time-freq feature from signal for phm purpose
Project description
phm-feature
介绍
- phm中的特征抽取任务
- 抽取振动信号中的各类时、频域业务特征值
软件架构
软件架构说明
.. code-block:: shell
.
├── build
│ ├── bdist.linux-x86_64
│ └── lib
│ └── phm_feature
│ └── __init__.py
├── dist
│ ├── phm_feature-0.0.2-py3-none-any.whl
│ └── phm_feature-0.0.2.tar.gz
├── LICENSE
├── phm_feature
│ ├── __init__.py
├── phm_feature.egg-info
│ ├── dependency_links.txt
│ ├── PKG-INFO
│ ├── SOURCES.txt
│ └── top_level.txt
├── README.md
├── setup.py
└── test.py
安装教程
- 使用pip进行安装
.. code-block:: python
pip install phm-feature
使用说明
获得振动信号的特征值
.. code-block:: python
import phm_feature
from phm_feature import *
enable_parallel(processnum=None) 开启多线程模式
disable_parallel() 开启单线程模式
feature_t(data) 获取时间域特征
feature_f 获取频率域特征
fft(data, 50) 快速离散傅里叶变换
power(data, 50) 功率谱
ifft(data, 50) 快速离散逆傅里叶变换
cepstrum(data, 50) 倒谱
envelope(data) 包络谱
window(data, 'hamming') 加窗-汉明窗
divide(data, 50, 25) 分帧
参与贡献
--------------------------------------------------------------------------------------------
2022-07-04 v0.0.1
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
pypi上传初版本
pypi上传phm-feature初版本
2022-07-04 v0.0.2
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
新增多线程模式
phm-feature
=========================================================================
介绍
-------------------------------------------------------------------------
使用torch torchaudio 构建PHM特征抽取功能层
功能层介绍
-----------------------------------------------------------------------------
torchphm.layers 将phm固定使用的数据操作,固化为如下层:
.. code-block:: python
1. STFT 离散傅立叶变换
2. Spectrogram 谱图
3. MelFilterbank mel譜过滤
4. AmplitudeToDb 幅度取分贝
5. TimeStretch 变速不变调
6. ComplexNorm 复数输出取模'范数'
7. ApplyFilterbank 过滤器应用
应用场景
------------------------------------------------------------------------------------
.. note::
组合上述功能层,实现不同场景
.. code-block:: shell
1. STFT
分帧==>加窗==>短时离散傅立叶变换
2. "变速不变调" Time scale modification
详细见 https://zhuanlan.zhihu.com/p/337193578
用于声谱图压缩、扩张处理,供后续分析
3. 梅尔mel谱转换
对于语谱图进行mel谱转换
4. HPSS
中值滤波,过滤出频率的谐波分量与冲击分量
为什么中值滤波,可以过滤出"数据轮廓"及发现"谐波分量和冲击分量",参见
https://blog.csdn.net/qq_38131594/article/details/80758567
软件结构说明
-----------------------------------------------------------------------------------
.. code-block:: shell
.
├── dist
│ ├── torchphm-1.1.7.tar.gz
│ └── torchphm-1.1.8.tar.gz
├── draw
│ ├── draw_functional.py
│ ├── draw_layers.py
│ ├── draw_torchphm_layers.ipynb
│ ├── torchphm -> ../torchphm
│ └── Untitled.ipynb
├── examples_torchphm.ipynb
├── README.md
├── setup.cfg
├── setup.py
├── tests
│ ├── test_functional.py
│ └── test_layers.py
├── torchphm
│ ├── beta_hpss.py
│ ├── functional.py
│ ├── __init__.py
│ ├── layers.py
└── torchphm.egg-info
======================== ===========
dist pypi上传包
======================== ===========
draw 画图
examples_torchphm.ipynb 应用例程
tests 测试文件
torchphm 实现源码
======================== ===========
安装教程
--------------------------------------------------------------------------------------
.. code-block:: python
pip install phm-feature
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
phm_feature-0.1.1.tar.gz
(17.2 kB
view hashes)
Built Distribution
Close
Hashes for phm_feature-0.1.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 84fff4808c59a2bc6dff3b4cefc5b899b04cdd28aab8c1ca57e36ffa06218eee |
|
MD5 | b2ea0cb14d081661897472f854884dfa |
|
BLAKE2b-256 | e7a04b5dceefa55b6666034a1802d7d6f8e230c5b1f3ce6a13dd5532643fcd0f |