!pip install siml import pandas as pd import numpy as np from matplotlib import pyplot as plt import seaborn as sns from scipy.fftpack import fft from scipy.stats import kstest as ks from siml.sk_utils import * from siml.signal_analysis_utils import * from collections import defaultdict, Counter from scipy.signal import chirp, spectrogram, stft from functools import reduce import heapq as hp import scipy.signal as signal

df=pd.read_csv(r'sinaisExemplo.csv',delimiter=";") br2=pd.read_csv(r'BR23_5118_TB23_P2.csv', delimiter=';') br3=pd.read_csv(r'BR24_5120_TB24_P3.csv', delimiter=';')

df

br2

br3

def plot_signal(signal_data, x, xlabel, ylabel): ax = sns.lineplot(y=signal_data, x=x) ax.set(xlabel=xlabel, ylabel=ylabel) plt.show()

plot_signal(df['Channel1WaveformData_1'], df['Unnamed: 0'], 'Time (s)', 'Velocity (mm/s)')

plot_signal(br2['Channel: 1'], br2['Time (ms)'], 'Time (ms)', 'Velocity (mm/s)')

def get_fft_values(y_values, T, N, f_s): f_values = np.linspace(0.0, 1.0/(2.0*T), N//2) fft_values_ = fft(y_values) fft_values = 2.0/N * np.abs(fft_values_[0:N//2]) return f_values, fft_values

def get_and_plot_fft(x, y, plot=False): signal_y = y.to_numpy() signal_x = x.to_numpy() max_x = signal_x[-1] if (max_x > 1000): max_x = max_x/1000 t_n = max_x # sampling rate N = len(signal_y) # sample size T = t_n / N #sampling time f_s = 1/T #sampling freq x_value = signal_x composite_y_value = signal_y t_n = 10 N = 1000 T = t_n / N f_s = 1/T f_values, fft_values = get_fft_values(composite_y_value, T, N, f_s) if plot: plt.plot(f_values, fft_values, linestyle='-', color='blue') plt.xlabel('Frequency [Hz]', fontsize=16) plt.ylabel('Amplitude', fontsize=16) plt.title("Frequency domain of the signal", fontsize=22) plt.grid() plt.show() return (f_values, fft_values)

def get_n_highest(freq, n=1): h = [] for i in range(len(freq)): if (len(h) == n): if(h[0][0] < freq[i]): hp.heappop(h) hp.heappush(h, (freq[i], i)) else: hp.heappush(h, (freq[i], i)) return h

plot_signal(df['Channel1WaveformData_1'], df['Unnamed: 0'], 'Time (s)', 'Velocity (mm/s)') t, v = get_and_plot_fft(df['Unnamed: 0'], df['Channel1WaveformData_1'], plot=True) dominant_frequency = get_n_highest(v)

def plot_stft(x, y, plot=False): signal_y = y.to_numpy() signal_x = x.to_numpy() max_x = round(signal_x[-1]) if (max_x > 1000): max_x = round(max_x/1000) fs = round(len(signal_x)/max_x) t_final = max_x t = np.linspace(1, t_final, fs*t_final) f, t, Zxx = stft(signal_y, fs, nperseg=256) if plot: plt.figure(figsize=(30, 10)) plt.pcolormesh(t, f, np.abs(Zxx), shading='gouraud') plt.title('STFT Magnitude', fontsize=22) plt.ylabel('Frequency [Hz]') plt.xlabel('Time [sec]') plt.axis([0, max_x, 0, 300]) plt.show() df_f = pd.DataFrame(f,columns = ['freq']).reset_index() matriz = pd.DataFrame(np.abs(Zxx)) df_matriz = pd.DataFrame(pd.DataFrame(np.abs(matriz)).idxmax(),columns=['idx'])\ .reset_index(drop=True)\ .merge(df_f,left_on = 'idx',right_on = 'index') ln_matriz = np.log(df_matriz.loc[df_matriz['freq']!=0])['freq'] mean = ln_matriz.mean() var = ln_matriz.std()**2 if plot: df_matriz.freq.hist() mean = ln_matriz.mean() var = ln_matriz.std()**2 return mean,var

plot_signal(df['Channel1WaveformData_1'], df['Unnamed: 0'], 'Time (s)', 'Velocity (mm/s)') t, v = get_and_plot_fft(df['Unnamed: 0'], df['Channel1WaveformData_1'], plot=True) dominant_frequency = get_n_highest(v) mean,var = plot_stft(df['Unnamed: 0'], df['Channel1WaveformData_1'], plot=True)

mean,var = plot_stft(br2['Time (ms)'], br2['Channel: 1'], plot=True)

mean,var = plot_stft(br3['Time (ms)'], br3['Channel: 3'], plot=True)

# Filtering a noisy signal. def pass_band_filter(signal_x, signal_y, highcut, plot=False): max_x = signal_x.iloc[-1] #final time if (max_x > 1000): max_x = round(max_x/1000) fs = round(len(signal_x)/max_x) order = 9 x= signal_y.to_numpy() a = 0.8 nsamples = len(signal_x.to_numpy()) t = np.linspace(0, max_x, nsamples, endpoint=True) #time array # and now let's filter it sys = signal.butter(order,[highcut], btype='lowpass', fs=fs, output='sos') y = signal.sosfilt(sys, x) # compute the output signal if plot: plt.figure() plt.clf() plt.plot(t, y, label='Filtered signal') plt.xlabel('Time (seconds)') plt.hlines([-a, a], 0, max_x, linestyles='--') plt.grid(True) plt.axis('tight') plt.legend(loc='upper left') plt.show() return y

highcut = 20

plot_signal(df['Channel1WaveformData_1'], df['Unnamed: 0'], 'Time (s)', 'Velocity (mm/s)') pass_band_filter(df['Unnamed: 0'], df['Channel1WaveformData_1'], highcut,plot = True)

plot_signal(br2['Channel: 1'],br2['Time (ms)'], 'Time (ms)', 'Velocity (mm/s)') pass_band_filter(br2['Time (ms)'], br2['Channel: 1'], highcut,plot = True)

plot_signal(br3['Channel: 2'],br3['Time (ms)'], 'Time (ms)', 'Velocity (mm/s)') pass_band_filter(br3['Time (ms)'], br3['Channel: 2'], highcut,plot = True)

plot_signal(br3['Channel: 3'],br3['Time (ms)'], 'Time (ms)', 'Velocity (mm/s)') pass_band_filter(br3['Time (ms)'], br3['Channel: 3'], highcut,plot = True)

plot_signal(df['Channel1WaveformData_5'],df['Unnamed: 0'], 'Time (s)', 'Velocity (mm/s)') pass_band_filter(df['Unnamed: 0'], df['Channel1WaveformData_5'],highcut,plot = True)

dfFinal = df.iloc[:,:5] dfFinal.rename(columns= {'Unnamed: 0': 'time', 'Channel1WaveformData_1':'wave1', 'Channel2WaveformData_1':'wave2', 'Channel3WaveformData_1':'wave3', 'Channel1WaveformData_2':'wave4'} ,inplace = True ) br2.rename( columns={ 'Time (ms)':'time', 'Channel: 1':'wave1', 'Channel: 2':'wave2', 'Channel: 3':'wave3' },inplace = True ) br3.rename( columns={ 'Time (ms)':'time', 'Channel: 1':'wave4', 'Channel: 2':'wave5', 'Channel: 3':'wave6' },inplace = True ) br = br2.merge(br3,on = 'time')

for i in range(len(dfFinal.columns)-1): dfFinal.iloc[:,i+1]= pass_band_filter(dfFinal.iloc[:,0], dfFinal.iloc[:,i+1],highcut) for i in range(len(br.columns)-1): br.iloc[:,i+1]= pass_band_filter(br.iloc[:,0], br.iloc[:,i+1],highcut)

#df featuresNames = ['Amplitude','Time','Frequency_domain','Mean','Variance'] features_row = [] for i in range(len(dfFinal.columns)-1): #Amplitude wave = dfFinal.iloc[:,i+1] amplitude = wave.max() #Time mom = dfFinal.iloc[dfFinal.iloc[:,i+1].idxmax(),0] #Frequency domain t, v = get_and_plot_fft(dfFinal['time'], dfFinal.iloc[:,i+1], plot=False) highest = get_n_highest(v) dominant_frequency= t[highest[0][1]] #Mean and variance mean,var = plot_stft(dfFinal['time'], dfFinal.iloc[:,i+1], plot=False) features_row = features_row + [[amplitude,mom,dominant_frequency,mean,var]] #-------------------------------------------------------- #br - Underwater for i in range(len(br.columns)-1): #Amplitude wave = br.iloc[:,i+1] amplitude = wave.max() #Time mom = br.iloc[br.iloc[:,i+1].idxmax(),0]/1000 #Frequency domain t, v = get_and_plot_fft(br['time'], br.iloc[:,i+1], plot=False) highest = get_n_highest(v) dominant_frequency= t[highest[0][1]] #Mean and variance mean,var = plot_stft(br['time'], br.iloc[:,i+1], plot=False) features_row = features_row + [[amplitude,mom,dominant_frequency,mean,var]]

dfFeatures = pd.DataFrame(features_row,columns=featuresNames) dfFeatures