Programming in Python

Plotting multicolored line in python with segments

FFT using Scipy Package

Data Normalization (Min-Max)

Conditional Replacement of number in NumPy array

Design a Low Pass Filter in Python

Programming in Python

Plotting multicolored line in python with segments

# Plotting different segment in different color

import numpy as np

import matplotlib.pyplot as plt

# Create test data

section_one_x1 = np.linspace(0,2.5,100)

section_one_y1 = section_one_x1**3

section_two_x2 = np.linspace(2.5,5,100)

section_two_y2 = section_two_x2**3

# Plot segment one in blue color

plt.plot(section_one_x1, section_one_y1, 'b')

# Plot segment two in red color

plt.plot(section_two_x2, section_two_y2, 'r')

plt.grid(True)

plt.legend(["No Arc", "Arc"])

plt.show()

Fig. Multicolored plot

# Plotting sine wave with different segment in different color

import numpy as np

from numpy import genfromtxt

import matplotlib.pyplot as plt

import math

# Create test data

section_one_x1 = np.linspace(0,3.1416*2,500)

section_one_y1 = np.sin(section_one_x1)

section_two_x2 = np.linspace(3.1416*2,4*3.1416,500)

section_two_y2 = np.sin(section_two_x2)

plt.figure(dpi=200)

# Plot segment one in blue color

plt.plot(section_one_x1, section_one_y1, 'b')

# Plot segment two in red color

plt.plot(section_two_x2, section_two_y2, 'r')

plt.grid(True)

plt.legend(["Blue Section", "Red Section"], loc="best")

plt.xlabel("Value of x")

plt.ylabel("Sine (x)")

plt.show()

Python tutorials:

FFT using Scipy Package

# Function fot FFT of samples from scipy.fft import fft, fftfreq # This fft perfectly matches with matlab fft, I checkeddef FFT(data, N=800, Fs=40): # N is Number of sample points, Fs is sampling rate in kHz; to show full Hz, set Fs as full # sample spacing T = 1.0 / Fs# x = np.linspace(0.0, N*T, N, endpoint=False)# print("x=" + str(x.shape)) yf = fft(data) xf = fftfreq(N, T)[:N//2]# import matplotlib.pyplot as plt # plt.plot(xf[1:N//2], 2.0/N * np.abs(yf[1:N//2]), '-r') yf1=2.0/N * np.abs(yf[2:N//2]) # extract the one sided values xf1 = xf[2:N//2]# print(xf1)# print(yf1) print(xf1.shape) print(yf1.shape) return (xf1, yf1)
# FFT of all the arc samples and stored valuesi1=17 # Signal numberi2=35 # Signal numberx1, f1=FFT(train_40[i1]) # i1 is Normal, i2 is Arcx2, f2=FFT(train_40[i2])print(train_label[i1])print(train_label[i2])plt.semilogy(x1, f1, "-b")plt.semilogy(x2, f2, "-r")plt.legend([train_label[i1], train_label[i2]])plt.xlabel("Frequency (kHz)")plt.ylabel("FFT Amplitude")plt.show()print(np.max(f1)-np.min(f1))
plt.plot(x1, f1, "-b")plt.plot(x2, f2, "-r")plt.legend([train_label[i1], train_label[i2]])plt.xlabel("Frequency (kHz)")plt.ylabel("FFT Amplitude")plt.show()
plt.plot(train_40[i1])plt.plot(train_40[i2])plt.legend([train_label[i1], train_label[i2]])plt.show()

Data Normalization (Min-Max)

# LOADING CLEAN DATABASEtrain = np.load("4class_db/dt_tr_1c_10kHz.npy")train_40 = np.load("4class_db/dt_tr_1c_40kHz.npy") train_label = np.load("4class_db/lb_tr_1c_40kHz.npy") valid = np.load("4class_db/dt_vl_1c_10kHz.npy") valid_40 = np.load("4class_db/dt_vl_1c_40kHz.npy") valid_label = np.load("4class_db/lb_vl_1c_40kHz.npy")test = np.load("4class_db/dt_ts_1c_10kHz.npy") test_40 = np.load("4class_db/dt_ts_1c_40kHz.npy") test_label = np.load("4class_db/lb_ts_1c_40kHz.npy")
print(train.shape)print(test.shape)print(valid.shape)
print(train_40.shape)print(test_40.shape)print(valid_40.shape)
# NORMALIZATION (Min Max)def norm(dt): dt=(dt-np.min(dt))/(np.max(dt)-np.min(dt)) return dt
train=norm(train)valid=norm(valid)test=norm(test)train_40=norm(train_40)valid_40=norm(valid_40)test_40=norm(test_40)

Conditional Replacement of number in NumPy array

def mult2binary(lb): '''Convert multiclass label into binary for arc (1) and normal (0)''' lb = np.where((lb == 1) | (lb == 3) | (lb==5) | (lb == 7), 10, lb) lb = np.where((lb == 0)| (lb == 2) | (lb==4) | (lb == 6), 11, lb) lb = np.where((lb == 10), 0, lb) lb = np.where((lb == 11), 1, lb) return lb
# Convert multiclass label into binary label arc (1) and normal (0)train_label = mult2binary(train_label)test_label = mult2binary(test_label)valid_label = mult2binary(valid_label)
# # converting all types of labels (y values) into categorical data# OneHot = np_utils.to_categorical(train_label, num_classes=8)# OneHot_valid_label = np_utils.to_categorical(valid_label, num_classes=8)# OneHot_test_label = np_utils.to_categorical(test_label, num_classes=8)

Design a Low Pass Filter in Python

My requirements of the filters are:

Sample Period — 5 sec (t)

Sampling Freq — 30 samples / s , i.e 30 Hz (fs)

Total Samples — (fs x t) = 150

Signal Freq = 6 signal / 5 sec = 1.2 Hz

Nyquist Frequency = 0.5 * fs

order = Polynomial order of the signal

Please do the followings:

1. Create some sample data with noise

2. Filter implementation using scipy

3. Filter and plot the input output data.

## CODE SNIPPET IS GIVEN BELOW

import numpy as np

import plotly.graph_objects as go

from scipy.signal import butter, filtfilt

# Filter requirements

T = 5.0 # Sample Period

fs = 30.0 # sample rate, Hz

cutoff = 2.0 # desired cutoff frequency of the filter, Hz

nyq = 0.5 * fs # Nyquist Frequency

order = 2 # sin wave can be approx represented as quadratic

n = int(T * fs) # total number of samples

# sin wave

t = np.linspace(0, T, n, endpoint=False)

sig = np.sin(1.2 * 2 * np.pi * t)

# Add some noise

noise = 1.5 * np.cos(9 * 2 * np.pi * t) + 0.5 * np.sin(12.0 * 2 * np.pi * t)

data = sig + noise

def butter_lowpass_filter(data, cutoff, fs, order):

normal_cutoff = cutoff / nyq

# Get the filter coefficients

b, a = butter(order, normal_cutoff, btype='low', analog=False)

y = filtfilt(b, a, data)

return y

# Filter the data, and plot both the original and filtered signals

y = butter_lowpass_filter(data, cutoff, fs, order)

fig = go.Figure()

fig.update_layout(width=800)

fig.add_trace(go.Scatter(

x=t,

y=data,

mode='lines',

name='signal with noise'

))

fig.add_trace(go.Scatter(

x=t,

y=y,

mode='lines',

name='filtered signal'

))

fig.show()

Actual output will be interractive if plotted in Jupyter Notebook. A snapshot of the output is given here.