This research paper present an efficient and robust way of tuning PID controller using five variants of compact Differential Evolution (cDE) algorithms for speed control of a dynamic system, with major focus on external disturbance mitigation. Control systems should not only be designed to track the command input (desired speed), but they should also be evaluated on the basis of their ability to mitigate the effects of external disturbance resulting from both natural sources (winds, thunder strike, etc), and man oriented events (e.g. explosive, etc), and also the effects of internal disturbance due to system parameters deviation from the optimal. Many catastrophic effects resulting from failure of control systems is often traced to inability of the system to maintain stability due to the effects of external disturbance or system parameter deviation. Out of the five variants implemented, cDE/rand/2/exp using exponential crossover appear to be more promising in addressing this problem with root mean square error (RMSE) of 16.45 RPM, maximum disturbance unit step amplitude of 0.00133 that rapidly decayed to zero within 1 second. The second best algorithm is cDE/rand/2/bin using binomial crossover with RMSE of 16.45 RPM, and maximum disturbance unit step amplitude of 0.00136. One of the major advantage of using compact differential evolution in this research as compared with the population based variants is that cDE can be implemented using single chip microcontrollers with limited memory space to effectively fine-tuned PID gains for real time control of complex systems. The controller implemented in this research depicted a robust performance not only in tracking the command input (desired speed), but also in mitigating the effect of external disturbance.
Iliya et al. (Mon,) studied this question.