An Approach Based on the Bacterial Foraging Algorithm for Induction Motor Parameters Identification

The accurate parameter identification of three-phase induction motors (IM) certainly becomes fundamental to the design of high-performance motor drive systems, the real-time digital simulation models and predictive maintenance platforms. This paper presents a systematic comparative study between two methodologically distinct approaches in other to identify the per-phase equivalent circuit parameters of a 5 HP (3.73 kW), 415 V, 50 Hz, four-pole, three-phase squirrel-cage IM: the classical experimental test method (covering: DC resistance test, no-load test and blocked-rotor test in accordance with the IEEE 112) and then a bio-inspired Bacterial Foraging Optimization Algorithm (BFOA). The experimental method yields five key parameters — stator resistance (R₁ = 1.143 Ω), rotor resistance referred to stator (Rr = 0.729 Ω), stator and rotor leakage reactances (X₁ = X₂ = 1.534 Ω), magnetizing reactance (X_m = 11.81 Ω) and the core-loss resistance (R_c = 107.8 Ω) — under idealized test conditions that necessitate several simplifying assumptions. Clearly, the BFOA treats all the five parameters simultaneously as decision variables within a bounded search space. This is seeded by experimental values, minimizing a multi-attributes objective function which penalizes prediction errors in the stator input current, electromagnetic torque and input power across multiple load points. The results obtained demonstrate that the BFOA-refined parameter set reduces the overall model prediction error by almost 96.3%, from J(φ) = 0.0842 based on the experimental parameters to J(φ*) = 0.0031 after the optimization. Also, there is a huge improvement in the accuracy of the torque-speed characteristic, power factor prediction and efficiency characteristics. Ten characteristic plots are presented and analyzed in detail. The plots illustrate both the motor’s fundamental performance behavior and the superiority of the BFOA-identified parameters over the experimentally derived baseline.