pywayne-dsp
/pywayne-dsp
Digital signal processing toolkit for filtering, peak detection, detrending, and curve similarity. Use when working with sensor data, signal preprocessing, feature extraction, noise suppression, or time-series analysis. Includes Butterworth filter, One-Euro filter, signal detrending, DTW curve similarity, Welford standard deviation, and sliding window extrema detection.
- Author
- wangyendt ↗
- Repository
- wayne-skills ↗
- Licence
- MIT
- Last repository change
- 23.10.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 23.10.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Digital signal processing toolkit for filtering, peak detection, detrending, and curve similarity. Use when working with sensor data, signal preprocessing, feature extraction, noise suppression, or time-series analysis. Includes Butterworth filter, One-Euro filter, signal detrending, DTW curve similarity, Welford standard deviation, and sliding window extrema detection.
Original author description
Digital signal processing toolkit for filtering, peak detection, detrending, and curve similarity. Use when working with sensor data, signal preprocessing, feature extraction, noise suppression, or time-series analysis. Includes Butterworth filter, One-Euro filter, signal detrending, DTW curve similarity, Welford standard deviation, and sliding window extrema detection.
How it works
- Quick Start
- Filtering - 滤波器
- Peak Detection - 峰值检测
- Detrending - 信号去趋势
- Curve Similarity - 曲线相似度
- Other Tools - 其他工具
- 应用场景
SKILL.md preview
---
name: pywayne-dsp
description: Digital signal processing toolkit for filtering, peak detection, detrending, and curve similarity. Use when working with sensor data, signal preprocessing, feature extraction, noise suppression, or time-series analysis. Includes Butterworth filter, One-Euro filter, signal detrending, DTW curve similarity, Welford standard deviation, and sliding window extrema detection.
---
# Pywayne Dsp
数字信号处理工具集,提供滤波器、峰值检测、去趋势、曲线相似度等信号处理功能。
## Quick Start
```python
from pywayne.dsp import butter_bandpass_filter, peak_det, SignalDetrend
# Butterworth 低通滤波
filtered = butter_bandpass_filter(signal, order=3, lo=0.5, hi=40, fs=250)
# 峰值检测
peaks, valleys = peak_det(signal, delta=0.5)
# 信号去趋势
detrender = SignalDetrend(method='linear')
detrended = detrender(raw_signal)
```
## Filtering - 滤波器
### butter_bandpass_filter
巴特沃斯带通滤波器。
```python
from pywayne.dsp import butter_bandpass_filter
# 带通滤波
filtered = butter_bandpass_filter(
signal=raw_signal,
order=4,
lo=1,
hi=50,
fs=250,
btype='bandpass'
)
```
**参数说明**:
| 参数 | 类型 | 说明 |
|------|------|------|
| `signal` | array | 输入信号 |
| `order` | int | 滤波器阶数 |
| `lo` | float | 下限截止频率 (Hz) |
| `hi` | float | 上限截止频率 (Hz) |
| `fs` | float | 采样频率,默认为 0(不归一化) |
| `btype` | str | 滤波器类型:'lowpass', 'highpass', 'bandpass', 'bandstop' |
| `realtime` | bool | 是否实时处理,默认 False |
### ButterworthFilter
基于 NumPy 的巴特沃斯滤波器类,同时支持传递函数 `ba` 和数值更稳定的二阶节级联 `sos`。
BA/SOS 执行核心均为 Direct Form II Transposed,对齐 SciPy 的 `lfilter`/`sosfilt` 和
`filtfilt`/`sosfiltfilt`。
```python
from pywayne.dsp import ButterworthFilter
# 方式 1:通过参数设计 BA
bf = ButterworthFilter.from_params(order=4, fs=200, btype='bandpass', cutoff=(1, 50))
y, zf = bf.lfilter(signal)
# 方式 2:通过系数构造
bf2 = ButterworthFilter.from_ba(b, a)
y, zf = bf2.lfilter(signal)
# 方式 3:SOS(高阶、窄带或接近 0/Nyquist 时推荐)
bf_sos = ButterworthFilter.from_params(
order=8,
fs=200,
btype='bandpass',
cutoff=(0.1, 10),
output='sos',
)
# 也可直接使用 scipy.signal.butter(..., output='sos') 的系数
bf_sos2 = ButterworthFilter.from_sos(sos)
# 零相位滤波(前向-后向)
y = bf.filtfilt(signal)
# 流式处理:首段稳态初始化,后续传递 zf
state = bf_sos.zi() * chunks[0][0]
outputs = []
for chunk in chunks:
y, state = bf_sos.lfilter(chunk, zi=state)
outputs.append(y)
# 去趋势
detrended = ButterworthFilter.detrend(signal, method='linear')
```
**参数设计方法**:
```python
ButterworthFilter.from_params(order, fs, btype, cutoff, cache_zi=True, output='ba')
ButterworthFilter.from_ba(b, a, cache_zi=True)
ButterworthFilter.from_sos(sos, cache_zi=True)
ButterworthFilter.lfilter_zi(b, a)
ButterworthFilter.sosfilt_zi(sos)
```
**参数说明**:
| 参数 | 类型 | 说明 |
|------|------|------|
| `order` | int | 滤波器阶数 |
| `fs` | float | 采样频率 (Hz) |
| `btype` | str | 'lowpass', 'highpass', 'bandpass', 'bandstop' |
| `cutoff` | float/Tuple | 截止频率 (Hz),带通为 (low, high) 元组 |
| `cache_zi` | bool | 是否预计算稳态初始条件 |
| `output` | str | `'ba'` 或 `'sos'`;高阶滤波推荐 `'sos'` |
**实例方法**:
| 方法 | 说明 |
|------|------|
| `zi(self)` | 返回稳态初始条件;SOS 形状为 `(n_sections, 2)` |
| `lfilter(self, x, zi=None)` | 单向因果滤波,返回 (y, zf) |
| `filtfilt(self, x, padtype='odd')` | 零相位滤波,可指定填充方式 |
SOS 模式的 `lfilter` 接受 SciPy 形状 `(n_sections, 2)` 或等价扁平 `zi`,返回的
`zf` 始终是 `(n_sections, 2)`。在线滤波首段建议使用 `zi() * x[0]`,后续必须持续传递
前一段的 `zf`。`filtfilt` 不可用于实时流式处理,且有限长度信号仍需检查边界效应。
## Peak Detection - 峰值检测
### peak_det
峰值检测函数,基于 MATLAB peakdet 转换。
```python
from pywayne.dsp import peak_det
max_peaks, min_peaks = peak_det(signal, delta=0.5)
```
**参数说明**:
| 参数 | 类型 | 说明 |
|------|------|------|
| `v` | array | 输入信号 |
| `delta` | float | 检测阈值,控制检测灵敏度 |
| `x` | array | 可选的 x 轴数据,若未提供则使用下标 |
**返回值**:`(maxima_indices, minima_indices)` - 峰值和谷值的索引位置
### find_extremum_in_sliding_window
在滑动窗口中查找极值。
```python
from pywayne.dsp import find_extremum_in_sliding_window
extrema = find_extremum_in_sliding_window(data, k=50)
```
**参数说明**:
| 参数 | 类型 | 说明 |
|------|------|------|
| `data` | list | 输入数据列表 |
| `k` | int | 滑动窗口大小 |
**返回值**:`[minima, maxima]` - 含局部极值的列表
### FindSlidingWindowExtremum
滑动窗口极值查找器类,用于实时数据流。
```python
from pywayne.dsp import FindSlidingWindowExtremum
detector = FindSlidingWindowExtremum(win=100, find_max=True)
# 应用新值
for sample in stream:
current_peak = detector.apply(sample)
# 处理 current_peak
```
**方法**:
| 方法 | 说明 |
|------|------|
| `__init__(win, find_max)` | 初始化,指定窗口大小和查找类型(最大值或最小值) |
| `apply(val)` | 更新窗口数据,返回当前极值 |
## Detrending - 信号去趋势
### SignalDetrend
信号去趋势处理器,支持多种去趋势算法。
```python
from pywayne.dsp import SignalDetrend
# 去除线性趋势
detrender = SignalDetrend(method='linear')
detrended = detrender(signal)
# 去除均值趋势
detrender = SignalDetrend(method='mean')
detrended = detrender(signal)
# LOESS 去趋势
detrender = SignalDetrend(method='loess', span=0.3)
detrended = detrender(signal)
```
**方法**:
| 方法 | 说明 |
|------|------|
| `method` | str | 去趋势方法:'none', 'mean', 'linear', 'poly', 'loess', 'wavelet', 'emd', 'ceemdan', 'median' |
| `__call__(x)` | 应用去趋势算法处理输入信号 |
**去趋势方法**:
| 方法 | 说明 |
|------|------|
| `none` | 不处理,返回原信号 |
| `mean` | 去除均值 |
| `linear` | 去除线性趋势 |
| `poly` | 去除多项式趋势 |
| `loess` | 局部加权回归平滑 |
| `wavelet` | 小波变换去趋势 |
| `emd` | EMD 去趋势 |
| `ceemdan` | CEEMDAN 去趋势 |
| `median` | 中值滤波去趋势 |
## Curve Similarity - 曲线相似度
### CurveSimilarity
曲线相似度计算,支持动态时间规整(DTW)。
```python
from pywayne.dsp import CurveSimilarity
cs = CurveSimilarity()
distance = cs.dtw(curve1, curve2, mode='global')
# 可选后端:auto(默认)、python、numba
fast_distance = cs.dtw(curve1, curve2, backend='numba')
```
**方法**:
| 方法 | 说明 |
|------|------|
| `dtw(x, y, mode='global', *params, backend='auto')` | 计算两条曲线的 DTW 距离 |
| `mode` | str | 'global'(全局)或 'local'(局部) |
| `backend` | str | 'auto'、'python' 或 'numba';auto 在 Numba 不可用时回退 Python |
Numba 是可选性能依赖,可通过 `pip install "pywayne[performance]"` 安装。首次调用会进行 JIT 编译,
重复调用或较长曲线收益更明显。Python 和 Numba 后端保持相同的评分语义。
## Other Tools - 其他工具
### OneEuroFilter
一欧罗滤波器,用于平滑信号并减少延迟。
```python
from pywayne.dsp import OneEuroFilter
# 初始化
euro_filter = OneEuroFilter(te=0.02, mincutoff=1.0, beta=0.007, dcutoff=1.0)
# 应用滤波
smooth_value = euro_filter.apply(new_measurement, te=0.02)
```
**参数说明**:
| 参数 | 类型 | 说明 |
|------|------|------|
| `te` | float | 采样时间(秒),自动推断默认值 |
| `mincutoff` | float | 最小截止频率 |
| `beta` | float | 调整速率参数 |
| `dcutoff` | float | 导数截止频率 |
### WelfordStd
使用 Welford 算法进行在线标准差计算。
```python
from pywayne.dsp import WelfordStd
std_calculator = WelfordStd(win=50)
for sample in data_stream:
current_std = std_calculator.apply(sample)
# 使用 current_std 进行判断
```
**方法**:
| 方法 | 说明 |
|------|------|
| `__init__(win)` | 初始化,指定窗口大小 |
| `apply(val)` | 更新标准差计算,返回当前窗口标准差 |
## 应用场景
| 场景 | 使用函数 |
|------|----------|
| 心电图信号分析 | `butter_bandpass_filter`, `peak_det` |
| 传感器数据平滑 | `OneEuroFilter`, `ButterworthFilter` |
| 数据预处理 | `SignalDetrend` |
| 曲线相似度比较 | `CurveSimilarity.dtw` |
| 质量监控 | `WelfordStd` |
Versions
The database has no other skills named pywayne-dsp.
Database snapshot: 13.11.2026 · package version 248e3a9f3285 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100timeseries-detrending
Tools and techniques for detrending time series data in macroeconomic analysis. Use when working with economic time series that need to be decomposed into trend and cyclical components. Covers HP filter, log transformations for growth series, and correlation analysis of business cycles. - 100dsp-audio-filter-effects
Apply digital signal processing (DSP) algorithms in Python and C++ for audio filtering (IIR, FIR, Butterworth, Chebyshev), spectral analysis via Fast Fourier Transform (FFT), dynamic range compression, and modulation effects. Trigger when developing audio DSP algorithms, sound… - 100aeon
This skill should be used for time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search. Use when working with temporal data, sequential patterns, or time-indexed observations…