that if any of the upper tanks is low, or any of the lower
tanks is low (i.e. BVL=1), then the beverage blending control
system must remain in the waiting state i.e. BWAIT. When it
is refilled (BVL=0), the valve for tank 1(HBV1) is opened to
dispense beverage 1 and simultaneously a timer for the valve
(HTV1) is triggered (by HTRGR). Once time timing is up it
moves to trigger the timing and dispensing of beverage 2, then
beverage 3 and beverage 4 in that order. After this the
predetermined quantity of flavor and colour is added at state
ST5 (i.e. PFLCR). When this is done (i.e. TFLCR=0) the
beverages are blended/mixed (HMXR) which is timed by
TMXR. At TMXR=0 the beverage is extruded (i.e. XTRUD)
when this is done (i.e. at TXTRUD=0) the run count signal is
triggered (HRUNCNT) and then the control system waits
(BWAIT) for the next production run. Thus, when this
beverage blending control system is realized for one mixer, it
is replicated for use with other columns of lower beverage
tanks & mixer.
HBV3
HBV2
HBV4
SBVH/L
BVH2
BVL1 000
001
010
011
100
101
110
111
BVL2
State agent 2 CL2
0 1 2 3
40 60 40 60
State agent 0 Cl2
0 1 2 3
50 30 10 10
State agent 1 CL2
0 1 2 3
28 48 28 48
State agent 3 CL2
0 1 3 4
64 84 64 84
State agent 4 CL2
0 1 3 4
C0 C0 A0 A0
State agent 5 CL2
0 1 3 4
A2 A2 C2 C2
State agent 6 CL2
0 1 3 4
00 00 E0 E0
State agent 7 CL2
0 1 3 4
E1 E1 01 01
PROCESS
AGENT
FFC
FFB
FFA
R CLK
Square
wave
generator
Power-
up one
shot
HBV1
WAITING
BVL4
BVH3
BVH4
STOPRUN
BVH1
C B A
1 OUT
OF 8
MUXBVL3

__________________________________________________________________________________________
Fig 10: ASM for beverage blending process
Table 5 shows the state agent oriented state transition table
corresponding to the ASM chart of fig 10. The optimized
ROM based implementation of the ASM chart is shown in fig
11. Through the use of a 1-out-of -8 multiplexer, the number
of qualifier lines that is connected directly to the control
system has been reduced to 1 namely Q. The state agent based
fully expanded transition table corresponding to the optimized
ROM implementation of fig 11 is shown in table 6. Again
using the multi-agent approach (i.e. state agents controlled by
a process agent) the implementation of the industrial blending
process control is as shown in fig 12. It features the use of a
decoder, as described by Tocci et al (2009), to decode state
outputs so as to limit the output line demand of this state
machine to 1 byte or less as required in the optimized ROM
design approach showed by Inyiama et al (in print).
CB
1
1
1
1
0
0
0
0
HBV4
HTV4
HTRGR
PFLCR
TFLCR
HTRGR
HMXR
THMXR
HTRGR
HXTRUD
TXTRUD
HRUNCNT
ST4
ST5
ST6
ST7
1
0
1
1
1
0
0
0
BWAIT
BVL
HTRGR
HBV1
HTV1
HTRGR
HBV2
HTV2
HBV3
HTV3
HTRGR
ST0
ST1
ST2
ST3
HTRGR
ST0 ST1
ST3ST2
ST4 ST5
ST6 ST7
0 1
00
01
10
11
A
State map
ST0 000
001ST1
ST2 010
ST3 011
100ST4
ST5
ST6
ST7
101
110
111
State
Name
State
Code
State assignment

__________________________________________________________________________________________
Table 5: Agent- STT corresponding to ASM of fig. 10
STATE PRESENT STATE QUALIFIERS NEXT STATE STATE OUTPUTS CONDITIONAL OUTPUT
C B A
BVL
HTV1
HTV2
HTV3
HTV4
PFLCR
TMXR
TXTRUD
C′ B′ A′
BWAIT
HBV1
HBV2
HBV3
HBV4
HFLCR
HMXR
HXTRUD
HTRGR
HRUNCNT
STO
STO
000
000
1 - - - - - - -
0 - - - - - - -
000
001
1 0 0 0 0 0 0 0
1 0 0 0 0 0 0 0
0 0
1 0
ST1
ST1
001
001
- 1 - - - - - -
- 0 - - - - - -
001
010
0 1 0 0 0 0 0 0
0 1 0 0 0 0 0 0
0 0
1 0
ST2
ST2
010
010
- - 1 - - - - -
- - 0 - - - - -
010
011
0 0 1 0 0 0 0 0
0 0 1 0 0 0 0 0
0 0
1 0
ST3
ST3
011
011
- - - 1 - - - -
- - - 0 - - - -
011
100
0 0 0 1 0 0 0 0
0 0 0 1 0 0 0 0
0 0
1 0
ST4
ST4
100
100
- - - - 1 - - -
- - - - 0 - - -
100
101
1 0 0 0 1 0 0 0
1 0 0 0 1 0 0 0
0 0
1 0
ST5
ST5
101
101
- - - - - 1 - -
- - - - - 0 - -
101
110
1 0 0 0 0 1 0 0
1 0 0 0 0 1 0 0
0 0
1 0
ST6
ST6
110
110
- - - - - - 1 -
- - - - - - 0 -
110
111
1 0 0 0 0 0 1 0
1 0 0 0 0 0 1 0
0 0
1 0
ST7
ST7
111
111
- - - - - - - 1
- - - - - - - 0
111
000
1 0 0 0 0 0 0 1
1 0 0 0 0 0 0 1
0 0
0 1
Fig- 11: Optimized ROM-based implementation of beverage blending/mixing control system
Fully expanded state transition table
Q
HTV2
BWAIT 000
001
010
011
100
101
110
111
HTV3
FFC
FFB
FFA
R CLK
Square
wave
generator
Power-
up one
shot
HRUNCNT
HTRGR
TMXR
HTV4
TXTRUD
HTV1
C B A
1 OUT
OF 8
MUXPFLCR
000
001
010
011
100
101
110
111
3 to 8 line demux
CBA
HBV3
HBV2
HBV4
HBV1
WAITING
HFLCR
HMXR
HXTRUD

__________________________________________________________________________________________
Table 6: Agent-based transition table derived from fig. 11
STATE PRESENT STATE QUALIFIERS
(Q)
NEXT STATE CONDITIONAL OUTPUT OUTPUT IN HEX
C B A C′ B′ A′
HTRGR
HRUNCNT
STO
STO
000
000
1
0
000
001
0 0
1 0
00
06
ST1
ST1
001
001
1
0
001
010
0 0
1 0
04
06
ST2
ST2
010
010
1
0
010
011
0 0
1 0
08
0D
ST3
ST3
011
011
1
0
011
100
0 0
1 0
0C
12
ST4
ST4
100
100
1
0
100
101
0 0
1 0
10
16
ST5
ST5
101
101
1
0
101
110
0 0
1 0
14
1A
ST6
ST6
110
110
1
0
110
111
0 0
1 0
18
1E
ST7
ST7
111
111
1
0
111
000
0 0
0 1
1C
01
Fig- 12: multi-agent-based implementation of beverage blending/mixing control system
Q
HTV2
BWAIT 000
001
010
011
100
101
110
111
HTV3
FFC
FFB
FFA
R CLK
Square
wave
generator
Power-
up one
shot
HRUNCNT
HTRGR
TMXR
HTV4
TXTRUD
HTV1
C B A
1 OUT
OF 8
MUXPFLCR
State agent 0 Cl1
00 06
State agent 1 Cl1
04 06
State agent 3 Cl1
0C 12
State agent 4 Cl1
10 16
State agent 2 Cl1
08 0D
State agent 7 Cl1
1C 01
State agent 6 Cl1
18 1E
State agent 5 Cl1
14 1A
Process agent
C′ B′ A′C B A Q
000
001
010
011
100
101
110
111
3to8linedemux
C B A
HBV3
HBV2
HBV4
HBV1
WAITING
HFLCR
HMXR
HXTRUD

__________________________________________________________________________________________
Fig 13 shows the multi-agent based linked state machine for
the industrial blender. It comprises of three processes namely:
upper tank control, the beverage level sensing and refill
control and the beverage blending operation.
Each unit is working independently but the upper tank control
passes a signal to the lower tank control when it runs short of
beverage. Similarly, the lower tank control sends a signal
‘BWAIT’ to the blending agent when it is waiting to refill the
tanks that have gone low on stock.
The state agents of each of the state machines do not have
communication with those of other units. It is the process
agents that do the communication between the state machines
.
Fig 13: Multi-agent based linked state machine
CONCLUSIONS
A systematic approach to the design of multi-agent based
linked state machines has been demonstrated in the foregoing.
The method reduces the effort required to realize complex
process control systems through the ‘divide –and-conquer
approach. The complex process control system is first broken
down into a number of simpler control systems which are
quickly realized and then linked up so as to yet constitute the
complex control system. In the multi-agent based approach,
each of the simpler process control systems is implemented as
a multi-agent based system comprised of one process agent
and a number of state agents. Only the process agent in each
state machine is used to perform the intelligent interlinking of
all the machines in order to realize a composite complex
process control system built by concatenating a number of
simpler systems.
Agent 0
Agent 1
Agent 2
Agent 3
Agent 4
Agent 5
Agent 6
Agent 7
Input
value
Process agent for
upper tank refill
Process
O/P
Agent 0
Agent 1
Agent 2
Agent 3
Agent 4
Agent 5
Agent 6
Agent 7
Input
value
Process agent for
lower tank refill
Process
O/P
Agent 0
Agent 1
Agent 2
Agent 3
Agent 4
Agent 5
Agent 6
Agent 7
Input
value
Process agent for
blending operatn
Process
O/P
The plant (beverage blending machine)
Upper tank control Beverage level sensing
control
Beverage blending
operation

__________________________________________________________________________________________
REFERENCES
[1]. Akpado K. A., Inyiama H. C., Okezie C.C. (2010),
“Simplifying industrial process control using linked state
machine approach”. International Journal of Electrical
and Telecommunication System Research. Vol 4. No 4
pp 72-79.
[2]. Inyiama H.C., C.C. Okezie and I.C. Okafo (2011),
“Digital Control of Palm Fruit Processing using ROM
Based Linked State Machines” European Journal of
Scientific Research,Vol.59 No.4 Pp. 597-606.
[3]. Inyiama H.C., C.C. Okezie and I.C. Okafo (2012),
“Agent based process control system design”.
Proceedings of the peer reviewed National Conference
on Infrastructural Development and Maintenance in the
Nigerian Environment. Pp 234-252
[4]. Inyiama H.C., C.C. Okezie and I.C. Okafo (In print),
“Complexity Reduction in Rom-Based Process Control
Systems Via Input Multiplexing and Output Decoding”
International Journal Of Engineering Innovations. Paper
No: AJUS-2011-115
[5]. Tocci R.J, N.S Widmer, G.L Moss, (2009), Digital
systems principles and applications 10th
ed; Pearson; Pp
435.

Designing multi agent based linked state machine

More Related Content

Viewers also liked (20)

Similar to Designing multi agent based linked state machine (20)

More from eSAT Publishing House (20)

Recently uploaded (20)

Designing multi agent based linked state machine