一种基于多层迭代奇异谱分析的信号分解方法

A Signal Decomposition Method Based on Multi-Layer Iteration Structured Singular Spectrum Analysis

针对复杂信号中相邻频率成分的分解问题,研究了一种基于多层迭代结构的奇异谱分析方法。通过频带细分和迭代筛选,对传统的奇异谱分析进行改进,设计出一种具有多层迭代结构的高精度分解算法,解决了原方法自适应分解能力不足、单次分解效果不佳的问题。仿真结果表明,改进方法的信号自适应分解能力得到了有效增强,能够准确地从频差/中心频率为0.2%的多模态振动信号中提取出各分量,且与理论值一致性好,波形失真度小,证明了其有效性。

To process the complex signals with concentrated frequency distribution,an adaptive decomposition method based on singular spectrum analysis with multi-layer iteration structure is researched.The traditional singular spectrum analysis is improved by frequency band subdivision and iterative filtering approach.A high-precision decomposition algorithm base on recursive structure is therefore designed,solving the problems such as insufficient adaptive capability and unsatisfied decomposition.Simulation results show that adaptive decomposition capability of the proposed method is effectively enhanced.For the multi-mode vibration signal with 0.2%ratio of spitting frequency to center frequency,the components are all accurately extracted,and consistent well with the theoretical value.