Torque Ripple Minimization in a Switched Reluctance Drive by Neuro-fuzzy Compensation

1
Torque Ripple Minimization in a Switched
Reluctance Drive by Neuro-fuzzy Compensation
Paulo J. C. Branco Joaquim A. Dente Mechatronics
Laboratory Instituto Superior Técnico
(IST) Portugal
Luís O. A. P. Henriques Luís G. B. Rolim Walter
I. Suemitsu Federal University of Rio de
Janeiro COPPE Brasil

• Advantages
• High efficiency
• Low manufacturing cost
• Fault tolerant
• Reliable
• Easy to repair
• Disadvantages
• Torque ripple
• Nonlinear model
• Applications
• Traction
• Heavy-duty applications
• Home appliances

Switched Reluctance Motor - Complete Simulated
System
Diagram of proposed SR torque ripple
compensation scheme
Torque signal without compensation (constant
current and velocity)
Torque signal with compensation (constant current
and velocity) after 10 learning iterations
Compensated Torque Harmonics Triangular fuzzy sets
Compensated Torque Harmonics Gaussian fuzzy sets
Compensated Current pulses after 10 learning
iterations
Conclusions
Support

• Neuro-fuzzy compensating mechanism to ripple
  reduction was investigated
• Compensating signal added in current waveform was
  used to minimize the torque ripple
• Bell shape function produces better ripple
  reduction in all harmonic content
• Future investigation
• Application of this concept in an experimental
  drive
• Incorporate another signal to be trained

Compensated Torque Harmonics Bell fuzzy sets