Article

Regression Models and Multi-Objective Optimization Using

the Genetic Algorithm Technique for an Integrated
Tillage Implement
Ganesh Upadhyay 1, * , Hifjur Raheman 2 and Rashmi Dubey 3

1 Department of Farm Machinery and Power Engineering, College of Agricultural Engineering and Technology,
CCS Haryana Agricultural University, Hisar 125004, Haryana, India
2 Department of Agricultural & Food Engineering, Indian Institute of Technology Kharagpur,
Kharagpur 721302, West Bengal, India; rahemanh@gmail.com
3 School of Engineering, University of Petroleum and Energy Studies, Dehradun 248007, Uttarakhand, India;
rashmirkd2012@gmail.com
* Correspondence: ganesh.upadhyay0@hau.ac.in

Abstract: This study presents an experimental and computational analysis of the specific
draft (SD) and specific torque (ST) requirements of an energy-efficient tillage implement,
the active–passive disk harrow (APDH). Soil bin trials were conducted to develop multiple
regression models predicting SD and ST based on operational parameters such as gang
angle (α), speed ratio (u/v), soil cone index, and working depth. Model’s accuracy was
assessed through statistical indices such as R2 , RMSE, MIE, and MAE. The high R2 and
low RMSE confirmed the reliability of the developed models in capturing the relationships
between input and output variables. A genetic algorithm-based multi-objective optimiza-
tion was implemented in MATLAB R2016a to determine optimal operational settings that
minimize total power consumption while maximizing soil pulverization. The optimized
values of α and u/v were determined to be in the ranges of 35.91◦ to 36.98◦ and 3.27 to 3.87,
respectively. Model validation with laboratory and field data demonstrated acceptable
prediction accuracy despite minor deviations attributed to soil variability and measure-
ment errors. The developed models provide a predictive framework for optimizing tillage
performance, aiding in tractor-implement selection, and enhancing energy efficiency in
Academic Editor: Marcello Biocca
agricultural operations.
Received: 6 March 2025
Revised: 8 April 2025 Keywords: genetic algorithm; active-passive tillage; regression models; soil bin; model validation
Accepted: 8 April 2025
Published: 11 April 2025

Citation: Upadhyay, G.;

Raheman, H.; Dubey, R. Regression 1. Introduction
Models and Multi-Objective
Optimization Using the Genetic
Optimizing the design and working parameters of agricultural implement is essential
Algorithm Technique for an Integrated for improving performance, production efficiency, and cultivation quality, while also min-
Tillage Implement. AgriEngineering imizing fuel consumption and its adverse environmental effects. Operating implements
2025, 7, 121. https://doi.org/ with optimum settings can enhance soil tilth quality with minimal energy use [1–3]. Nu-
10.3390/agriengineering7040121
merous studies have emphasized the significant energy requirements of tillage equipment
Copyright: © 2025 by the authors. during field operations, highlighting the importance of efficient energy management to
Licensee MDPI, Basel, Switzerland. reduce fuel consumption and costs.
This article is an open access article
Estimation models provide the ability to estimate the draft and torque requirements
distributed under the terms and
of tillage tools using minimal data, reducing the reliance on cost-intensive and time-
conditions of the Creative Commons
Attribution (CC BY) license
consuming field experiments. These models also aid in enhancing tool design by analyzing
(https://creativecommons.org/ and comparing various factors that influence draft requirements. In recent years, soft com-
licenses/by/4.0/). puting techniques have gained widespread recognition for their accuracy and efficiency.

AgriEngineering 2025, 7, 121 https://doi.org/10.3390/agriengineering7040121

AgriEngineering 2025, 7, 121 2 of 17

Predictive modeling approaches, such as regression models, response surface methodology,

and artificial neural networks (ANNs), have delivered promising results in agricultural
engineering [4–6]. These techniques enable the analysis and optimization of complex
agricultural processes by identifying relationships between input parameters and desired
outcomes. Regression models provide a straightforward statistical framework for predict-
ing system behavior based on historical data, while RSM facilitates the optimization of
multiple variables simultaneously to enhance performance. ANNs, inspired by the human
brain’s neural structure, offer a powerful machine learning-based approach capable of
capturing nonlinear patterns and improving predictive accuracy.
The sophisticated and stochastic tool–soil interaction, combined with the limitations
of traditional methods like linear regression, calls for advanced optimization techniques.
Among these, the genetic algorithm (GA) has been established as a resilient global opti-
mization method. By simulating the process of natural selection, the GA efficiently explores
large solution spaces, making it particularly well suited for addressing complex, nonlin-
ear problems, such as optimizing the operational parameters of agricultural machinery.
The GA has been widely adopted in agricultural research for tasks such as tool perfor-
mance optimization and energy management due to its ability to handle both discrete and
continuous systems effectively [7,8]. In a multi-objective GA, a population of potential
solutions evolves over multiple generations. The algorithm evaluates each solution based
on objective functions and applies Pareto dominance principles to identify a set of optimal
trade-offs, known as the Pareto front. The final solution set allows decision-makers to
choose an optimal configuration based on specific priorities. GAs outperform ANNs in
optimization tasks due to their ability to explore complex, multi-modal solution spaces
without requiring gradient information. Unlike ANNs, which rely on large datasets and
risk overfitting, GAs use evolutionary principles to find optimal solutions efficiently. They
are more interpretable, work well for multi-objective optimization, and do not require
predefined problem structures [9–12].
Researchers have utilized various approaches, including regression models, ANNs,
GAs, and hybrid techniques like genetic particle swarm optimization (GAPSO), to model
and optimize the operational parameters of tillage and seeding machinery [12–16]. GAPSO
leverages the GA’s ability to explore a diverse set of solutions and PSO’s fast exploitation
capability, leading to improved global optimization performance. The integration of pre-
dictive models with optimization tools has been successfully implemented across multiple
disciplines [17–20]. The GA’s standalone capability to optimize parameters, coupled with
its flexibility and computational efficiency, offers a distinct advantage over alternative
techniques. In an optimization problem, one is interested in minimizing the undesirable
effect(s) and maximizing the desirable effect(s). Multi-objective optimization (MOO) us-
ing the GA involves optimizing a number of objectives simultaneously and leveraging
resources efficiently to ensure the best possible outcomes. In MOO, the final solution of the
objective functions reflects a trade-off between multiple objectives, which may be entirely
conflicting, partially conflicting, or non-conflicting [21–23]. Multi-objective evolutionary
algorithm, i.e., genetic algorithm-based search for an optimal solution, among other meta-
heuristic approaches is among the most widely used stochastic search methods for solving
multi-objective optimization problems and is used in this study for performance optimiza-
tion of an integrated tillage implement. It is essentially a combined active–passive disk
harrow (APDH), featuring a front gang driven by the tractor PTO and a free-rolling rear
gang. Its benefits include enhanced soil tilth with fewer field passes, improved penetration
capability, reduced draft, reduced wheel slippage, and better utilization of the tractor’s
engine power [2,3,16,24].
AgriEngineering 2025, 7, 121 3 of 17
AgriEngineering 2025, 7, x FOR PEER REVIEW 3 of 18

In this study, experimental data from soil bin trials were utilized to develop regression
models
improved topenetration
predict specific draftreduced
capability, and specific torque of
draft, reduced APDH
wheel basedand
slippage, onbetter
inpututili-
parameters
such
zationasofgang angle, engine
the tractorʹs speed power
ratio, [2,3,16,24].
soil cone index (CI), and working depth. Subsequently,
In this study,
a GA-based experimental
optimization data fromwas
framework soilimplemented
bin trials were to
utilized to develop
identify regres-combina-
the optimal
sion models to predict specific draft and specific torque of APDH based
tions of parameters that minimize energy expenditure while maximizing performance. on input param-
etersfindings
The such as are
gangintended
angle, speed ratio,
to offer soil cone
practical index (CI),
guidance anddesign
for the workinganddepth. Subse-of active–
operation
quently, a GA-based optimization framework was implemented to identify the optimal
passive tillage machinery, promoting energy efficiency and environmentally sustainable
combinations of parameters that minimize energy expenditure while maximizing perfor-
agricultural practices.
mance. The findings are intended to offer practical guidance for the design and operation
of active–passive tillage machinery, promoting energy efficiency and environmentally
2. Materials and Methods
sustainable agricultural practices.
2.1. Indoor Soil Bin and Test Setup
2. Materials
Research and
was Methods
conducted in an indoor soil bin measuring 15 m in length, 1.80 m in
2.1. Indoor
width, andSoil Bin m
0.60 andinTest Setupwith an integrated active–passive offset-type disk harrow
depth
(APDH). The was
Research bin consisted
conductedof in aan
soil bin track,
indoor soil
soil bin processing
measuring 15 mcarriage, gear
in length, 1.80reduction
m in unit
width, and 0.60 m in depth with an integrated active–passive offset-type
of a power tiller for forward/reverse movement, wire rope unit, an instrument panel, and disk harrow
a(APDH). The binsystem
data logging consisted of The
[16]. a soilsoil
bin processing
track, soil processing
carriage carriage,
(Figure 1) gear reduction
helped unit
in preparing the
of a power tiller for forward/reverse movement, wire rope unit, an instrument
test bed at different soil compaction levels. It comprised a rotavator powered by a 3.75 kW, panel, and
a data logging system [16]. The soil processing carriage (Figure 1) helped in preparing the
three-phase, and 1420 rpm induction motor, a leveling blade, and a roller for tilling, leveling,
test bed at different soil compaction levels. It comprised a rotavator powered by a 3.75
and compacting the soil, respectively. The APDH (Figure 2) featured 03 powered disks in
kW, three-phase, and 1420 rpm induction motor, a leveling blade, and a roller for tilling,
the front and 03 free-rolling disks in the rear set of gangs. The disk diameter is 510 mm
leveling, and compacting the soil, respectively. The APDH (Figure 2) featured 03 powered
with
disksainconcavity of 6003mm
the front and and a spacing
free-rolling disks inof the225
rearmm
set between
of gangs. disks.
The diskA diameter
10 hp, three-phase
is
motor
510 mm was used
with to transmit
a concavity of 60power
mm and to the front gang.
a spacing of 225The
mmdesigned APDH
between disks. A had a working
10 hp,
width of 630motor
three-phase mm, with its operating
was used to transmit depth
power becoming adjustable
to the front gang. Theusing a hydraulic
designed APDH cylinder.
The
had angle of the
a working front-powered
width of 630 mm, with disksits (α) could depth
operating be adjusted
becoming by adjustable
altering the angle
using a of the
hydraulic
shafts cylinder. The
transmitting angle of
motion the front-powered
through U-joints and disksutilizing
(α) couldabeseries
adjusted by altering
of holes in the frame
the angle
while of the shaftsthe
repositioning transmitting
mountedmotionbearings. through
The U-joints
frame ofand theutilizing
APDH awas series of holes
fitted with slotted
in the frame while repositioning the mounted bearings. The frame
channels (C-shaped), allowing it to be attached to the soil processing carriage. of the APDH was fitted
with slotted channels (C-shaped), allowing it to be attached to the soil processing carriage.

AgriEngineering 2025, 7, x FOR PEER REVIEW 4 of 18

Figure 1.
Figure 1. Soil
Soilprocessing
processingcarriage inside
carriage soil soil
inside bin. bin.

Figure 2.
Figure Views of
2. Views of integrated
integrated active–passive
active–passive offset
offset type
type disk
disk harrow
harrow (APDH).

2.2. Experimental Plan

The soil type was sandy clay loam with particle density 2650 kg m−3, angle of internal
friction 22°, and was taxonomically grouped as Alfisol (Oxyaquic haplustalf). The proto-
AgriEngineering 2025, 7, 121 4 of 17

2.2. Experimental Plan

The soil type was sandy clay loam with particle density 2650 kg m−3 , angle of internal
friction 22◦ , and was taxonomically grouped as Alfisol (Oxyaquic haplustalf). The prototype
of the considered implement was tested in the soil bin at different α (25◦ , 30◦ , 35◦ , and 40◦ ),
speed ratios (2.4, 3.0, 3.6, and 4.6), cone index (CI) (0.50, 0.80, and 1.10 MPa), and at different
depths (100, 120, and 140 mm). Speed ratio (u/v) is the ratio between the peripheral speed
of the disk (u) to the forward speed of the machine (v). In this study, forward speed was
maintained at 0.89 m s−1 , the maximum permissible speed in the bin, while peripheral
speed was adjusted by changing the disk speed between 80 and 150 revolutions per minute
to obtain the required speed ratios. All of these mentioned parameters influence the power
requirements of actively rotating disks [25–27].

2.3. Determination of Soil and Operating Parameters

The data on soil moisture content were taken using an infrared moisture meter.
A hydraulically operated soil cone penetrating device was used to record the penetration
resistance (or soil CI) as per ASAE Standards [28]. The forward speed was measured using
a proximity sensor attached to the processing carriage. The draft requirement was recorded
using a 2000 kg load transducer (S-type) positioned between the processing carriage and
implement. The torque requirement of APDH was measured with a 1 kN m capacity torque
sensor. The data of all sensors used were fed to a data logging unit (HBM-Spider8) and
then saved to a laptop.

2.4. Test Procedure

The soil bed was formed using the soil processing carriage prior to each experiment.
A rotavator was employed to relieve soil stresses before the experiments, subsequently
accompanied by water sprinkling to achieve the desired moisture of 9–11% (db). The soil
was evened out with a leveler and compressed with a hydraulic roller to attain the desired
soil CI. Tests were conducted after verifying the uniformity of the soil bed (p ≤ 0.01) by
recording moisture and the CI. Tests were conducted over a seven-meter track in the central
section of the soil bed with each test repeated three times. Using the load transducer, torque
sensor, and proximity sensor, draft, torque, and operating speed data were continuously
recorded. After the operation, the CI of the tilled soil (CItilled ) was also measured at five
points about one-meter distance apart along the test bed. After each test, the bed was
disturbed and re-prepared using the same procedure for subsequent tests.

2.5. Statistical Analysis and Development of Regression Models

The experimental design was fully randomized with 03 replications for each parameter.
ANOVA and post hoc analysis using Duncan’s test were performed to assess differences
between treatment means. SPSS software (version 22.0) was used to perform regression
analysis using the multiple regression method. Curve fitting was applied to analyze the
relationship between independent (CI, u/v ratio, depth, and α) and dependent (specific
draft and specific torque) parameters and these relationships were incorporated into the
ultimate regression model. The regression models were fitted in an iterative manner to
the data using nonlinear least-squares optimization following the Levenberg–Marquardt
algorithm. To refine the model, stepwise or backward elimination regression methods were
applied to identify the most relevant predictors. Residual analysis was also conducted to
check for heteroscedasticity and ensure the assumption of normality in residuals, which val-
idates the model’s reliability. If issues arise, transforming variables or applying interaction
terms might improve model accuracy. Interactions between variables were excluded in the
development models, as their inclusion added complexity without significantly enhancing
AgriEngineering 2025, 7, 121 5 of 17

accuracy. To evaluate model efficiency, statistical indices: mean index error (MIE), variance
account for (VarAF), root mean square error (RMSE), R2 , and mean absolute error (MAE)
were calculated.
∑ n ( E − Mi )
MIE = i=1 i (1)
n
Variance (Mi ) − Variance (Ei )
 
VarAF = 1 − × 100 (2)
Variance (Mi )
v( )
u ∑n (Ei − Mi )2
u
2 i=1
RMSE = t (3)
n

2
∑ni=1 (Mi − Ei )
R2 = 1 − 2
(4)
∑ni=1 (Mi − Ai )
∑ni=1 |(Mi − Ei ) / Mi |
MAE = × 100 (5)
n
MAD = max [(E i − Mi )/ Ei ] × 100 (6)

where Ei , Mi , and Ai are the values of estimated, measured, and average specific draft
or specific torque, and n is the number of observations. MIE quantifies the non-diagonal
estimation and should ideally be close to zero for an accurate model. A VarAF of 100%
signifies that the developed model perfectly estimated the measured output, while a VarAF
of 0% indicates poor estimation accuracy. The RMSE is the standard deviation of the
residuals reflecting the prediction power of the model. If the VarAF is 100% and the RMSE
is zero, the model would be considered perfect, as it would indicate that the predicted
values exactly match the measured data with no deviation. R² is widely used to assess
goodness of fit. The MAE, denoting accuracy in percent, was also used to compare the
model’s estimated performance.

2.6. Optimization of Operating Parameters

Specific draft and specific torque requirement estimation models were developed
using multiple regression technique to ultimately predict the total power requirement (Pe).
A model was also developed to predict the CI of the tilled soil (CItilled ). These developed
estimation models were used to carry out multi-objective optimization (MOO) in MATLAB
software to determine the optimum ‘α’ and speed ratio before developing a field prototype
with the aim to carry out tillage with minimum Pe and maximum reduction in CItilled .
A MOO problem for achieving optimum tillage performance with minimum power
consumption was formulated for the implement operating in three different soil conditions
(CI values of 500 ± 30, 800 ± 30, and 1100 ± 30 kPa representing soft, tilled, and firm soil,
respectively) as discussed earlier. Five independent variables (decision variables) namely
α (degrees), operating depth (mm), forward speed (km h−1 ), front gang axle rpm (NFGA ),
and soil CI (MPa) denoted, respectively, as X1 , X2 , X3 , X4 , and X5 were considered such that
the vector of decision variables is as given in Equation (7).

X = {X1 , X2 , X3 , X4, X5 } (7)

The optimization problem of interest had two objective functions which were mini-
mizing the estimated total power requirement of the tractor (Pe) and minimizing the cone
index of the tilled soil (CItilled ). Two functions were defined, f1 (X) for Pe and f2 (X) for CItilled
given in the results section. With these functions defined, the optimization problem was
formulated (Equation (8)):  
min
fi (X), i = 1, 2. (8)
X ϵXn
AgriEngineering 2025, 7, 121 6 of 17

where Xi is the decision variable space limited by upper and lower bounds (lb ≤ X ≤ ub ;
lb for lower bound and ub for upper bound) on the decision variables. The optimization
problem under consideration was constrained by upper and lower bounds for decision
variables and are expressed by Equations (9)–(13).

20 ≤ X1 ≤ 45 (9)

80 ≤ X2 ≤ 140 (10)

3 ≤ X3 ≤ 6 (11)

80 ≤ X4 ≤ 250 (12)

0.8 ≤ X5 ≤ 1.1 (13)

The MOO problem was solved using the ‘gamultiobj’ function in MATLAB, which
efficiently finds a set of Pareto-optimal solutions for conflicting objectives. Since the GA
is inherently a minimization algorithm, the maximum reduction in CItilled objective is
transformed into a minimization problem by negating it. To implement this, an objective
function file is created, defining Pe and CItilled as functions of key variables. The optimiza-
tion problem is then constrained within defined upper and lower bounds to ensure practical
feasibility. The ‘gamultiobj’ function is then executed with these constraints, generating a
set of Pareto-optimal solutions, where no single solution is universally better—trade-offs
exist between reducing energy consumption and improving soil conditions.
To analyze the results, a Pareto front is plotted, visually depicting the trade-offs
between power requirement and soil improvement. The decision-maker can select an
optimal solution based on operational priorities. This approach ensures an efficient and
data-driven selection of tillage parameters, leveraging the GA’s ability to explore complex
solution spaces. MATLAB’s GA-based MOO approach provides a robust framework for
optimizing agricultural processes, balancing competing objectives for sustainable and
energy-efficient tillage operations.

2.7. Validation of the Developed Regression Models

The specific draft and torque requirement estimation models were validated using
data collected during a separate set of soil bin tests, distinct from those used for model
development and from the field data collected during tillage operation. The maximum
absolute deviation (MAD) quantified the maximum deviation of model outputs from the
data recorded from soil bin and field tests.

3. Results and Discussion

The APDH test rig was operated in the soil bin at different α, u/v settings, CI, and
working depths and the results are presented in Figures 3 and 4. The effect of these operating
parameters on specific draft (SD) and specific torque (ST) requirements is discussed below.
 3. Results and Discussion
The APDH test rig was operated in the soil bin at different α, u/v settings, CI, and
working depths and the results are presented in Figures 3 and 4. The effect of these oper-
AgriEngineering 2025, 7, 121 7 of 17
ating parameters on specific draft (SD) and specific torque (ST) requirements is discussed
below.

AgriEngineering 2025, 7, x FOR PEER REVIEW 8 of 18

Figure 3. Variations in specific draft requirement across different operating variables.
Figure 3. Variations in specific draft requirement across different operating variables.

Figure4.4. Variations
Figure Variations in
in specific
specific torque requirement
requirement across
across different
differentoperating
operatingvariables.
variables.

3.1. Impact of Operating Parameters on Specific Draft (SD) Requirement

ANOVA results revealed that the individual effects of all independent variables on
the SD requirement were statistically significant at the 5% significance level. The F-values
indicated that α had the greatest influence on the SD requirement, followed by the CI, u/v
ratio, and depth of operation. The findings indicate that the SD of the APDH decreased
AgriEngineering 2025, 7, 121 8 of 17

3.1. Impact of Operating Parameters on Specific Draft (SD) Requirement

ANOVA results revealed that the individual effects of all independent variables on
the SD requirement were statistically significant at the 5% significance level. The F-values
indicated that α had the greatest influence on the SD requirement, followed by the CI, u/v
ratio, and depth of operation. The findings indicate that the SD of the APDH decreased with
an increasing u/v ratio, with the most significant reduction occurring up to a u/v of 3.6,
beyond which any increase in rotational speed yields negligible benefits. This suggests that
increasing disk speed improves soil pulverization and reduces draft until a threshold where
the soil shear strength reaches a minimum. Duncan’s test confirms significant differences
in mean SD values across different u/v ratios at a 5% significance level, reinforcing that
optimal performance is achieved at a u/v of around 3.6.
The study reveals that the SD of the APDH increased with an increase in gang angle
(α), with a moderate rise up to 35◦ followed by a sharp increase beyond this point. This
trend is attributed to the decreasing rear-side bearing area and increasing soil volume
managed by the disks. Duncan’s test confirms significant differences in mean SD values
across different α levels at a 5% significance level, indicating that higher gang angles result
in increased draft requirements, particularly beyond the critical threshold of 35◦ .
The findings indicate that the SD of the APDH decreased with an increment in working
depth. This reduction, ranging from 1.53% to 19.27%, is attributed to the relatively lower
increase in the draft at greater depths due to the enhanced pushing force generated by the
front set of powered disks and a higher increase in the soil displacement area. These results
suggest that deeper operations may improve efficiency by reducing the SD requirement.
The study shows that the SD of the APDH increased with higher soil CI due to greater
soil resistance. The mean SD rose significantly as soil CI increased from 0.5 to 1.1 MPa,
with a percentage increase ranging from 42.18% to 64.50% across different gang angles at
a working depth of 120 mm. Comparable trends were found at 100 and 140 mm depths,
confirming that higher soil resistance leads to increased draft requirements, emphasizing
the need for appropriate operational adjustments in denser soils.

3.2. Impact of Operating Parameters on Specific Torque (ST) Requirement

ANOVA results revealed that the individual effects of all independent variables on
the ST requirement were statistically significant at the 5% significance level. The F-values
indicated that CI had the greatest influence on the ST requirement, followed by u/v ratio,
α, and depth of operation. The findings indicate that the ST requirement of the APDH
decreased with an increasing u/v ratio, with the most significant reduction occurring
up to a u/v of 3.6; thereafter, rotational speed has a negligible effect. This reduction is
attributed to lower cutting force requirements and frequent engagement with already tilled
soil at higher rotational speeds. Duncan’s test confirms no significant difference in ST
between u/v ratios of 3.6 and 4.6, suggesting that increasing the rotational speed beyond
3.6 provides no additional torque-related benefits.
The results show that the ST requirement of the APDH decreased with increasing α,
with a sharp reduction observed from 25◦ to 35◦ and a smaller reduction beyond 35◦ . This
trend is attributed to the lesser frictional force resulting from the decreasing contact area
between the rear side of the disks and the furrow wall. However, an increase in the volume
of soil managed beyond 35◦ counteracts this effect, leading to a smaller reduction in ST.
Duncan’s test confirms significant differences in ST for different α values, reinforcing the
impact of disk angle on torque requirements.
The findings suggest that the ST requirement of the APDH decreased with an increase
in operating depth across all tested conditions. The reduction is attributed to a compara-
tively lesser increase in torque demand alongside a greater increase in soil disturbance area
AgriEngineering 2025, 7, 121 9 of 17

at higher depths. However, the absolute torque requirement increased with depth due to
the greater amount of soil handled and higher penetration resistance. These results align
with previous studies by [25,29], reinforcing the observed trend across different soil types.
The results indicate that the ST requirement of the APDH increased with an increase
in soil CI. This is due to the greater soil resistance associated with greater CI values. The
trend was consistent across different α values, u/v ratios, and working depths. Notably,
the percentage increase in ST was more pronounced at higher α values, with the highest
increase (76.82% to 98.62%) observed at α of 40◦ . These findings suggest that soil com-
paction significantly impacts the torque demand, which is crucial for optimizing APDH
performance in varying soil conditions.

3.3. Specific Draft and Specific Torque Estimation Models

To develop regression models, the Levenberg–Marquardt algorithm was followed
and the interactions between operational variables were excluded as it increased the
model sophistication without improving its prediction power. A multi-collinearity test
was conducted using tolerance and variance inflation factor (VIF) as signs of possible
interdependencies among the operational variables, revealing no sign of multicollinearity
as the VIF value was calculated to be one. The high VIF values (typically >10) suggest
strong correlations between independent variables, potentially distorting the regression
estimates. The developed models are discussed below.
Soil CI and operating depth were found to have a linear relationship with both specific
draft and specific torque requirements. However, α had a quadratic relationship, and the
speed ratio had a negative exponential relationship with both specific draft and specific
torque requirements. The developed models for estimating specific draft (Equation (14))
and specific torque (Equation (15)) requirements are as follows.
u
SDAPDH = C0 + C1 × α + C2 × α2 + C3 × CI+ C4 × d+C5 × e(− v ) (14)
u
STAPDH = C0 + C1 × α + C2 × α2 + C3 × CI+C4 × d+C5 × e(− v ) (15)

where SDAPDH is in kN m−2 ; STAPDH is in kN-m m−2 ; α is in degrees; CI is in MPa; d

is the working depth in mm; u/v is the speed ratio; v is in km h−1 ; and Ci = regression
coefficients whose values are given in Table 1 with standard errors for both SD and ST
models, i = 0, 1, 2, 3, 4, 5.

Table 1. Regression coefficients and standard error (SE) for the developed models.

SDAPDH Model (Equation (14)) STAPDH Model (Equation (15))

Regression (R2 = 0.93; RMSE = 1.43) (R2 = 0.91; RMSE = 0.43)
Coefficients
Estimate SE Estimate SE
C0 35.665 2.912 9.222 0.849
C1 −2.355 0.180 −0.323 0.052
C2 0.046 0.003 0.003 0.001
C3 11.763 0.282 3.506 0.082
C4 −0.044 0.004 −0.015 0.001
C5 82.876 2.285 24.114 0.666

The total power requirement (Pe) may be predicted from the SDAPDH and STAPDH
models as follows (Equation (16)):

SD ×v 2 × π ×N × ST
 
Pe = + × Asd (16)
3.6 × ηPTO to DB 60 × ηtrans
AgriEngineering 2025, 7, 121 10 of 17

where Pe is the total power requirement of the tractor in kW; SD is the specific draft in kN
m−2 ; v is in km h−1 ; ST is in kN-m m−2 ; Asd is the area of soil disturbance in m2 ; N is the
rpm of powered disks; ηPTO to DB is the transmission efficiency from PTO to drawbar [30];
and ηtrans is the transmission efficiency from PTO to the front gang.
The results of regression analysis for the specific draft and specific torque estimation
models are given in Table 2.

Table 2. Results of regression analysis for the developed SDAPDH and STAPDH models.

SDAPDH Model STAPDH Model

(R2 = 0.93; RMSE = 1.43) (R2 = 0.91; RMSE = 0.43)
Source
Sum of Mean Sum of Mean
DoF DoF
Squares Square Squares Square
Regression 137,837.7 6 22,972.9 8295.1 6 1382.5
Residual 875.6 426 2.05 74.3 426 0.17
Uncorrected
138,713.3 432 - 8369.5 432 -
total
Corrected
13,500.4 431 - 858.6 431 -
total

3.4. Estimation Model for Cone Index of the Tilled Soil (CItilled )
The average values of the cone index of the tilled soil (CItilled ) obtained during soil
bin tests were analyzed and were used to develop a model for estimating CItilled using the
multiple regression technique. The aim of developing the model for CItilled (Equation (17))
was to use this model for carrying out the multi-objective optimization. In general, the
CItilled obtained after the operation was found to decrease with an increase in α, u/v,
and operating depth at all tested soil cone indices. The ANOVA results indicated that
all variables had a significant effect on the CItilled at a 1% level of significance with the
initial soil CI having the greatest influence on the CItilled followed by u/v, depth, and α
in that order.
Soil CI, working depth, and α were found to have a linear relationship while speed
ratio was found to have a negative exponential relationship with CItilled . The developed
model has high R2 and low RMSE of 0.97 and 0.032, respectively. The results of regression
analysis for the developed CItilled model are given in Table 3. The estimated values of
regression coefficients and their standard error for the developed CItilled model are given
in Table 4.
u
CItilled = C0 + C1 × α + C2 × CIbefore + C3 × d+C4 × e(− v ) (17)

where CItilled is in MPa; α is in degrees; CIbefore is the cone index of the soil before tillage in
MPa; d is in mm; u/v is the speed ratio; and Ci = regression coefficients, i = 0, 1, 2, 3, 4.

Table 3. Results of regression analysis for the developed CItilled model.

CItilled Model (R2 = 0.97; RMSE = 0.032)

Source
Sum of Squares DoF Mean Square
Regression 63.86 5 12.77
Residual 0.10 427 0.00
Uncorrected total 63.96 432
Corrected total 3.63 431
AgriEngineering 2025, 7, 121 11 of 17

Table 4. Regression coefficients and standard error (SE) for the CItilled model.

Regression CItilled Model (R2 = 0.97; RMSE = 0.032)

Coefficients Estimate SE
C0 0.248 0.008
C1 −0.003 0.0001
C2 0.344 0.003
C3 −0.001 0.00004
C4 0.849 0.025

3.5. Statistical Indices for the Developed Models

The statistical indices for the specific draft and specific torque models were calculated
using Equations (1)–(5) for analyzing the efficiency of developed models, and the results
are given in Table 5. The indices calculated for the developed CItilled model are also given in
Table 5. The high VarAF and R2 , combined with low MIE, RMSE, and MAE, confirmed the
accuracy of all developed models and demonstrated that the selected parameters effectively
explained the majority of the variability in the test data.

Table 5. Statistical indices for the developed models.

Model MIE RMSE VarAF MAE R2

SDAPDH (Equation (14)) −0.06 1.43 92.80 6.98 0.93
STAPDH (Equation (15)) −0.11 0.43 92.83 8.92 0.91
CItilled (Equation (17)) −0.028 0.032 98.96 7.99 0.97
Note: MIE = mean index error, VarAF = variance accounted for, RMSE = root mean square error, MAE = mean
absolute error; The units of both MEI and RMSE for the SDAPDH , STAPDH , and CItilled models are kN m−2 , kN-m
m−2 , and MPa, respectively.

3.6. Multi-Objective Optimization to Obtain the Optimum Combination of α and Speed Ratio
The multi-objective optimization (MOO) based on the genetic algorithm approach
was performed in MATLAB to obtain the optimized values of α and speed ratio that
satisfied the objective functions (i.e., minimizing Pe (Equation (16)) and minimizing CItilled
(Equation (17)). Pareto fronts or sets of non-dominated solutions obtained for Pe vs. CItilled
at some set operating conditions (CIbefore , operating depth, and forward speed) are shown
in Figure 5. In this figure, moving from one solution to another in the Pareto set involves
trade-offs between multiple objectives. One optimal solution among the possible solutions
can be selected on the Pareto front based on the choice/preference of the decision-maker
for any particular condition.
The numerical solutions of MOO obtained for different operating conditions are listed
in Table 6. In the present study, the aim was to test the implement in the field at the
optimum settings of α and speed ratio obtained from the results of MOO. Based on soil bin
data and following the MOO technique using the genetic algorithm, the optimum values
of α and speed ratio were found to vary in the range of 35.91◦ to 36.98◦ and 3.27 to 3.87,
respectively. Considering the optimum range of these variables, the field prototype of the
APDH was designed and developed by fixing α at 36◦ , whereas the speed ratio was varied
in the field. Further, it was noticed that with an increase in the forward speed from 3 to
6 km h−1 , the corresponding NFGA needed to be increased from 92.9 rpm to 220.1 rpm to
achieve optimized values of objective functions (i.e., minimum Pe with maximum reduction
in CItilled values) for most of the non-dominated solutions.
 at some set operating conditions (CIbefore, operating depth, and forward speed) are shown
in Figure 5. In this figure, moving from one solution to another in the Pareto set involves
trade-offs between multiple objectives. One optimal solution among the possible solutions
AgriEngineering 2025, 7, 121 12 of 17
can be selected on the Pareto front based on the choice/preference of the decision-maker
for any particular condition.

Figure 5.
Figure Some Pareto
5. Some Pareto fronts
fronts obtained
obtained through
through MOO
MOO using
using GA
GA in
in Matlab
Matlab for
for estimated
estimated equivalent
equivalent
PTO power (Pe) and CI tilled
PTO power (Pe) and CItilled. .

Table 6. Results of multi-objective optimization performed in MATLAB.

The numerical solutions of MOO obtained for different operating conditions are
listed in Table Variables
Design 6. In the present study, the aim was to test the implement
Objectivein Functions
the field at the
CIbefore , Operating Forward
optimumSpeed, α, speedNratio
settings of α and FGA , obtained
Estimated
from the results of MOO.
−1 Pe, kW CIBased on soil
tilled , MPa
MPa Depth, mm km h Degree rpm Speed Ratio
bin data and following the MOO technique using the genetic algorithm, the optimum val-
0.8 100 ues of3.00
α and speed ratio 35.97 99.7 to vary in 3.51
were found 8.89 to 36.98° and
the range of 35.91° 0.34 3.27 to
0.8 120 3.87, respectively. Considering the optimum range of these variables, the field0.32
3.00 36.29 104.5 3.68 9.89 prototype
0.8 140 3.00 36.63 110.0 3.87 10.59 0.29
of the APDH was designed and developed by fixing α at 36°, whereas the speed ratio was
1.1 100 3.00 36.98 98.6 3.47 10.94 0.44
1.1 120 varied3.00
in the field. Further,
36.78 it was noticed
98.3 that with an increase 12.37
3.46 in the forward speed
0.42 from
1.1 140 3 to 6 3.00
km h−1, the corresponding
35.98 N 92.9
FGA needed to be
3.27increased from
13.5392.9 rpm to 220.1 rpm
0.41
0.8 100 to achieve
4.00 optimized 35.98values of objective
132.9 functions (i.e., minimum
3.51 11.86Pe with maximum
0.34 re-
0.8 120 duction4.00 36.29
in CItilled values) for most 139.2 3.67 solutions.
of the non-dominated 11.90 0.35
0.8 140 4.00 36.64 146.6 3.87 14.12 0.29
1.1 100 4.00 36.86 135.3 3.57 14.67 0.44
1.1 120 4.00 35.91 125.4 3.31 16.43 0.43
1.1 140 4.00 36.50 126.9 3.35 18.05 0.41
0.8 100 5.00 35.97 166.0 3.50 14.82 0.34
0.8 120 5.00 36.28 174.1 3.67 16.48 0.32
0.8 140 5.00 36.63 183.4 3.87 17.65 0.29
1.1 100 5.00 36.33 164.7 3.48 18.25 0.44
1.1 120 5.00 36.44 157.7 3.33 20.55 0.43
1.1 140 5.00 36.52 160.1 3.38 22.57 0.41
0.8 100 6.00 36.00 200.1 3.52 17.79 0.34
0.8 120 6.00 36.29 209.0 3.68 19.78 0.32
0.8 140 6.00 36.63 220.1 3.87 21.18 0.29
1.1 100 6.00 36.82 188.0 3.31 21.76 0.45
1.1 120 6.00 36.23 196.9 3.46 24.76 0.42
1.1 140 6.00 36.19 187.8 3.30 27.07 0.41

3.7. Validation of the Regression Models from the Soil Bin Data
The developed models for estimating SD and ST were validated using data collected
during a separate set of soil bin tests, distinct from those used for model development. The
rotational speed of the disk, α, depth, and forward speed was varied randomly between 80
AgriEngineering 2025, 7, 121 13 of 17

and 150 rpm, 25◦ to 40◦ , 60 to 140 mm, and 0.8 to 3.2 km h−1 , respectively, at different cone
indices of soil (0.65 to 1.2 MPa). The observed and estimated values of the SD and ST are
compared in Figures 6a and 6b, respectively. It is evident that the slope of the best-fit line
is close to unity with high R2 for all the models developed. The MAD between observed
and estimated values of parameters was found to be 18.50% and 13.11% for the SD and
ST, respectively. These variations are considered acceptable because of the variations that
existed in preparing the soil bed with the desired CI and also the experimental14errors
AgriEngineering 2025, 7, x FOR PEER REVIEW of 18
incurred while measuring the draft and torque.

(a) Comparison of the measured and estimated specific draft of the APDH (SDAPDH).

(b) Comparison of the measured and estimated specific torque of the APDH (STAPDH).

Figure
Figure6.6.Validation
Validationof
ofthe
thedeveloped
developedregression
regression models
models with
with the
the soil
soil bin
bin data.
data.

3.8.Validation
3.8. Validationofofthe
theDeveloped
DevelopedRegression
RegressionModels
Modelsfrom
fromthe
theField
FieldData
Data
Theestimation
The estimationmodels
modelsdeveloped
developed from
from soilsoil
binbin
datadata were
were alsoalso validated
validated against
against the
the mean values of draft and torque data collected from the field tests. The
mean values of draft and torque data collected from the field tests. The observed and es- observed
and estimated
timated values of values
the SDof and
the SD
ST and ST requirements
requirements are compared
are compared in Figurein7a
Figures
and 7b,7arespec-
and 7b,
respectively. It can be seen in these figures that the slope of the best-fitted line is
tively. It can be seen in these figures that the slope of the best-fitted line is close to unityclose to
unitygood
withRgood 2
R for
with 2 for both theboth the models
models developed.
developed. However, However, the developed
the developed modelsmodels were
were found
found to underestimate the SD and overestimate the ST requirement.
to underestimate the SD and overestimate the ST requirement. The MADs between ob- The MADs between
served and estimated values of the specific draft and specific torque requirement were
found to be 19.79% and 19.47%, respectively. The presence of loose and anchored crop
residues, variations in soil conditions in the field, and the experimental errors incurred
while measuring the draft and torque values could be the reasons behind the variations
between observed and estimated values.
AgriEngineering 2025, 7, 121 14 of 17

observed and estimated values of the specific draft and specific torque requirement were
found to be 19.79% and 19.47%, respectively. The presence of loose and anchored crop
residues, variations in soil conditions in the field, and the experimental errors incurred
AgriEngineering 2025, 7, x FOR PEER REVIEW 15 of 18
while measuring the draft and torque values could be the reasons behind the variations
between observed and estimated values.

(a) Comparison of the measured and estimated specific draft of the APDH (SDAPDH).

(b) Comparison of the measured and estimated specific torque of the APDH (STAPDH).

Figure
Figure7.7.Validation
Validationofofthe
thedeveloped
developedregression models
regression with
models thethe
with field data.
field data.

4.4.Conclusions
Conclusions
Thespecific
The specificdraft
draftand
andspecific
specific torque
torque requirement
requirement of of
thethe APDH
APDH waswas found
found to increase
to increase
withan
with anincrease
increaseininα αandand soil
soil CI,CI,
andand it decreased
it decreased with
with an increase
an increase in ratio
in u/v u/v ratio
at anyatαany
and operating
α and depth.
operating UsingUsing
depth. multiple regression
multiple technique,
regression models were
technique, modelsdeveloped incor-
were developed
porating the soilthe
incorporating CI,soil
front
CI,gang
frontangle,
gang operating depth, forward
angle, operating speed, and
depth, forward u/v and
speed, ratiou/v
fromratio
soil bin data to estimate the specific draft and torque requirements of the
from soil bin data to estimate the specific draft and torque requirements of the APDH inAPDH in sandy
clay loam
sandy soil
clay withsoil
loam maximum absolute variations
with maximum of 18.50%,of
absolute variations 13.11%,
18.50%, and 12.04%,and
13.11%, respec-
12.04%,
tively. Soil CI and operating depth were found to have a linear relationship
respectively. Soil CI and operating depth were found to have a linear relationship with both with both
specific
specificdraft
draftand
andspecific
specifictorque
torque requirements.
requirements. However,
However, α had a quadratic
α had a quadraticrelationship,
relationship,
and the speed ratio had a negative exponential relationship with both specific draft and
specific torque requirements. Based on soil bin data and following the MOO technique
using genetic algorithm, the optimum values of front gang angle and u/v ratio for the
APDH were found to vary in the range of 35.91° to 36.98° and 3.27 to 3.87, respectively.
Further, it was found that with an increase in the forward speed (3 to 6 km h−1), the corre-
AgriEngineering 2025, 7, 121 15 of 17

and the speed ratio had a negative exponential relationship with both specific draft and
specific torque requirements. Based on soil bin data and following the MOO technique
using genetic algorithm, the optimum values of front gang angle and u/v ratio for the
APDH were found to vary in the range of 35.91◦ to 36.98◦ and 3.27 to 3.87, respectively.
Further, it was found that with an increase in the forward speed (3 to 6 km h−1 ), the
corresponding NFGA of APDH needed to be increased (92.9 to 220.1 rpm) to achieve
optimized values of objective functions, i.e., minimum Pe and maximum reduction in
CItilled values. The field trials demonstrated that by carefully setting the gang angle and
the speed ratio, significant improvements in tillage efficiency were achieved, reducing
overall fuel consumption and improving soil conditions. The combination of multiple
regression modeling, GA-based multi-objective optimization, and field validation provided
a comprehensive approach to optimizing tillage operations.
The developed models for estimating specific draft and specific torque will offer
valuable insights to farmers and manufacturers regarding the necessary tractor power,
facilitating the optimal selection of tractor-implement combinations for their farms. The
findings of this research will serve as a useful resource for engineers and scientists involved
in farm machinery development and management, as well as for manufacturers and farmers
across India. Furthermore, this study provides a strong foundation for adopting active-
passive combination tillage implements in the Indian farming system, enhancing tractor
power utilization efficiency. Future studies involving diverse soil types and humidity levels
will aid in confirming the model’s general applicability.

Author Contributions: Conceptualization, G.U.; methodology, G.U. and H.R.; software, R.D.; valida-
tion, G.U.; formal analysis, G.U. and R.D.; investigation, G.U. and H.R.; resources, H.R.; data curation,
R.D.; writing—original draft preparation, G.U.; writing—review and editing, G.U.; visualization,
H.R.; supervision, H.R.; project administration, G.U. and H.R.; funding acquisition, H.R. All authors
have read and agreed to the published version of the manuscript.

Funding: This research received no external funding.

Data Availability Statement: The data are contained in the article.

Acknowledgments: The funding and cooperation received from MHRD, Govt. of India, to carry out
this research work is sincerely acknowledged.

Conflicts of Interest: The authors declare no conflicts of interest.

Abbreviations
The following abbreviations are used in this manuscript:
α Angle of the front powered disks
APDH Active–passive disk harrow
ASAE American Society of Agricultural Engineers
CI Cone index
Ci Regression coefficient
CItilled Cone index of the tilled soil
MAD Maximum absolute deviation
MIE Mean index error
MAE Mean absolute error
MOO Multi-objective optimization
Pe Total power requirement of the tractor
RMSE Root mean square error
SD Specific draft
ST Specific torque
u/v Speed ratio
VarAF Variance account for
AgriEngineering 2025, 7, 121 16 of 17

References
1. Mehta, C.R.; Singh, K.; Selvan, M.M. A decision support system for selection of tractor–implement system used on Indian farms.
J. Terramech. 2011, 48, 65–73. [CrossRef]
2. Upadhyay, G.; Raheman, H. Effect of velocity ratio on performance characteristics of an active-passive combination tillage
implement. Biosyst. Eng. 2020, 191, 1–12. [CrossRef]
3. Sarkar, P.; Upadhyay, G.; Raheman, H. Active-passive and passive-passive configurations of combined tillage implements for
improved tillage and tractive performance: A review. Span. J. Agric. Res. 2021, 19, e02R01. [CrossRef]
4. Kundu, P.; Paul, V.; Kumar, V.; Mishra, I.M. Formulation development, modeling and optimization of emulsification process
using evolving RSM coupled hybrid ANN-GA framework. Chem. Eng. Res. Des. 2015, 104, 773–790. [CrossRef]
5. Ranjbarian, S.; Askari, M.; Jannatkhah, J. Performance of tractor and tillage implements in clay soil. J. Saudi Soc. Agric. Sci.
2017, 16, 154–162. [CrossRef]
6. Askari, M.; Abbaspour-Gilandeh, Y.; Taghinezhad, E.; Hegazy, R.; Okasha, M. Prediction and optimizing the multiple re-
sponses of the overall energy efficiency (OEE) of a tractor-implement system using response surface methodology. J. Terramech.
2022, 103, 11–17. [CrossRef]
7. Damanauskas, V.; Velykis, A.; Satkus, A. Efficiency of disc harrow adjustment for stubble tillage quality and fuel consumption.
Soil Tillage Res. 2019, 194, 104311. [CrossRef]
8. Fawzi, H.; Mostafa, S.A.; Ahmed, D.; Alduais, N.; Mohammed, M.A.; Elhoseny, M. TOQO: A new tillage operations quality
optimization model based on parallel and dynamic decision support system. J. Clean. Prod. 2021, 316, 128263. [CrossRef]
9. Katoch, S.; Chauhan, S.S.; Kumar, V. A review on genetic algorithm: Past, present, and future. Multimed. Tools Appl.
2021, 80, 8091–8126. [CrossRef]
10. Kotyrba, M.; Volna, E.; Habiballa, H.; Czyz, J. The influence of genetic algorithms on learning possibilities of artificial neural
networks. Computers 2022, 11, 70. [CrossRef]
11. Sexton, R.S.; Dorsey, R.E.; Johnson, J.D. Optimization of neural networks: A comparative analysis of the genetic algorithm and
simulated annealing. Eur. J. Oper. Res. 1999, 114, 589–601. [CrossRef]
12. Sablani, S.S.; Ramaswamy, H.S.; Sreekanth, S.; Prasher, S.O. Neural network modeling of heat transfer to liquid particle mixtures
in cans subjected to end-over-end processing. Food Res. Int. 1997, 30, 105–116. [CrossRef]
13. Kalogirou, S.A. Artificial neural networks in renewable energy systems applications: A review. Renew. Sustain. Energy Rev.
2001, 5, 373–401. [CrossRef]
14. Hassan, R.; Cohanim, B.; De Weck, O.; Venter, G. A comparison of particle swarm optimization and the genetic algo-
rithm. In Proceedings of the 46th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference,
Austin, TX, USA, 18–21 April 2005; p. 1897.
15. Wang, S.; Zhao, B.; Yi, S.; Zhou, Z.; Zhao, X. GAPSO-Optimized fuzzy PID controller for electric-driven seeding. Sensors
2022, 22, 6678. [CrossRef]
16. Upadhyay, G.; Kumar, N.; Raheman, H.; Dubey, R. Predicting the power requirement of agricultural machinery using ANN
and regression models and the optimization of parameters using an ANN–PSO technique. AgriEngineering 2024, 6, 185–204.
[CrossRef]
17. Kumar, M.; Yadav, N. Multilayer perceptrons and radial basis function neural network methods for the solution of differential
equations: A survey. Comput. Math. Appl. 2011, 62, 3796–3811. [CrossRef]
18. Zeng, Z.; Chen, Y.; Zhang, X. Modelling the interaction of a deep tillage tool with heterogeneous soil. Comput. Electron. Agric.
2017, 143, 130–138. [CrossRef]
19. Alzoubi, I.; Delavar, M.R.; Mirzaei, F.; Nadjar Arrabi, B. Comparing ANFIS and integrating algorithm models (ICA-ANN,
PSO-ANN, and GA-ANN) for prediction of energy consumption for irrigation land leveling. Geosyst. Eng. 2018, 21, 81–94.
[CrossRef]
20. Rabbani, A.; Samui, P.; Kumari, S. A novel hybrid model of augmented grey wolf optimizer and artificial neural network for
predicting shear strength of soil. Model. Earth Syst. Environ. 2023, 9, 2327–2347. [CrossRef]
21. Vanderplaats, G.N. Numerical Optimization Techniques for Engineering Design with Applications; McGraw-Hill: New York, NY, USA, 1984.
22. Mastinu, G.; Gobbi, M.; Miano, C. Optimal Design of Complex Mechanical Systems: With Applications to Vehicle Engineering; Springer
Science, Business Media: Berlin/Heidelberg, Germany, 2007.
23. Goh, C.K.; Tan, K.C. Evolutionary Multi-Objective Optimization in Uncertain Environment; Springer: Berlin/Heidelberg, Germany, 2009.
24. Manian, R.; Kathirvel, K. Development and evaluation of an active-passive tillage machine. Agric. Mech. Asia Afr. Lat. Am.
2001, 32, 9–18.
25. Hoki, M.; Burkhardt, T.H.; Wilkinson, R.H.; Tanoue, T. Study of PTO driven powered disk tiller. Trans. ASAE 1988, 31, 1355–1360.
[CrossRef]
26. Hann, M.J.; Giessibl, J. Force measurements on driven discs. J. Agric. Eng. Res. 1998, 69, 149–157. [CrossRef]
AgriEngineering 2025, 7, 121 17 of 17

27. Nalavade, P.P.; Salokhe, V.M.; Niyamapa, T.; Soni, P. Performance of free rolling and powered tillage discs. Soil Tillage Res.
2010, 109, 87–93. [CrossRef]
28. ASAE S313.3; Soil Cone Penetrometer. ASABE Standards: St. Joseph, MI, USA, 2001.
29. Kumar, S.; Singh, T.P. Assessment of power, energy and torque of powered disc through soil bin study. J. Agric. Eng. 2016, 53, 1–9.
30. ASABE D497.5; Agricultural Machinery Management Data. ASABE Standards: St. Joseph, MI, USA, 2006.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.