Abstract: The traditional filtered-x least mean square (FxLMS) algorithm commonly used in active noise control in vehicles often results in insufficient hardware computing power and unsatisfactory noise reduction effects due to its high computational complexity. An adaptive notch local-secondary-path filtered-x least mean square (LFxLMS) algorithm based on an improved local secondary path modeling method and the corresponding narrowband active noise control (LFx-NANC) system are proposed in this paper. The improved local secondary path modeling method has a higher modeling accuracy, and the LFx-NANC system can greatly reduce the computational complexity compared to traditional systems. The noise reduction effects of the system on the steady-state and unsteady-state multi-harmonic noises are verified through the simulation analysis of Matlab. Based on the ADSP-21489 controller, a dual channel LFx-NANC system is built in the vehicle, and the second, fourth, and sixth order noise reductions of 34.67 dB(A), 21.41 dB(A), and 10.29 dB(A) at the driver's position are achieved respectively under the steady-state condition. Under the acceleration condition, the total sound pressure level and second-order noise reductions at the driver's position reach 6.01 dB(A) and 20.40 dB(A) respectively, while having good noise reduction effects at other positions. The proposed method provides a reference for the application of active noise control engineering.