Page 166 - IJOCTA-15-4

Oleiwi et al. / IJOCTA, Vol.15, No.4, pp.706-727 (2025)
In their study, Cheyun et al. 26 proposed a proposed two improved PID controller structures
robust adaptive controller that integrates SMC designed to enhance the con-PID’s performance
with an RBF NN. In this approach, the unknown in guiding a human leg robot along a predefined
dynamics are approximated using the RBF net- path. The grasshopper optimization algorithm
work, while the adaptive update law for the net- was utilized to optimize the improved PID pa-
work’s parameters is derived based on Lyapunov rameters. In the study by Yadav et al., 35 var-
stability theory. Mohamed et al. 27 introduced ious soft computing techniques were applied to
six control structures utilizing FOPID and con- fine-tune and implement a hybrid controller for
ventional PID (con-PID) controllers, all of which robotic manipulator motion control. This hy-
were enhanced through NN integration. These brid controller combined traditional PID control
structures included recurrent NN-like FOPID, re- with an “overwhelming” controller, and the PID
current NN-like PID, set point-weighted PID, set parameters were optimized using methods such
point-weighted FOPID, hybrid NN–FOPID, and as Ziegler–Nichols tuning, PSO, ant colony opti-
NN–PID controllers. The zebra optimization al- mization, and genetic algorithms.
gorithm was employed to fine-tune the param-
Over recent decades, numerous control strate-
eters of these controllers with the aim of min-
gies have been explored in the literature, includ-
imizing the integral time square error (ITSE).
Similarly, Chotikunnan and Chotikunnan 28 pre- ing classical, adaptive, and SMC theories. How-
ever, these approaches often face limitations in
sented a dual-design PID controller, which uti-
adaptability, as they typically account for only
lized two PID controllers—one operating in real-
a narrow range of disturbances. Moreover, due
time and the other with a one-time-step delay.
to the inherent mechanical complexity of robotic
The performance of this dual design was eval- systems, they exhibit highly nonlinear dynamics,
uated against a con-PID controller, using the making the modeling and control processes partic-
Chien–Hrones–Reswick tuning method to achieve ularly challenging. This has motivated the devel-
minimal overshoot. Mary et al. 29 proposed a
opment of more resilient and effective controllers
fuzzy-like PD controller combined with SMC.
capable of handling such complexities.
This controller was developed based on fuzzy
rules, membership functions, and the nominal The primary objective of this work is to de-
model of the robotic manipulator. sign hybrid controllers that integrate NNs with
PID operations to leverage the advantages of
Coutinho and Hess-Coelho 30 proposed two
both techniques in control system design. The
hybrid control techniques: one combining pure
proposed controllers require parameter tuning to
computed torque control with a modified SMC,
achieve specific performance objectives. To ac-
and the other integrating pure PD control with complish this, the African vultures optimization
SMC. In the study by Yahya and Abbas, 31 an en-
algorithm (AVOA) is employed to identify the op-
hanced version of the parallel Harris hawks opti-
timal set of parameters for the hybrid controllers.
mization algorithm was introduced along with in-
This study focuses on designing a controller for
tegral SMC to identify optimal controller param-
a 3-LRRM to enable precise position control by
eters. The proposed method was compared with tracking a desired trajectory, while also demon-
other algorithms, including the standard Harris
hawks optimization, demonstrating improved per- strating robustness against external disturbances
formance. Meanwhile, C¸etinkaya et al. 32 con- and uncertainties within the system model. The
main contributions of this work are as follows:
ducted positioning analysis using NNs for robot
manipulator systems in the textile industry and (i) This work designs two hybrid control struc-
refined the most accurate NN model for high- tures that combine NNs with PID opera-
precision positioning tasks. In the study by Saeed tions. The first structure, STNN–PID, is
Al-Khayyt, 33 a self-tuning NN–PID (STNN–PID) a self-tuning PID controller based on NNs.
controller was developed, combining the adaptive The second structure, NN–PID, is a newly
learning capabilities of NNs with a con-PID con- proposed hybrid controller that incorpo-
troller. The NN output, derived from the system’s rates PID operations as activation func-
control error, dynamically adjusts the propor- tions within the NN architecture.
tional, integral, and derivative gains in real time. (ii) A new technique is proposed to eliminate
To enhance dynamic performance—particularly controllers that exhibit complex control ac-
to minimize overshoot and achieve accurate tra- tions with excessive chattering, which are
jectory tracking with minimal steady-state er- impractical for real-world implementation.
ror—the con-PID controller was replaced with This technique is applied and shown to be
this STNN–PID controller. Adnan and Karam 34 highly effective.
708