Filtered‐x least mean square (FxLMS) algorithm and its variants have been prevalently applied in the field of active noise control. However, these algorithms still suffer from relatively low convergence rates and steady‐state noise reduction. In this article, we propose a new idea to improve the performance of the conventional filtered‐x least mean square/fourth (FxLMS/F) algorithm by weighting reference signal samples, thus a weighted reference signal FxLMS/F (WRS‐FxLMS/F) algorithm is presented. Besides, to solve the contradiction between convergence speed and steady‐state noise reduction, the convex combination method is combined with the WRS‐FxLMS/F algorithm, and a convex combination WRS‐FxLMS/F (CWRS‐FxLMS/F) algorithm is developed. The computational complexity of the proposed algorithms is analyzed, and the simulation results demonstrate that the proposed algorithms have better performance both in the steady‐state noise reduction and convergence rate than the traditional FxLMS, NFxLMS, and FxLMS/F algorithm.