Abstract: To optimize the ship cabin environment and reduce the noise level in the ship cabin, the existing active noise reduction algorithm filtered-X least mean square（FXLMS） is studied. Through research, it is found that the selection of variable step-length control parameters in the existing FxLMS method is mainly based on trial and error method or empirical method, which requires a lot of time, manpower and material resources. In this paper, Taguchi’s method is used to improve the variable step length control function, which is used to optimize the normalized arctangent variable step length control function in active noise reduction, so as to find the optimal parameter combination. Finally, through orthogonal experiments, the optimal parameter combination of the step-length control function in the active noise reduction of ship cabin noise is obtained. The noise of the steering engine cabin is reduced, and the sound pressure level is reduced by 5 dB（A）.