Active noise control integrating the least mean squares algorithm and support vector machine algorithm

Abstract During vehicle operation, noise signals present challenges such as a broad frequency range, strong non-stationarity, and the limited effectiveness of traditional passive noise reduction methods in the low-frequency band. To address these issues, this study proposes an active noise control model that integrates the least mean square algorithm with support vector machines. The noise control architecture incorporates a least mean squares branch for real-time suppression of linear noise components, while a support vector machine branch models and compensates for nonlinear residuals and frequency drift. Outputs from both branches are coupled via a hybrid coefficient, forming a closed-loop control mechanism integrated with secondary channel modeling. Experimental results demonstrated that the proposed model achieved significantly faster convergence speed and reduced steady-state error compared to the standard FxLMS system across both datasets, with in-band noise reduction increasing by approximately 2.0–2.5 dB. Furthermore, it achieved faster stabilization control while maintaining an average computational delay of around 3.3 ms. Under frequency drift conditions, the method exhibited markedly shorter error recovery times during step and ramp transitions than the benchmark algorithm, indicating superior rapid tracking capability and control stability in complex non-stationary environments. In summary, this research not only effectively enhances noise reduction performance and nonlinear adaptability under complex automotive conditions but also provides an efficient and feasible technical approach for low-frequency noise mitigation in intelligent driving environments.