Fixed Versus Fuzzy Self-Tuning PID Speed Control of a Geared DC Motor: Experimental Identification and Robustness Evaluation

Abstract

Fixed-gain proportional-integral-derivative (PID) control is attractive for low-cost direct-current motor drives, but performance can degrade when load and plant parameters depart from the tuning condition. This paper compares a fixed PID controller with a zero-order Sugeno fuzzy self-tuning PID controller for speed regulation of a 12 V JGA25-370 geared motor. An Arduino Uno, L298N driver, and Hall-effect encoder were used to acquire a 100 ms open-loop step-response record. A filter-aware output-error identification procedure yielded the first-order model G(s)=10.8826/(0.024254s+1) rpm/V, with a root-mean-square fit error of 0.1553 rpm and R²=0.999938. Both controllers were evaluated in MATLAB/Simulink with identical sampling, encoder quantisation, measurement filtering, command saturation, and anti-windup assumptions. At the nominal 100 rpm operating point, both controllers achieved a 0.4 s rise time, while the fixed PID settled in 0.6 s rather than 0.8 s and produced lower integral error indices. Under an equivalent 2 V opposing disturbance, fuzzy scheduling reduced maximum speed deviation by 26.88%, recovery time by 40%, and post-event integral absolute error by 51.57%. Under parameter variation, the corresponding reductions were 28.32%, 40%, and 49.63%. Across 500 paired Monte Carlo trials, mean ITAE and settling time decreased by 17.94% and 13.54%, respectively. The fuzzy controller nevertheless increased control-signal total variation by 6.68 times. The results support fuzzy self-tuning as a robustness enhancement, not a universal replacement for a well-tuned fixed PID controller.