The Hall sensor signals of brushless DC motors (BLDCM) are susceptible to PWM switching noise, electromagnetic interference, and other factors, leading to commutation errors and torque ripple, which seriously affect control performance. This paper proposes a signal correction method based on adaptive noise cancellation (ANC) technology. Using the PWM synchronization signal as the reference input, the least mean square (LMS) adaptive filtering algorithm is employed to estimate and cancel the noise components in the Hall signal in real time. The physical mechanism of Hall signal distortion is revealed through theoretical analysis, and a variable-step LMS algorithm and system architecture suitable for motor control systems are designed. Simulation and experimental results show that the proposed method reduces the commutation timing deviation by more than 69%, phase current ripple by 56%, and improves system efficiency by 2.6 percentage points. The algorithm is simple, has good real-time performance, requires no additional hardware cost, and possesses high engineering application value.