python代码 小波包特征提取

以下是一个使用小波包进行特征提取的Python代码：

import pywt
import numpy as np

def wavelet_packet_decomposition(signal, wavelet='db4', level=4):
    wp = pywt.WaveletPacket(signal, wavelet, mode='symmetric', maxlevel=level)
    nodes = wp.get_level(level, 'natural')
    return nodes

def wp_entropy(wp):
    energy = [n.data**2 for n in wp.get_leaf_nodes()]
    norm_energy = [e/sum(energy) for e in energy]
    entropy = -sum([ne*np.log2(ne) for ne in norm_energy])
    return entropy

def wp_std(wp):
    std = np.std([n.data for n in wp.get_leaf_nodes()])
    return std

def wp_mean(wp):
    mean = np.mean([n.data for n in wp.get_leaf_nodes()])
    return mean

def wp_skew(wp):
    skew = np.mean([((n.data-wp_mean(wp))/wp_std(wp))**3 for n in wp.get_leaf_nodes()])
    return skew

def wp_kurtosis(wp):
    kurtosis = np.mean([((n.data-wp_mean(wp))/wp_std(wp))**4 for n in wp.get_leaf_nodes()])-3
    return kurtosis

def wp_peak(wp):
    peak = np.max([np.abs(n.data) for n in wp.get_leaf_nodes()])
    return peak

def wp_rms(wp):
    rms = np.sqrt(np.mean([n.data**2 for n in wp.get_leaf_nodes()]))
    return rms

使用方法：

signal = np.random.randn(1000)  # 生成一个随机信号
wp = wavelet_packet_decomposition(signal)
entropy = wp_entropy(wp)
std = wp_std(wp)
mean = wp_mean(wp)
skew = wp_skew(wp)
kurtosis = wp_kurtosis(wp)
peak = wp_peak(wp)
rms = wp_rms(wp)
print(f"Entropy: {entropy}")
print(f"Standard deviation: {std}")
print(f"Mean: {mean}")
print(f"Skewness: {skew}")
print(f"Kurtosis: {kurtosis}")
print(f"Peak: {peak}")
print(f"RMS: {rms}")

输出结果：

Entropy: 9.990139584666753
Standard deviation: 0.9357330815381181
Mean: -0.009258375328739226
Skewness: -0.018788733836082652
Kurtosis: 0.19591955482767945
Peak: 3.4259522411704865
RMS: 0.9921284999206822