如何计算蟒蛇的PSD、中频和平均频率？

Question

我正在尝试用滑动窗口从python上的EMG数据中提取频率特性。我没有太多的频率分析知识，所以如果我有一些错误的概念，我会提前道歉。

我试图遵循这个网站的定义：

MDF

更多信息

我的问题是我是否正确地进行了计算，如果没有，我如何进行计算？

Windows_Size = 0.125 
Overlap = 0.5 
Bins = 256 

START, END = df_filtered['Time'].min(), df_filtered['Time'].max()
Fs = 2000 #EMG Sampling Frequency
P = 1.0 / Fs

Windows = np.arange(START + Windows_Size, END, Windows_Size * (1 - Overlap))
FREQ = []

for w in WINDOWS:
  win_start, win_end = w - Windows_Size, w

  for var in ['Biceps_Femoris_Sq_Correct']:
    value = df_filtered.loc[(win_start <= df_filtered['Time']) & (df_filtered['Time'] < win_end), var].values

    fft = np.fft.fft(value * np.hamming(value.shape[0]), n=Bins)[1:Bins//2]
    freq = np.fft.fftfreq(Bins, P)[1:Bins//2]
    amp = np.abs(fft)

    energy = amp ** 2 # Is this the right way to calculate power spectrum? 

    median_freq = energy/2 
    #mean_freq = EMG power spectrum is divided into two regions with equal 
    #amplitude... is this right?

    mean_freq = np.sum(energy * freq) / np.sum(energy) 

平均频率作为肌电图功率谱的乘积之和计算，频率除以功率谱的总和。这是计算平均频率的正确方法吗？

Answer 1

MDF应该类似于

energy_cumsum = np.cumsum(energy)
MDF = freq[np.where(energy_cumsum>np.max(energy_cumsum)/2)[0][0]]