Page 18 - IJOCTA-15-4

M. A. Aman et al. / IJOCTA, Vol.15, No.4, pp.549-577 (2025)

Figure 18. Adaptive neuro-fuzzy inference system (ANFIS) strategy for switched reluctance motors (SRMs).
Adapted from Pushparajesh et al. 58
Abbreviation: PI: Proportional integral.

of SRMs. 64 The proposed pre-trained ANN mod- have been applied to torque control, utilizing di-
els were employed to estimate the actual torque rect flux control rather than relying on phase cur-
using motor current and position data, in addi- rent or torque control. 69 While this method is
tion to calculating the suitable reference currents well-suited for digital control implementations, it
for each phase to mitigate torque ripple. Mean- requires precise information regarding the motor’s
while, a novel intelligent technique was proposed features and rotor position.
to manage SRM speed, with a focus on minimiz- The feedback linearization approach employs
ing torque ripple. 63 This technique was based on state feedback to transform the nonlinear system
a computational model of the mammalian limbic into a linearized closed-loop system, 70 efficiently
system and emotional processes—the brain’s emo- addressing the nonlinear features of the motor.
tional learning based intelligent controller. This However, the significant drawbacks of this method
method employs machine learning to achieve sim- include the requirement for an accurate motor
ple and effective controls that are fully indepen- model that necessitates large currents while op-
dent of motor characteristics and eliminate the erating at low speeds and the monitoring of state
need for conventional controllers. The proposed variables (stator currents, position, and velocity).
technique demonstrated high tracking capabil- To address these limitations, an adaptive feedback
ity, rapid auto-learning, enhanced speed response, linearization method has been employed, lever-
and significant reduction of torque ripple. aging multi-objective optimization through a ge-
netic algorithm. 71
3.1.4. Other torque controlling strategies
This section presents a comprehensive overview of
various other techniques aimed at reducing torque 3.2. Comparison of the torque control
ripple. For example, a variable structure control strategies
technique is employed for SRM to enhance its per-
formance and reduce torque ripple compared with All the mentioned torque control strategies have
conventional control techniques. 64 However, the the potential to mitigate torque ripple, although
effects of phase coupling and magnetic saturation their implementation and computing complexity
were ignored in this approach. may differ. For example, the average torque con-
For SRM current controllers, studies have em- trol strategy maintains a constant reference phase
ployed the corresponding sliding mode variable current during excitation. This approach facil-
structure control theory. 65 Furthermore, the it- itates ease of implementation and reduces costs,
erative learning control approach was applied to thereby achieving high precision in torque estima-
the SRMs, yielding favorable control effects with- tion and enabling precise torque control. How-
out the need for measuring the motor’s magnetic ever, it is associated with notable drawbacks, in-
features or high model precision. 66 Another study cluding the generation of significant speed oscil-
proposed a self-learning method that enables on- lations and fluctuations at low speeds that arise
line optimal current determination for each phase from torque ripples during the phase commuta-
to fulfill the overall torque command. 67 Mean- tion process.
while, voltage feedback was incorporated to en- The average torque of SRMs is not directly
hance bandwidth control. 68 Genetic algorithms obtained from phase currents. Consequently,
560