Abstract: As a core component of high-end equipment, the control precision of motorized spindles is directly affected by the time-varying characteristics of stator resistance. During actual operation, stator resistance drift occurs due to factors such as temperature rise, leading to degraded control performance. Furthermore, traditional identification methods suffer from sensitivity to initial values and a tendency to converge to local optima. To address these issues, a stator resistance identification method (TFOA-BP) that integrates an improved fruit fly optimization algorithm (TFOA) with a BP neural network is proposed, aiming to enhance identification accuracy and robustness. Simulation results demonstrate that the identification error of the proposed TFOA-BP method remains stable within ±0.0046 Ω, which is 68.2% lower than that of the traditional BP neural network. In comparison with various mainstream methods, the mean squared error (MSE) shows an average reduction by 42.7%. The proposed method exhibits significant advantages in identification accuracy, convergence speed, and stability, offering both theoretical reference value and practical engineering applicability for intelligent motor parameter identification.