Noise‐Adaptive SOGI–HOSM Observer for Sensorless Speed Control of Induction Machines

This study addresses the limitations of traditional sliding‐mode observers in sensorless induction motor control under complex, non‐Gaussian noise profiles. Existing methods often assume white Gaussian noise, which fails to represent real‐world disturbances such as colored noise (e.g., pink or brown), heavy‐tailed distributions (e.g., Laplacian), and band‐limited noise due to sensor constraints. These challenges lead to instability and degraded estimation accuracy. A noise‐adaptive second‐order generalized integrator–based higher‐order sliding‐mode observer (NA SOGI–HOSM) framework is proposed to overcome these limitations. The approach dynamically adjusts observer parameters based on real‐time noise classification and magnitude estimation. Key innovations include an adaptive hyperbolic tangent nonlinearity to suppress chattering while ensuring robust convergence, noise‐dependent gain exponent adjustments, and enhanced SOGI filtering with real‐time resonance frequency tuning and damping coefficients for precise noise attenuation. A Lyapunov‐based stability analysis establishes the robust convergence of the observer under various noise conditions. Simulation results demonstrate that the proposed method achieves superior state estimation accuracy and stability compared to conventional super‐twisting and baseline SOGI–HOSM observers, even under significant noise and parameter variations. The NA SOGI–HOSM is a significant advancement in sliding‐mode observer design, enabling accurate, sensorless speed estimation and enhanced performance in challenging real‐world scenarios.