agriculture

Article
Measuring and Comparing Forces Acting on
Moldboard Plow and Para-Plow with Wing to Replace
Moldboard Plow with Para-Plow for Tillage and
Modeling It Using Adaptive Neuro-Fuzzy Interface
System (ANFIS)
Mohammadreza Abbaspour-Gilandeh 1, *, Gholamhossein Shahgoli 1 ,
Yousef Abbaspour-Gilandeh 1 , Miguel Apolonio Herrera-Miranda 2 ,
José Luis Hernández-Hernández 3, * and Israel Herrera-Miranda 4
1 Department of Biosystems Engineering, College of Agriculture and Natural Resources,
University of Mohaghegh Ardabili, Ardabil 56199-11367, Iran; gshahgoli@uma.ac.ir (G.S.);
abbaspour@uma.ac.ir (Y.A.-G.)
2 Graduate Program in Applied Statistics, Faculty of Mathematics, Autonomous University of Guerrero,
Acapulco 39650, Mexico; 07088@uagro.mx
3 Division of Research and Graduate Studies, TecNM/Technological Institute of Chilpancingo,
Chilpancingo 39070, Mexico
4 Graduate Program in Applied Statistics, Government and Public Management Faculty,
Autonomous University of Guerrero, Chilpancingo, Guerrero 39087, Mexico; israelhm@uagrovirtual.mx
* Correspondence: m_abbaspour1366@yahoo.com (M.A.-G.);
joseluis.hernandez@itchilpancingo.edu.mx (J.L.H.-H.); Tel.: +98-3355072 (M.A.-G.)




Received: 3 December 2020; Accepted: 10 December 2020; Published: 14 December 2020


Abstract: The objective of this study was to measure the draft, vertical, and lateral forces acting on
the moldboard plow, para-plow without a wing, para-plow with forward-bent wing, and para-plow
with a backward-bent wing at three working depths and three forward speeds in clay loam soil
to investigate the use of a suitable para-plow instead of the moldboard plow. Also, modeling the
draft, vertical, and lateral forces acting on the implements using Adaptive Neuro-Fuzzy Interface
System (ANFIS) was another objective of this research. To measure the draft, vertical, and lateral
forces, a three-point hitch dynamometer was used. The results showed that, with the increment of
the forward speed and working depth, the draft force required by the used implements increased.
This increase was also true for vertical and lateral forces acting on the implements. Modeling of
the draft, vertical, and lateral forces acting on the implements was performed using the effective
parameters of the implements, working depth, and forward traveling speed using the (ANFIS) fuzzy
neural system model. The root mean square error (RMSE) for the draft, vertical, and lateral forces for
the above models were obtained equal to 0.121, 0.014, and 0.016, respectively.

Keywords: moldboard plow; para-plow; draft force; fuzzy neural system

1. Introduction
One of the fundamental and important issues in the production cycle of agricultural products is
the type of tillage operations [1]. Along with trying to do proper tillage operations, which consumes
a significant portion of the energy in crop production [2], it is highly important to optimize the
operations and the implements used to cultivate agricultural products [3]. The most common and
oldest tillage implement is the moldboard plow. In addition to plows, there are some types of subsoilers,

Agriculture 2020, 10, 633; doi:10.3390/agriculture10120633 www.mdpi.com/journal/agriculture

Agriculture 2020, 10, 633 2 of 12

called para-plows, that have a lower draft than conventional subsoilers [4,5], which nowadays are
widely used in the form of para-plows with wing and without a wing. Determining the force acting
on these agricultural implements by the soil (soil resistance), estimating the energy required to carry
out tillage operations, and choosing a favorable traction force to carry out the desired operations,
as well as managing the decisions required in the implementation of large-scale mechanization to save
fuel and energy consumption are of great importance [6]. Therefore, measuring the forces acting on
tillage implements is one of the main goals of many researchers in the field of tillage and soil dynamics.
Determining the horizontal and vertical components of soil cutting and draft force has attracted the
attention of many researchers [7–9]. This is in direct relationship between the forces acting on tillage
implements and their stability during tillage operations, and the energy applied to the soil, as well as
the size of the tractor (power source) to carry out tillage operations [6,10].
Researchers use agricultural dynamometers to measure the forces acting on the implements during
agricultural operations. Some dynamometers measure all force components acting on the implements.
In some other designs, only the horizontal and vertical forces are measured [11]. In many other systems,
only the horizontal force (draft force) is measured [12]. Godwin et al. [13] used a three-point hitch
dynamometer to measure the torque and force acting on tillage equipment. The dynamometer was
tested and evaluated in both field and laboratory modes. It was able to measure a maximum force of
100 kN in three directions of the coordinate axes and a maximum torque of 100 kN/m in three directions
about the coordinate axes. Khan et al. [14] used a bi-axial extended octagonal ring (EOR) transducer
system to measure the horizontal and vertical forces acting on the implements. The system was used for
a maximum horizontal load of 80 kN and a maximum vertical load of 60 kN. Askari et al. [15] developed
an adjustable three-point hitch dynamometer, with a capacity of 50 kN, to measure the forces acting on
the tractor and mounted implements. This dynamometer provided a variable width and height of
dynamometer links to meet a wide range of implementations. The force-sensing element in this system
consisted of a rubber (load cell) that was installed between the frames. Irshad Ali et al. [16] studied
the horizontal, vertical, and lateral forces acting on a moldboard plow in a laboratory environment.
Studies were conducted at different depths, speeds, and humidity. With increasing speed and depth,
the highest horizontal force of 5.113 kN was obtained. Also, with an increase in humidity from 27% to
33%, the horizontal force was reduced.
Godwin et al. [17] used a mathematical model to predict the draft force acting on the moldboard
plow. The geometric parameters of the plow body, plowing speed, and soil characteristics were used
to determine the above model. The results showed that the predicted and measured forces differed by
2.8%. Roul et al. [18] developed an artificial neural network with a background learning algorithm to
predict the draft force of various tillage equipment in a sandy loam clay soil under different soil and
operational conditions. Al-Hamed et al. [19] used an artificial neural network (ANN) to predict the draft
force required by a disk plow. The proposed neural network model used 10 input parameters: plowing
depth, plowing speed, sand content, clay content, silt content, disk diameter, disk attack angle, tilt angle,
soil moisture content, and soil density. An algorithm based on backpropagation was used to train the
network. The coefficients of determination for the testing points of this model were obtained to be between
0.915 and 0.934. Akbarnia et al. [20] also used a neural network model with a backpropagation training
algorithm to predict the draft force required by two winged share tillage tools in loam soil. The input
parameters for the above network were the working depth, speed, and the width of the wing used in the
implement. The developed model predicted the draft force with a mean relative error of 0.56 compared to
the measured values. Pentoś and Pieczarka [21] developed a mathematical model using an artificial neural
network (ANN) to predict the draft force. Soil texture, soil moisture, compaction, horizontal deformation,
and vertical load were used as input parameters of the neural network. The coefficient of determination
of the model for the validation data set for the draft force was 0.945.
The ANFIS system creates a fuzzy inference system using a set of input–output data. The ANFIS
system is a combination of fuzzy logic and neural network systems. In other words, the ANFIS is
a fuzzy system that gives us the power of qualitative expression and, in this respect, it is similar to
 Agriculture 2020, 10, x FOR PEER REVIEW 3 of 13
Agriculture 2020, 10, 633 3 of 12

fuzzy system that gives us the power of qualitative expression and, in this respect, it is similar to
neuralneural networks
networks withwith variablestructure
variable structuresystem.
system. This
Thismeans
means its its
structure is determined
structure according
is determined to
according
the data. The determination of the parameters of the membership functions in this algorithm is
to the data. The determination of the parameters of the membership functions in this algorithm is
performed using the backpropagation algorithm or its combination with the least squares method.
performed using the backpropagation algorithm or its combination with the least squares method.
This adjustment method allows fuzzy systems to achieve their data acquisition structure [22,23].
This adjustment method
In this study, theallows fuzzyvertical,
horizontal, systemsandto achieve their acting
lateral forces data acquisition
on differentstructure [22,23].
implements were
In this study, the horizontal, vertical, and lateral forces acting on different implements
measured at different speeds and depths. Using ANFIS system, models were presented to predict were measured
at different
each ofspeeds and depths.
these forces by takingUsing ANFISthe
into account system,
type ofmodels were presented
the implement to predict
and working eachspeed
depth and of these
forcesasby takinginputs.
system into account the type of the implement and working depth and speed as system inputs.

2. Materials andand
2. Materials Methods
Methods

FieldField experiments
experiments were
were carriedout
carried outononthe
the research
research fields
fieldsofofMohaghegh
Mohaghegh Ardabili University
Ardabili University
(longitude E 48◦E17
(longitude 0 35”, latitude
48°17′35″, latitude ◦ 120 40”, and
3838°12′40″, and altitude
altitude 1350
1350mmabove
abovesea level)
sea where
level) wheat
where residues
wheat residues
from previous cultivation had remained on the ground. The experiment was performed
from previous cultivation had remained on the ground. The experiment was performed at three working at three
working depths of 10 cm, 20 cm, 30 cm; three forward speeds of 2 km/h, 5 km/h, 7 km/h; and using a
depths of 10 cm, 20 cm, 30 cm; three forward speeds of 2 km/h, 5 km/h, 7 km/h; and using a moldboard
moldboard plow and para-plows with three different wing levels [24] including without wing, with
plow and para-plows with three different wing levels [24] including without wing, with forward-bent
forward-bent wing, and with backward-bent wing (Figure 1).
wing, and with backward-bent wing (Figure 1).

Figure 1. (a)1.Para-plow.
Figure Bend
(a) Para-plow. angle
Bend ofofpara-plow
angle para-plow leg Ω was
leg or Ω 45◦(b)
was45°; ; (b)Top
Top view
view of of
thethe forward
forward andand
(c) backward wings.
(c) backward wings.

The soil
Theofsoil
theofexperimental site was
the experimental siteclay
wasloam
clay (Cambisols in World
loam (Cambisols Reference
in World Base for
Reference Soilfor
Base Resources
Soil
(WRB), International
Resources (WRB), Union of Soil Sciences
International Union of (IUSS), 2014, Vienna,
Soil Sciences (IUSS)Austria). The soil
, 2014, Vienna, moisture
Austria). The and other
soil
moisture
properties and
were other properties
measured at three were
rangesmeasured at three
of soil depth ranges
(0–10 cm,of10–20
soil depth (0–10
cm, and cm, cm).
20–30 10–20Soil
cm, samples
and
were 20–30
carriedcm). Soilsoil
to the samples were
science carried toweighed,
laboratory, the soil science laboratory,
oven dried ◦ C for 24oven
at 105weighed, dried
h, and at 105 again
weighed °C for [25].
24 h, and weighed again [25]. Some characteristics of Ardabil agricultural
Some characteristics of Ardabil agricultural research center was presented in Table 1. research center was
presented in Table 1.
Table 1. Analysis of soil at the experiment site.
Table 1. Analysis of soil at the experiment site.
Property
Property
SandSand
(0.02–2 mm)mm) 41 −1
(0.02–2 41 gg 100
100 gg−1
Silt (0.002–0.02 mm)mm) 28 g 100 g −1
Silt (0.002–0.02 28 g 100 g−1
Clay Clay
(<0.002 mm) 33 −1
(<0.002 mm) 33 gg 100
100 gg−1−1
Organic carbon 0.29 g 100 g −1
Organic carbon 0.29 g 100 −1g
Electrical Conductivity (EC) 0.41 ds m −1
Electrical Conductivity (EC) 0.41 ds m−1
Liquid limit 30 g 100 g −1
Liquid limit 30 g 100 g
Plastic limit 20.05 g 100 g−1
Plastic(ds)
Field capacity limit 25.13 gg 100
20.05 100 gg−1
−1

Field capacity
Dry bulk density (0–10 cm) (ds) 25.13 g 100
948 kg m−3 g −1

Dry bulk
Dry bulk density
density (0–20(0–10
cm) cm) 948 kg
1004 kg mm
−3
−3
Dry bulk
Dry bulk density
density (0–30(0–20
cm) cm) 1004 kg
1112 kg m
m−3−3

Drycontent
Moisture bulk density (0–30
(ds) (0–10 cm)cm) 1112g kg
7.76 −1
100mg−3
Moisture content (ds) (0–20 cm) 11.24 g 100 g−1
Moisture content (ds) (0–30 cm) 18.46 g 100 g−1
ds—dry soil.
 Agriculture 2020, 10, x FOR PEER REVIEW 4 of 13
Agriculture 2020, 10, 633 4 of 12

Moisture content (ds) (0–10 cm) 7.76 g 100 g−1

Moisture content (ds) (0–20 cm) 11.24 g 100 g−1
Massey Ferguson (MF) 285 tractor (Iran Tractor Manufacturing Company, Tabriz, Iran) with a
Moisture content (ds) (0–30 cm) 18.46 g 100 g−1
single differential with 75 hp, equipped with precision measuring instrument systems, was used to
ds—dry soil.
conduct field trials. Field experiments were performed, by taking into account the type of tillage
implement, working
Massey speed,
Ferguson (MF)and
285 depth as three
tractor (Iran effective
Tractor and independent
Manufacturing parameters,
Company, Tabriz, with
Iran) with a five
replications for each treatment. A total of 180 experiments were performed in the form
single differential with 75 hp, equipped with precision measuring instrument systems, was used to of randomized
conduct
complete field
block trials. (RCBD).
design Field experiments were performed,
Each experimental by taking
plot was 2.5 m into
wideaccount
and 15them type
long,ofand,
tillage
at each
stage,implement, working and
the draft, vertical, speed, and depth
lateral forcesas threeon
acting effective and independent
the implements parameters,The
were measured. with fivewheel
fifth
replications
mechanism wasfor eachtotreatment.
used measureAthe total of 180 experiments
forward were
speed of the performed
tractor. Thisinwheel
the formis of randomized
equipped with an
complete block design (RCBD). Each experimental plot was 2.5 m wide and 15 m long, and, at each
electronic tachometer (pulse meter) that senses the gear teeth mounted on the center of the wheel and
stage, the draft, vertical, and lateral forces acting on the implements were measured. The fifth wheel
generates pulses. To obtain the actual speed of the tractor, the pulsed signal (Hz) was measured and
mechanism was used to measure the forward speed of the tractor. This wheel is equipped with an
then electronic
was converted into the real forward speed using the calibration relationship (Equation (1)).
tachometer (pulse meter) that senses the gear teeth mounted on the center of the wheel and
generates pulses. To obtain the actual speed of the tractor, the pulsed signal (Hz) was measured and
1
then was converted into the real forward (HZ −the530.09
S = speed using )
calibration relationship (Equation (1)). (1)
26.96
1
where S is the actual forward velocity, S in m/s. HZ 530.09 (1)
26.96
To measure the depth, a standard depth stabilization wheel mounted on the frame of the
where S is the actual forward velocity, in m/s.
implements was used. The arm, which provided the connection between the wheel and the frame,
To measure the depth, a standard depth stabilization wheel mounted on the frame of the
had three positions to provide the three desired depths of 10, 20, and 30 cm, which allowed the
implements was used. The arm, which provided the connection between the wheel and the frame,
implements
had three to positions
work at the specified
to provide thedepth
three and prevent
desired depthstheofimplement
10, 20, and from penetrating
30 cm, to the
which allowed thelower
depths. To ensure the correct working depth of the implements, the desired depth
implements to work at the specified depth and prevent the implement from penetrating to the lower was also measured
manually and
depths. Tothe accuracy
ensure of the
the correct operation
working depthofofthe
the depth stabilization
implements, wheel
the desired was
depth wasassured.
also measured
manually
To measure andthe
the draft,
accuracy of the operation
vertical, of the
and lateral depthacting
forces stabilization
on thewheel
tineswas assured.
of the para-plow and the
moldboard To plows
measure bythe
thedraft,
soil, vertical, and lateral
the three-point forces
hitch acting on thedesigned
dynamometer tines of the
bypara-plow and the
Abbaspour-Gilandeh
moldboard plows by the soil, the three-point hitch dynamometer designed by
and Khanmaraki [26] was used. It includes three octagonal load cells, strain gauges mounted on Abbaspour-Gilandeh
and Khanmaraki [26] was used. It includes three octagonal load cells, strain gauges mounted on
octagonal load cells and wires connected between the installed strain gauges, the data logger (DT800,
octagonal load cells and wires connected between the installed strain gauges, the data logger (DT800,
Data Taker company, Australia), and the laptop, which are shown in Figure 2. The data logger and
Data Taker company, Australia), and the laptop, which are shown in Figure 2. The data logger and
laptop received their power from an extra battery in the cabin. Before the experiments, dynamometer
laptop received their power from an extra battery in the cabin. Before the experiments, dynamometer
calibration waswas
calibration performed
performedon onthe
the field [27].Calibration
field [27]. Calibration showed
showed a higha correlation
high correlation
betweenbetween
draft ( R draft
2 2 2
(R ==0.98),
0.98), vertical R =
vertical ((R 0.98),and
= 0.98), andlateral ( R(R = =
lateral 0.98)
0.98) forces
forces in kN
in kN andand dynamometer
dynamometer outputoutput
in µV. in µV.

Figure 2. (a) The mechanism used for field experiments includes: (b) the three-point connection
dynamometer to measure the force applied to devices; (c) fifth wheel mechanism for measuring
real progressive speed; (d) para-plow with the backward wings; (e) data logger and laptop for data
recording; (f) the battery built into the cabin to power the data logger and laptop.
Agriculture 2020, 10, x FOR PEER REVIEW 5 of 13

Figure 2. (a) The mechanism used for field experiments includes: (b) the three-point connection
dynamometer to measure the force applied to devices; (c) fifth wheel mechanism for measuring real
progressive
Agriculture 2020, 10, speed;
633 (d) para-plow with the backward wings; (e) data logger and laptop for data5 of 12
recording; (f) the battery built into the cabin to power the data logger and laptop.

After collecting
After collecting the
the data,
data, the
the statistical
statistical analysis
analysis of
of the obtained data was performed
performed using SAS
software (version 9.1, North Carolina, USA). A comparison of
software (version 9.1, North Carolina, USA). comparison of the meansthe means of the data was performed
using Duncan’s test (p <<5%).
5%).

Designing the
Designing the ANFIS
ANFIS Fuzzy
Fuzzy Neural
Neural System
System
In this
In this section,
section, the
the three
three fuzzy
fuzzy neural
neural models
models forfor predicting
predicting draft,
draft, vertical,
vertical, and
and lateral
lateral forces
forces
acting on
acting on the
theimplements
implementsininthe thesoil under
soil thethe
under influence of the
influence of three factors
the three of velocity,
factors working
of velocity, depth,
working
and type of implement were assessed. Out of a total of 180 data obtained, 80% of
depth, and type of implement were assessed. Out of a total of 180 data obtained, 80% of the data were the data were
considered as
considered as training
training data
data and
and 20%
20% ofof the
the data
data were
were considered
considered as as checking
checking data. After importing
data. After importing
data, the initial FIS (fuzzy inference system) structure had to be determined. Using
data, the initial FIS (fuzzy inference system) structure had to be determined. Using the grid the grid partitioning
discretizationdiscretization
partitioning method, the type and number
method, the typeof and
membership
number functions for input
of membership and output
functions for parameters
input and
were determined. In all models the same structure was used, and the employed
output parameters were determined. In all models the same structure was used, and the employedpartitioning structure
used can be seen in Figure 3.
partitioning structure used can be seen in Figure 3.

Figure 3. Partitioning of the membership functions of the input parameters and the membership
function3.ofPartitioning
Figure the output parameter.
of the membership functions of the input parameters and the membership
function of the output parameter.
As can be seen in Figure 3, five membership functions were considered for all three input parameters
of implement type, working speed, and working depth and the type of membership functions for the
As can be seen in Figure 3, five membership functions were considered for all three input
input parameters were selected as triangular membership function (trimf). The membership function
parameters of implement type, working speed, and working depth and the type of membership
type for the output parameter in the proposed models was linear. The hybrid optimization method,
functions for the input parameters were selected as triangular membership function (trimf). The
which is a combination of the least squares method and (gradient) descent backpropagation method,
membership function type for the output parameter in the proposed models was linear. The hybrid
was used to train the ANFIS structure. The optimal error for the above models was considered to
optimization method, which is a combination of the least squares method and (gradient) descent
be zero. The number of training intervals was taken to be 30 intervals. Data training was stopped
backpropagation method, was used to train the ANFIS structure. The optimal error for the above
by achieving the required number of intervals or error level. From the training and validation of
models was considered to be zero. The number of training intervals was taken to be 30 intervals. Data
the ANFIS model, the error value was obtained and plotted and the output value from ANFIS was
training was stopped by achieving the required number of intervals or error level. From the training
compared with the training and checking data.
and validation of the ANFIS model, the error value was obtained and plotted and the output value
from ANFIS
3. Results was
and compared with the training and checking data.
Discussion
Tableand
3. Results 1 shows the results obtained from the analysis of the variance of the data related to
Discussion
the measurement of the draft, vertical, and lateral forces acting on different tillage implements.
The coefficient of variation (CV) of the data was 0.888% for draft force, 2.5% for vertical force, and 5.59%
for lateral force, where this value is appropriate and acceptable in terms of the number of treatments
and field test conditions. According to Table 2, it can be seen that the main impact of implement type,
Agriculture 2020, 10, 633 6 of 12

depth, and speed, individually and separately, on measured forces was significant at the probability
level of 1% (p < 0.01). Besides, the mutual effects of forward speed on working depth, type of tillage
implement on forward speed, and type of tillage implement on working depth on measured forces
were significant (p < 0.01). The triple interaction effect of parameters on draft, vertical, and horizontal
forces was significant, which is considered a desirable outcome and justifies the use of advanced
modeling systems such as ANFIS.

Table 2. Analysis of variance related to the measurement of forces acting on tillage equipment.

Mean Square
Degree of Freedom Source of Variation
Lateral Force (Fl ) Vertical Force (Fv ) Draft Force (Fd )
41.8088 ∗∗ 4.19177 ∗∗ 2423.4941 ∗∗
20.0341 ∗∗ 36.8328 ∗∗ 543.7585 ∗∗ 3 Tools
4.61797 ∗∗ 6.9408 ∗∗ 175.0482 ∗∗ 2 Speed
4.5637 ∗∗ 3.04847 ∗∗ 76.329 ∗∗ 2 Depth
0.2329 ∗∗ 2.6498 ∗∗ 1.5481 ∗∗ 6 Tools × Speed
0.17838 ∗∗ 1.2186 ∗∗ 1.41198 ∗∗ 6 Tools × Depth
0.35314 ∗∗ 0.60144 ∗∗ 4.3113 ∗∗ 4 Speed × Depth
0.007409 0.00127 0.01021 144 Error
** p < 0.01; ns: not significant.

Figure 4 shows the variations of draft force (Fd) in different treatments. The results showed
that the maximum draft force of 21.09 kN occurred for moldboard plow at a forward speed of
7 km/h and working depth of 30 cm. The least draft force of 4.11 kN was related to the para-plow
without a wing, which was observed at a working depth of 10 cm and forward speed of 2 km/h.
The results showed that with the increment of forward speed, the draft required force increased.
As the working depth increased, the draft force for each of the tillage implements increased. The draft
force increment, under increasing forward speed and working depth, was reported in the studies
of many researchers [10,28,29]. In all treatments, the draft force of the moldboard plow was higher
compared to other implements. The draft force of the para-plow equipped with a backward-bent wing
in all working speeds and depths was lower than para-plow equipped with backward-bent wings,
which was predictable due to the shape of the wings and the higher interaction of wings with the soil
in the para-plow equipped with forward-bent wings, and the results showed good consistency with
the research by Askari et al. [24]. The trend of the increase of draft force with the increase of working
depth and forward speed occurred without sudden increase and linearly. This behavior of draft force
variations was consistent with the report by Sharifat and Kushwaha [30]. Also, by increasing the depth
at the same speed, an increase in the draft force was observed in all treatments and for all implements,
which was consistent with findings of Kirisci et al. [12], Godwin et al. [13], and Askari et al. [15].
Figure 5 shows the effect of tillage type, forward speed, and working depth on the vertical resistance
force. The highest vertical force of 2.82 kN was obtained using para-plow equipped with forward-bent
wings, which occurred at forward speed of 7 km/h and working depth of 30 cm. Also, the least vertical
force of 0.82 kN acted on the para-plow without wing at a speed of 2 km/h and working depth of 10 cm.
For all implements, the increment in the vertical force happened with increasing working depth and
forward speed. In the case of the moldboard plow, this increase occurred at a slower pace. However,
at higher speeds, the increase in the vertical force acting on the moldboard plow showed a specific trend.
Increasing the depth at the same speed led to larger increases in the vertical force, while this increase
occurred with a slower slope with increasing speed at the same depth, which is consistent with the
reports by Askari et al. [24] and Kasisira and du Plessis [31,32]. The high vertical force in the para-plow
equipped with forward-bent wings is an indication of the tendency of this implement to penetrate to the
soil. This fact is also true for other types of para-plow (without wing and equipped with backward-bent
wings). For this reason, in order to control the working depth of these implements, a depth control
wheel was used to allow the implements to penetrate up to the desired depth. The need for using it was
especially felt at high speeds, where the vertical force increased significantly.
 Agriculture 2020, 10, 633 7 of 12
Agriculture 2020, 10, x FOR PEER REVIEW 7 of 13

paraplow without
25 wings
paraplow with
20 backward wings
paraplow with
Draft force, kN

forward wings
15 Moldboard plow

10 Speed:
S1= 2 km/h
S2= 5 km/h
5 S3= 7 km/h

Depth:
0 D1= 10 cm
S1,D1 S1,D2 S1,D3 S2,D1 S2,D2 S2,D3 S3,D1 S3,D2 S3,D3 D2= 20 cm
Treatments D3= 30 cm

Figure 4. Mean draft forces under different implements and treatments, S and D show speed and
Agriculture 2020, 10, x FOR PEER REVIEW 8 of 13
depth, respectively.

3
Figure 4. Mean draft forces under different implements and treatments, S and D show speed and
depth, respectively. Paraplow without
2.5 wings
Paraplow with
Figure 5 shows the effect of tillage type, forward speed, and working depth on the vertical
Vertical force, kN

2 backward wings
resistance force. The highest vertical force of 2.82 kN was obtained using para-plow equipped Paraplowwithwith
forward-bent
1.5 wings, which occurred at forward speed of 7 km/h and working depth of 30forward wings
cm. Also,
Moldboard plow
the least vertical force of 0.82 kN acted on the para-plow without wing at a speed of 2 km/h and
working1 depth of 10 cm. For all implements, the increment in the vertical force happened Speed:with
S1= 2 km/h
increasing working depth and forward speed. In the case of the moldboard plow, thisS2= increase
5 km/h
occurred0.5 at a slower pace. However, at higher speeds, the increase in the vertical force acting S3=on7 km/h
the
moldboard plow showed a specific trend. Increasing the depth at the same speed led to larger
0 Depth:
increases in the vertical force, while this increase occurred with a slower slope with increasing D1=speed
10 cm
S1,D1 S1,D2 S1,D3 S2,D1 S2,D2 S2,D3 S3,D1 S3,D2 S3,D3
at the same depth, which is consistent with the reports by Askari et al. [24] and Kasisira and du D2=Plessis
20 cm
Treatments D3= 30 cm
[31,32]. The high vertical force in the para-plow equipped with forward-bent wings is an indication
of the tendency of thisvertical
Figure 5. Mean implement forcestounder
penetrate to the
different soil. This fact
implements andistreatments,
also true for other
S and typesspeed
D show of para-
and
plow (without wing and equipped with backward-bent wings). For this reason, in order to control
depth, respectively.
the working depth of these implements, a depth control wheel was used to allow the implements to
Figure
Figure
penetrate up 6 the
5. to shows
Mean theforces
vertical
desired effect ofThe
under
depth. the tillage
different
need implement
implements
for using it and type, working
wastreatments,
especially S felt speed,
and at
D show and working
high speed
speeds,andwhere thedepth on
depth,
the lateral
vertical respectively.
forceforces.
increasedThe highest lateral force of 3202 kN acted on the moldboard plow at a working
significantly.
depth of 30 cm and forward speed of 7 km/h, and the least lateral force of 0.598 kN occurred using
Figure 6 shows the effect of the tillage implement type, working speed, and working depth on
para-plow equipped with backward-bent wings at a working depth of 10 cm and forward speed of
the lateral forces. The highest lateral force of 3202 kN acted on the moldboard plow at a working
7 km/h. The increasing trend of the lateral force with the increase of working depth and forward speed
depth of 30 cm and forward speed of 7 km/h, and the least lateral force of 0.598 kN occurred using
was quiteequipped
para-plow obvious,withwhich is consistentwings
backward-bent withat the report by
a working Irshad
depth Ali
of 10 cm[16].
and The impact
forward speed ofof
adding
7 wings
to para-plow
km/h. was significant,
The increasing trend of the and greatly
lateral decreased
force with the lateral
the increase forces
of working and and
depth led forward
to the greater
speed balance of
the quite
was para-plow.
obvious,This
whichfactisisconsistent
quite evident
with thein report
the comparison
by Irshad Aliof [16].
the lateral forces
The impact of acting
addingon the para-plow
wings
to para-plow
without wasand
wings significant, and greatly
the para-plow decreased
equipped the wings
with lateral (forward-bent
forces and led toand the backward-bent).
greater balance In most
of the para-plow. This fact is quite evident in the comparison of the lateral forces
working conditions, para-plow with backward-bent wings showed lower lateral force compared to acting on the para-
plow withoutwith
para-plow wingsforward-bent
and the para-plow wings.equipped with wings
The lateral force(forward-bent
acting on the andmoldboard
backward-bent).
plowInin different
most working conditions, para-plow with backward-bent wings showed lower lateral force
working conditions was almost twice as much as the lateral force acting the implements with wings.
compared to para-plow with forward-bent wings. The lateral force acting on the moldboard plow in
It happened due to the high volume of soil displaced by the plow, and it seemed that as the draft force
different working conditions was almost twice as much as the lateral force acting the implements
increased,
with wings. Itsohappened
did the lateral
due to force acting
the high volumeon the implements.
of soil displaced by the plow, and it seemed that as
the draft force increased, so did the lateral force acting on the implements.
 3.5 Paraplow without
wing
3 Paraplow with
backward wings
Agriculture
2.5 2020, 10, 633 Paraplow8 ofwith
12

Lateral force, kN
Agriculture 2020, 10, x FOR PEER REVIEW 9 of 13
forward wings
2 Moldboard plow
3.5 Paraplow without
1.5 wing Speed:
3 Paraplow with
S1= 2 km/h
backward wings
1
2.5 ParaplowS2=
with5 km/h
forward wings7 km/h
S3=
Lateral force, kN

0.5 2 Moldboard plow

Depth:
D1= 10 cm
01.5 Speed: D2= 20 cm
S1,D1 S1,D2 S1,D3 S2,D1 S2,D2 S2,D3 S3,D1 S3,D2 S3,D3 S1= 2 km/h
D3= 30 cm
1 S2= 5 km/h
Treatments S3= 7 km/h
0.5
Depth:
D1= 10 cm
0
D2= 20 cm
S1,D1 S1,D2 S1,D3 S2,D1 S2,D2 S2,D3 S3,D1 S3,D2 S3,D3 D3= 30 cm
Treatments
Figure
Figure 6.
6. Mean
Meanlateral
lateralforces
forcesunder
underdifferent
different implements
implements and
and treatments,
treatments,SS and
and D
D show
show speed
speed and
and
depth, respectively.
depth, respectively.

Predicting Draft,
Predicting Draft, Vertical,
Vertical, and Lateral Forces Acting on Implements Using ANFIS Model
Figure7 6.
Figure Mean lateral forces underbetween
different implements and treatments, S and D show and
speedthe
andoptimization
Figure 7shows
showsthetherelationship
relationship between the root
themeanrootsquare
meanerror (RMSE)
square error (RMSE) and the
depth, respectively.
steps of training
optimization stepsandof checking
training anddatachecking
for draft force, where
data for draftthe errorwhere
force, of thethetraining
error data
of theistraining
marked data
with
(*)marked
is and the error
with of
(*) the
and checking
the error data
of theis marked
checking with
data (♦)
is symbol.
marked with (◊)
Predicting Draft, Vertical, and Lateral Forces Acting on Implements Using ANFIS Model symbol.
Figure 7 shows the relationship between the root mean square error (RMSE) and the
optimization steps of training and checking data for draft force, where the error of the training data
is marked with (*) and the error of the checking data is marked with (◊) symbol.

Figure7.
Figure Changesin
7.Changes inthe
thetraining
trainingand
andchecking
checkingdata
dataerror.
error.

After training
After training and
and validating
validating the
theANFIS
ANFISmodel modelfor fordifferent
different input
inputparameters,
parameters, thethe
error values
error for
values
Figure 7. Changes in the training and checking data error.
the ANFIS model for the draft, vertical, and lateral forces were obtained equal
for the ANFIS model for the draft, vertical, and lateral forces were obtained equal to 0.376, 1.38, and to 0.376, 1.38, and 2.4,
respectively.
2.4, respectively.TheThe
After trainingoutput
and obtained
output obtained
validating from the ANFIS
from
the ANFIS the was
ANFIS
model compared
was compared
for different with
input the training
with and checking
the training
parameters, the error and data,
checking
values
and
data, the
forand results
the ANFIS are shown
modelare
the results in
forshown Figure
the draft, 8.
in vertical,In
Figure 8. Figure
and 8a,
Inlateral
Figure the
forces training data
weretraining
8a, the are
obtaineddata
equalmarked
to marked
are with
0.376, 1.38, (o) and the
and(o) and
with
modeling
the2.4, data are
respectively.
modeling data Themarked with the
output obtained
are marked with the (*) sign.
from(*) theIn Figure
ANFIS
sign. 8b, the
was compared
In Figure validation data are
with the training
8b, the validation data aremarked
and with
checking
marked the (+)
with the
(+) data,
sign
sign and
andand thethe
theresults
predictive are
predictive shown
data in
are are
data Figure
markedmarked 8. Inwith
with Figure
the8a,
the (*) (*)the
sign. training
Table
sign. data are the
3 presents
Table marked
3 presents thewith
root root(o)
mean and square
square
mean error
forthe
all modeling
three data are
models, marked
both for with
the the (*) sign.
training data In Figure
and the 8b, the validation
validation data. data are marked with the
error for all
Agriculture three
2020, 10, x models,
FOR PEERboth for the training data and the validation data.
REVIEW 10 of 13
(+) sign and the predictive data are marked with the (*) sign. Table 3 presents the root mean square
error for all three models, both for the training data and the validation data.

Figure 8.
Figure Comparisonof
8. Comparison ofpredicted
predicted data
data (a)
(a) with
with training
training data
data and
and (b)
(b) with
with checking
checking data.
data.

Table 3. Root mean square error table for training and validation of the models offered.

Root Mean Square Error (RMSE)

FIS Models for Forces
Test Data Train Data
Agriculture 2020, 10, 633 9 of 12

Figure 8. Comparison of predicted data (a) with training data and (b) with checking data.

Table 3. Root mean square error table for training and validation of the models offered.
Table 3. Root mean square error table for training and validation of the models offered.
Root
RootMean
Mean Square Error(RMSE)
Square Error (RMSE)
FIS Models for Forces
FIS Models for Forces
TestData
Test Data Train
TrainData
Data
0.037602
0.037602 0.02149
0.02149 DraftDraft
force (Fd
force (Fd)
0.018901 0.01426 Vertical force (Fv)
0.018901 0.01426 Vertical force (Fv)
0.021046 0.01658 Lateral force (Fl)
0.021046 0.01658 Lateral force (Fl

The output of the ANFIS system is presented in the two forms of surface graphs and rules. In the
surfaceThe output
case, of the of
the output ANFIS system was
the system is presented
obtainedinin
the two
the forms
form of surface graphs and
of three-dimensional rules.versus
graphs In the
the surface case, the output of the system was obtained in the form of three-dimensional graphs
input and output parameters. Figure 9a shows the mutual effect of the type of implements and the
versus the input and output parameters. Figure 9a shows the mutual effect of the type of implements
working depth on the draft force. It was found that the highest draft force occurred for the moldboard
and the working depth on the draft force. It was found that the highest draft force occurred for the
plow at a working depth of 30 cm, and the least draft force was observed for para-plow without wings
moldboard plow at a working depth of 30 cm, and the least draft force was observed for para-plow
at awithout
working depth
wings at a of 10 cm.depth
working The draft force
of 10 cm. required
The forrequired
draft force the para-plow equippedequipped
for the para-plow with a forward-bent
with
winga forward-bent wing was greater than the para-plow equipped with a backward-bentwas
was greater than the para-plow equipped with a backward-bent wing, which wing,true at different
which
speeds andatworking
was true different depths. Figure
speeds and 9b,cdepths.
working also shows
Figurethe9b,cmutual effect
also shows theof the type
mutual ofof
effect implements
the type and
theofworking
implements andon
depth thethe
working depthvertical
measured on the measured vertical
and lateral forces, andrespectively.
lateral forces, respectively.

Figure 9. Surface graph showing the relationship of implements and depth with (a) draft force (Fd),
(b)Figure 9. Surface
vertical graph
force (Fv), showing
and the relationship
(c) lateral force (Fl). of implements and depth with (a) draft force (Fd),
(b) vertical force (Fv), and (c) lateral force (Fl).
In order to determine the relationship between measured and predicted values in the presented
models,Inaorder to determine
regression the relationship
diagram was drawnbetween measured
and is shown and predicted
in Figure values
10, which in the
is an presented
indication of a very
models, a regression diagram was drawn and is shown 2 in Figure 10, which is an indication of a very
high and acceptable coefficient of determination (R ) for all three presented models.
Agriculture 2020, 10, x FOR PEER REVIEW 11 of 13
high and acceptable coefficient of determination (R ) for all three presented models.

Figure 10. The

Figure10. The relationship betweenthe
relationship between thevalues
values measured
measured andand predicted
predicted by the
by the ANFIS
ANFIS models
models for
for (a)
(a) draft
draft force
force (Fd),
(Fd), (b)
(b) vertical
vertical force
force (Fv),
(Fv), and
and (c)(c) lateral
lateral force
force (Fl);
(Fl); y indicates
y indicates the
the regression
regression relationship
relationship
betweenthe
between themeasured
measuredvalues
valuesandand the
the predicted
predicted values.

Table 4 shows the characteristics of the models created and the evaluation of these models based
on the root mean square, the relative error percentage, and the coefficient of determination (R )
parameters. These results show a high accuracy compared to the neural network models presented
for draft force by Roul et al. [18], with a coefficient of determination of 0.934, and Akbarnia et al. [20],
with a mean relative error of 0.56. Also, a comparison of the results of ANFIS modeling with the
neural network model of Pentoś and Pieczarka [21], which was presented to predict the draft force
Agriculture 2020, 10, 633 10 of 12

Table 4 shows the characteristics of the models created and the evaluation of these models based on
the root mean square, the relative error percentage, and the coefficient of determination (R2 ) parameters.
These results show a high accuracy compared to the neural network models presented for draft force
by Roul et al. [18], with a coefficient of determination of 0.934, and Akbarnia et al. [20], with a mean
relative error of 0.56. Also, a comparison of the results of ANFIS modeling with the neural network
model of Pentoś and Pieczarka [21], which was presented to predict the draft force using a neural
network with the coefficient of determination of R2 = 0.945, indicates the high accuracy of ANFIS fuzzy
neural systems.

Table 4. Characteristics of ANFIS models and evaluation based on statistical parameters.

Type of Membership Number of Membership Learning ANFIS Models

RMSE (%)ε R2
Functions Functions Method for Forces
Input Output Input Epoch - - - - -
- - - -
- Linear - - - Fd 0.3756 0.0214 0.9999
Trimf - 5 30 Hybrid Fv 0.0142 0.0142 0.9981
Fl 0.0166 0.0166 0.9977

4. Conclusions
In this research, the measurement of draft, vertical, and lateral forces acting on two tillage
implements was studied at three forward speeds and three working depths. Modeling of the forces
acting on the implements was also performed using the ANFIS fuzzy neural system. The results of
the measurement of the forces showed that the main effects of the implements, the working depth,
and the working speed each individually, on the measured forces, were significant. Also, the mutual
and triple interaction effects of these parameters were found to be significant at a probability level of
1%. According to the obtained diagrams, it can be said that with the increment of the working speed
and depth, the draft, vertical, and lateral forces increased. The highest draft force of 21.09 kN and
lateral force of 3.2 kN was related to the moldboard plow, and the highest vertical force of 2.82 kN
occurred in the case of para-plow equipped with forward-bent wings. The least draft force of 0.598 kN
was observed in para-plow equipped with backward-bent wings. At all working speeds and depths,
the lateral force acting on the para-plow without wing was almost twice as much as the force acting on
the para-plows with wing, indicating the impact of adding the wing to the para-plow, which reduced
the lateral force to a great extent and made the para-plow more balanced. Obtained data indicated that
winged para-plow is a proper implement that can be used in tilling agricultural soils and showed a
promising prospective in the future of primary tillage in terms of power saving.
In the continuation of this research, the modeling of draft, vertical, and lateral forces acting on the
implements was performed using the effective parameters of the forward speed, the working depth,
and the type of implements, using (ANFIS) fuzzy neural system. The prediction results showed the
coefficient of determination (R2 ) in all three models presented was above 0.99. The error rates for the
models of the prediction of the draft, vertical, and lateral forces were obtained equal to 0.3756%, 1.385%,
and 2.45%, respectively. The results showed that the (ANFIS) fuzzy neural system is a powerful tool
for predicting the forces acting on the implements, and the data predicted by this model have very
high accuracy.

Author Contributions: Conceptualization, M.A.-G., G.S., Y.A.-G.; methodology, M.A.-G., G.S., Y.A.-G.; software,
M.A.-G.; validation, G.S., Y.A.-G., J.L.H.-H.; formal analysis, M.A.-G., G.S.; investigation, M.A.-G., G.S., Y.A.-G.;
resources, M.A.-G., G.S.; data curation, M.A.-G., Y.A.-G.; writing—original draft preparation, M.A.-G., G.S.;
writing—review and editing, G.S., Y.A.-G., J.L.H.-H., M.A.H.-M., I.H.-M.; visualization, M.A.-G.; supervision, G.S.;
project administration, G.S.; funding acquisition, Y.A.-G., J.L.H.-H., M.A.H.-M., I.H.-M. All authors have read and
agreed to the published version of the manuscript.
Funding: This study was financially supported by University of Mohaghegh Ardabili, Ardabil, Iran.
Agriculture 2020, 10, 633 11 of 12

Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to
publish the results.

References
1. Jacobs, C.O.; Finnery, J.B. Soil Managemen; Farming Press Publication: Totnes, UK, 1993; pp. 187–189.
2. Dexter, A.R.; Czyz, O.P.; Gat, E. A method for soil penetration resistance. Soil Till. Res. 2007, 93, 412–419.
[CrossRef]
3. Kheiralla, A.F.; Yahya, A.; Zohadie, M.; Ishak, W. Design and development of three point auto hitch
dynamometer for an agricultural tractor. AJSTD 2003, 20, 271–288. [CrossRef]
4. Harrison, H.; Licsko, Z. Soil reacting forces for models of three bentleg plows. Soil Tillage Res. 1989, 15, 125–135.
[CrossRef]
5. Pidgeon JDSoane, B.D. Effects of tillage and direct drilling on soil properties during the growing season in a
long-term barely mono-culture system. J. Agric. Sci. 1977, 88, 431–442. [CrossRef]
6. Al-Janobi, A.; Al-Suhaibani, S.A. Draft of primary tillage implements in sandy loan soil. Appl. Eng. Agric.
1998, 14, 343–348. [CrossRef]
7. Fervers, C. Improved FEM simulation model for tire–soil interaction. J. Terramech. 2004, 41, 87–100. [CrossRef]
8. Nakashima, H.; Oida, A. Algorithm and implementation of soil–tire contact analysis code based on dynamic
FE–DE method. J. Terramech. 2004, 41, 127–137. [CrossRef]
9. Yong, R.; Boonsinsuk, P.; Fattah, E. Tyre flexibility and mobility on soft soils. J. Terramech. 1980, 17, 43–58.
[CrossRef]
10. Alsuhaibani, S.A.; Aljnobi, A.A.; Almajhadi, Y.N. Tractors and Tillage Implements Performance. In Proceedings
of the CSBE/SCGAB 2006 Annual Conference, Edmonton, AB, Canada, 16–19 July 2006.
11. Scholtz, D.C. A three-point hitch dynamometer for restrained linkages. J. Agric. Eng. Res. 1966, 11, 33–37.
[CrossRef]
12. Kirisci, V.; Blackmore, B.S.; Kilgour, J.; Godwin, R.J.; Babier, A.S. A Three-Point Linkage Dynamometer System.
In Proceedings of the International Conference on Agriculture Engineering, Uppsala, Sweden, 1–4 June 1992.
13. Godwin, R.; Reynolds, A.; O’Dogherty, M.; Al-Ghazal, A. A Triaxial Dynamometer for Force and Moment
Measurements on Tillage Implements. J. Agric. Eng. Res. 1993, 55, 189–205. [CrossRef]
14. Khan, J.; Godwin, R.J.; Kilgour, J.; Blackmore, B.S. Design and Calibration of a BI-Axial Extended Octagonal
Ring Transducer System for the Measurment of Tractor-Implement Forces. J. Eng. Appl. Sci. 2007, 2, 16–20.
15. Askari, M.; Komarizade, M.; Nikbakht, A.; Nobakht, N.; Teimourlou, R. A novel three-point hitch dynamometer
to measure the draft requirement of mounted implements. Res. Agric. Eng. 2011, 57, 128–136. [CrossRef]
16. Irshad Ali, M.; Changying, J.; Leghari, N.; Ghulam Ali, B.; Farman Ali Ch ChuadryArslan Ch Sattar, A.;
Huimin, F. Analyses of 3-dimensional draught and soil deformation forces caused by mouldboard plough in
clay loam soil. Glob. Adv. Res. J. Agric. Sci. 2015, 4, 259–269.
17. Godwin, R.; O’Dogherty, M.; Saunders, C.; Balafoutis, A. A force prediction model for mouldboard ploughs
incorporating the effects of soil characteristic properties, plough geometric factors and ploughing speed.
Biosyst. Eng. 2007, 97, 117–129. [CrossRef]
18. Roul, A.; Raheman, H.; Pansare, M.; Machavaram, R. Predicting the draught requirement of tillage implements
in sandy clay loam soil using an artificial neural network. J. Biosyst. Eng. 2009, 104, 476–485. [CrossRef]
19. Al-Hamed, S.A.; Wahby, M.F.; Al-Saqer, S.M.; Aboukarima, A.M.; Sayedahmed, A.A. Artificial Neural
Network Model for Predicting Draft and Energy Requirements of a Disk Plow. J. Anim. Plant Sci. 2013,
23, 1714–1724.
20. Akbarnia, A.; Mohammadi, A.; Farhani, F.; Alimardani, R. Simulation of draft force of winged share tillage
tool using artificial neural network model. Agric Eng. Int. CIGR J. 2014, 16, 57–65.
21. Pentoś, K.; Pieczarka, K. Applying an artificial neural network approach to the analysis of tractive properties
in changing soil conditions. Soil Tillage Res. 2017, 165, 113–120. [CrossRef]
22. Abbaspour-Gilandeh, Y.; Sedghi, R. Predicting soil fragmentation during tillage operation using a fuzzy
logic approach. J Terrramech. 2015, 57, 61–69. [CrossRef]
23. Abbaspour-Gilandeh, S.S. Automatic grading of emperor apples based on image processing and ANFIS.
Tarım Bilim. Derg. J. Agric. Sci. 2015, 21, 326–336. [CrossRef]
Agriculture 2020, 10, 633 12 of 12

24. Askari, M.; Shahgholi, G.; Abbaspour-Gilandeh, Y. New wings on the interaction between conventional
subsoiler and paraplow tines with the soil: Effects on the draft and the properties of soil. Arch. Agron. Soil Sci.
2018, 65, 88–100. [CrossRef]
25. Page, A.L.; Miller, R.H.; Keeney, D.R. Methods of Soil Analysis; American Society of Agronomy (Publ.):
Madison, WI, USA, 1982.
26. Abbaspour-Gilandeh, Y.; Khanramaki, M. Design, Construction and Calibration of a Triaxial Dynamometer
for Measuring Forces and Moments Applied on Tillage Implements in Field Conditions. Met. Soc. India 2013,
28, 119–127. [CrossRef]
27. Askari, M.; Shahgholi, G.; Abbaspour-Gilandeh, Y.; Tash-Shamsabadi, H. The Effect of New Wings on
Subsoiler Performance. Appl. Eng. Agric. 2016, 32, 353–362.
28. De Souza, E.G.; Lima, J.S.S.; Milanez, L.F. Overall Efficiency of Tractor Operating in the Field. Trans. ASAE
1994, 106, 771–775. [CrossRef]
29. Raheman, H.; Jha, S. Wheel slip measurement in 2WD tractor. J. Terramech. 2007, 44, 89–94. [CrossRef]
30. Sharifat, K.; Kushwaha, R.L. Modeling soil movement by tillage tools. Can. Agric. Eng. 2000, 42, 165–172.
31. Kasisira, L.; Du Plessis, H. Energy optimization for subsoilers in tandem in a sandy clay loam soil. Soil Tillage Res.
2006, 86, 185–198. [CrossRef]
32. Al-janobi, A. A data-acquistion system to monitor performance of fully mounted implements. J. Agric.
Eng. Res. 2000, 75, 167–175. [CrossRef]

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).