中文摘要：

在分析基于SVD的平稳信号降噪原理的基础上，提出了一种基于SVD的非平稳信号重叠分段降噪算法。非平稳信号通过数据分段可近似处理为多段平稳信号。为降低连续分段降噪误差，进行一定比例的重叠分段，应用基于SVD的平稳信号降噪方法对各分段信号进行降噪，相邻段之间的重叠信号经过多次降噪后求平均得到最终降噪信号。仿真结果表明，该算法具有良好的降噪能力且稳定性较好。

英文摘要：

Based on the analysis of SVD-based noise reduction method for stationary signal, an SVD-based overlapping fragment noise reduction algorithm for nonstationary signal is proposed. Nonstationary signal can be divided into many sections of stationary signal approximately by intercepting. In order to reduce the error caused by continuous fragment noise reduction, nonstationary signal is intercepted with some proportional superposition. The noise in each fragment signal will be reduced by SVD-based noise reduction method for stationary signal. The final noise-reduced nonstationary signal is attained by computing the mean of the adjacent overlapping signals which have been de-noised in many times. Simulation results show that the algorithm is of better performance and stronger stability in noise reduction for nonstationary signal.