基于EMD-VMD-LSTM的地震信号分类研究

Research on seismic signal classification based on EMD-VMD-LSTM

针对地震信号分类问题,提出了一种基于经验模态分解—变分模态分解—长短期记忆(EMD-VMD-LSTM)的地震信号分类研究的模型。首先利用EMD和VMD分别提取地震信号的前5个本征模态分量;然后对提取出来的每个本征模态分量求出其熵值,作为分类特征;最后把分类特征输入到LSTM网络中,构成EMD-VMD-LSTM分类模型,对地震信号进行分类实验。实验结果表明:该分类模型对比单一分解方法模型,对地震信号进行分类研究更为有效。

Aiming at the problem of seismic signal classification, a model based on empirical mode decomposition variational mode decomposition long short-term memory(EMD-VMD-LSTM)is proposed.Firstly, the first five intrinsic mode components of seismic signal are extracted by using EMD and VMD.Secondly, the entropy value of each of the extracted intrinsic mode components is derived as a classification feature.Finally, the classification features are input into the LSTM network to form EMD-VMD-LSTM classification model, and take classification experiments on seismic signal.The experimental results show that the classification model, compared with the single decomposition model, is more effective in classifying study on seismic signal.