PSO-Tuned Sugeno Fuzzy Control of Partial Rotor-Reactance Compensation During Starting of a 5 MW Slip-Ring Induction Motor

Abstract

High-power wound-rotor induction motors remain useful in oil-and-gas installations because an external liquid resistance starter (LRS) provides controllable starting torque. However, rotor leakage reactance contributes to lagging displacement power factor during acceleration. This paper proposes a particle-swarm-optimised zero-order Sugeno fuzzy supervisor that coordinates LRS electrode travel with bounded rotor-side capacitive reactance cancellation. The physically consistent partial-compensation relation is Ceff = Cres/α, where α = |XC|/XL and 0 < α ≤ 0.72; the capacitor branch is disconnected when α is below the enable threshold, hardware-equivalent capacitance exceeds its ceiling, or independent protection permissives fail. A reproducible fundamental-frequency reduced-order model compares conventional LRS, an independently optimised fixed capacitor, an untuned fuzzy supervisor and the PSO-tuned fuzzy design for a 5 MW, 6.6 kV, four-pole motor. Twelve independently seeded PSO runs provide repeatability evidence. Under combined conductivity, temperature and load stress, mean starting displacement power factor increases from 0.8506 to 0.8662 (+1.84% relative), while mean reactive power decreases by 7.51% relative to conventional LRS and by 4.19% relative to the optimised fixed-capacitor comparator. Peak-current change is +0.28% and LRS dissipated-energy change is +0.36%. A five-level 150-250 mF effective-capacitance screening study and a three-profile re-ranking of the twelve realised PSO candidates quantify hardware-limit and objective-preference sensitivity. The findings are reproducible numerical design-study results; they do not represent measured plant performance, a trained ANFIS, a certified converter design or experimental validation.