目前,在基于医疗大数据与机器学习的心音识别系统研究中,对于单心动周期的提取大多依赖人工截取或基于同步心电信号进行分割,大大降低了整个系统的实用性和易用性。针对以上问题提出了一种基于低频提取的单心动周期分割及MFCC(Mel Frequ...目前,在基于医疗大数据与机器学习的心音识别系统研究中,对于单心动周期的提取大多依赖人工截取或基于同步心电信号进行分割,大大降低了整个系统的实用性和易用性。针对以上问题提出了一种基于低频提取的单心动周期分割及MFCC(Mel Frequency Cepstral Coefficients)特征提取的嵌入式硬件系统,能够更高效地实现单心动周期分割并计算其MFCC特征参数,综合分割准确率达98.3%,解决了单心音周期分割中对心音信号纯净度要求较高和没有成熟系统的问题,并且降低了数据存储成本,具有较好的实用性和潜在的应用前景。展开更多
An algorithm for detecting low-frequency seismic events is developed and applied to the detection of low-frequency events before the 2008 Wenchuan and the 2013 Lushan earthquakes. Continuous vertical-component wavefor...An algorithm for detecting low-frequency seismic events is developed and applied to the detection of low-frequency events before the 2008 Wenchuan and the 2013 Lushan earthquakes. Continuous vertical-component waveforms of some broadband stations in the few months before the Wenchuan and Lushan earthquakes are processed by applying a bandpass filter in 2- 8Hz,and then converted to envelopes with a smoothing time of 10 s window and a median filter with a 20 min window. As a result,teleseismic,long-period noise and local small earthquakes are removed,the filtered amplitude is obviously larger than that of the noise and lasts for a dozen minutes to several hours during a few days in a few stations before the Wenchuan and Lushan earthquakes,respectively. The waveform and envelope are similar to that of a non-volcanic tremor( NVT). There are suspected NVT before the two earthquakes. Preliminary application demonstrates that this algorithm is potentially useful for extracting NVT signals from continuous seismic waveforms.展开更多
3 Summary and discussion In the generalized drift resonance theory[17],a characteristic signature of the ULF wave-particle interactions is the increasingly-tilted stripes in the particle energy spectrum.The phase diff...3 Summary and discussion In the generalized drift resonance theory[17],a characteristic signature of the ULF wave-particle interactions is the increasingly-tilted stripes in the particle energy spectrum.The phase difference across different energy channels is relatively small展开更多
文摘目前,在基于医疗大数据与机器学习的心音识别系统研究中,对于单心动周期的提取大多依赖人工截取或基于同步心电信号进行分割,大大降低了整个系统的实用性和易用性。针对以上问题提出了一种基于低频提取的单心动周期分割及MFCC(Mel Frequency Cepstral Coefficients)特征提取的嵌入式硬件系统,能够更高效地实现单心动周期分割并计算其MFCC特征参数,综合分割准确率达98.3%,解决了单心音周期分割中对心音信号纯净度要求较高和没有成熟系统的问题,并且降低了数据存储成本,具有较好的实用性和潜在的应用前景。
基金funded by the National Science & Technology Pillar Program in the 12th "Five-year Plan" Period,China(2012BAKI9B02)
文摘An algorithm for detecting low-frequency seismic events is developed and applied to the detection of low-frequency events before the 2008 Wenchuan and the 2013 Lushan earthquakes. Continuous vertical-component waveforms of some broadband stations in the few months before the Wenchuan and Lushan earthquakes are processed by applying a bandpass filter in 2- 8Hz,and then converted to envelopes with a smoothing time of 10 s window and a median filter with a 20 min window. As a result,teleseismic,long-period noise and local small earthquakes are removed,the filtered amplitude is obviously larger than that of the noise and lasts for a dozen minutes to several hours during a few days in a few stations before the Wenchuan and Lushan earthquakes,respectively. The waveform and envelope are similar to that of a non-volcanic tremor( NVT). There are suspected NVT before the two earthquakes. Preliminary application demonstrates that this algorithm is potentially useful for extracting NVT signals from continuous seismic waveforms.
基金supported by the National Natural Science Foundation of China(Grant Nos.41421003&41474140)
文摘3 Summary and discussion In the generalized drift resonance theory[17],a characteristic signature of the ULF wave-particle interactions is the increasingly-tilted stripes in the particle energy spectrum.The phase difference across different energy channels is relatively small
