We Keep Coding

Interpolating measured sine wave using python

I have 2 sampled sine waves obtained as a measurement from a DSO. The sampling rate of the DSO is 160 GSa/s and my signal is 60 GHz. I need to find the phase difference between the two sine waves. Both are the same frequency. However, the sampling rate is not enough to accurately determine the phase. Is there any way to interpolate the measured signal to get a better sine wave and then calculate the phase difference?

You may fit to sine functions, but for the phase difference (delta phi=2pi frequency delta t) it would be sufficient to detect and compare the zero-crossings (respective a possible constant offset), which may be found from a segment of your series by an interpolation like
w=6.38 # some radian frequency
t = np.linspace(0, 0.5) # time interval containing ONE zero-crossing
delta_phi=0.1 # some phase difference
x = np.sin(w*t-delta_phi) # x(t)
f = interpolate.interp1d(x, t) # interpolate t(x), default is linear
delta_t = f(0) # zero-crossing time referred to t=0
delta_phi_detected= w*delta_t
You need to relate two adjacent zero-crossings of your signals.
Alternatively, you may obtain an average value by multiplication of both signals and numerical integration over time T which converges to (T/2)cos(delta_phi), if both signals have (or are made to) zero mean value.

I have an array of samples at equal intervals of 25 ms. I want to determine the frequency spectrum for the fundamental

From say N=1000 voltage samples at 1 ms sample rate. I need to find precisely with python/numpy the amplitude and angle of the fundamental, which is between 45 and 55 Hz as well as any side bands, that may exist.
Do I need a phase lock loop to do that, or can it be done without ?

Your measurement frequency is fundamentally too low and should be more than double your expected event frequency!
measured: 0.025s
event range: 0.0182-0.0222s
more here: https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem

A phase lock loop is be a reasonable approach to estimate the frequency etc. of the fundamental. Supposing you have collected the N samples up front, another way to do the analysis is:
Apply a window, like np.hanning(N), by multiplying it pointwise with the samples.
Compute the spectrum with np.fft.rfft. For a sampling interval of Ts seconds, the nth element of the resulting array is the DFT coefficient for frequency n * N * Ts in units of Hz (or for the values N=1000, Ts=0.001, simple n Hz).
Find the bin with the peak magnitude between 45 and 55. The location of the peak gives the frequency of the fundamental. You can interpolate a polynomial (np.polyfit) across a few neighboring bins and find its peak to get a more precise estimate. Magnitude and complex phase of the peak give the amplitude and phase (angle) of the fundamental.
Plot the magnitude of the spectrum to look for sidebands.

How to extract features from FFT?

I am gathering data from X, Y and Z accelerometer sensors sampled at 200 Hz. The 3 axis are combined into a single signal called 'XYZ_Acc'. I followed tutorials on how to transform time domain signal into frequency domain using scipy fftpack library.
The code I'm using is the below:
from scipy.fftpack import fft
# get a 500ms slice from dataframe
sample500ms = df.loc[pd.to_datetime('2019-12-15 11:01:31.000'):pd.to_datetime('2019-12-15 11:01:31.495')]['XYZ_Acc']
f_s = 200 # sensor sampling frequency 200 Hz
T = 0.005 # 5 milliseconds between successive observation T =1/f_s
N = 100 # 100 samples in 0.5 seconds
f_values = np.linspace(0.0, f_s/2, N//2)
fft_values = fft(sample500ms)
fft_mag_values = 2.0/N * np.abs(fft_values[0:N//2])
Then I plot the frequency vs the magnitude
fig_fft = plt.figure(figsize=(5,5))
ax = fig_fft.add_axes([0,0,1,1])
ax.plot(f_values,fft_mag_values)
Screenshot:
My difficulty now is how to extract features out of this data, such as Irregularity, Fundamental Frequency, Flux...
Can someone guide me into the right direction?
Update 06/01/2019 - adding more context to my question.
I'm relatively new in machine learning, so any feedback is appreciated. X, Y, Z are linear acceleration signals, sampled at 200 Hz from a smart phone. I'm trying to detect road anomalies by analysing spectral and temporal statistics.
Here's a sample of the csv file which is being parsed into a pandas dataframe with the timestamp as the index.
X,Y,Z,Latitude,Longitude,Speed,timestamp
0.8756,-1.3741,3.4166,35.894833,14.354166,11.38,2019-12-15 11:01:30:750
1.0317,-0.2728,1.5602,35.894833,14.354166,11.38,2019-12-15 11:01:30:755
1.0317,-0.2728,1.5602,35.894833,14.354166,11.38,2019-12-15 11:01:30:760
1.0317,-0.2728,1.5602,35.894833,14.354166,11.38,2019-12-15 11:01:30:765
-0.1669,-1.9912,-4.2043,35.894833,14.354166,11.38,2019-12-15 11:01:30:770
-0.1669,-1.9912,-4.2043,35.894833,14.354166,11.38,2019-12-15 11:01:30:775
-0.1669,-1.9912,-4.2043,35.894833,14.354166,11.38,2019-12-15 11:01:30:780
In answer to 'francis', two columns are then added via this code:
df['XYZ_Acc_Mag'] = (abs(df['X']) + abs(df['Y']) + abs(df['Z']))
df['XYZ_Acc'] = (df['X'] + df['Y'] + df['Z'])
'XYZ_Acc_Mag' is to be used to extract temporal statistics.
'XYZ_Acc' is to be used to extract spectral statistics.
Data 'XYZ_Acc_Mag' is then re sampled in 0.5 second frequency and temporal stats such as mean, standard-deviation, etc have been extracted in a new dataframe. Pair plots reveal the anomaly shown at time 11:01:35 in the line plot above.
Now back to my original question. I'm re sampling data 'XYZ_Acc', also at 0.5 seconds, and obtaining the magnitude array 'fft_mag_values'. The question is how do I extract temporal features such as Irregularity, Fundamental Frequency, Flux out of it?

Since 'XYZ_Acc' is defined as a linear combination of the components of the signal, taking its DFT makes sense. It is equivalent to using a 1D accelometer in direction (1,1,1). But a more physical energy-related viewpoint can be adopted.
Computing the DFT is similar to writing the signal as a sum of sines. If the acceleration vector writes :
The corresponding velocity vector could write:
and the specific kinetic energy writes:
This method requires computing the DFT a each component before the magnitude corresponding to each frequency.
Another issue is that the DFT is intended to compute the Discrete Fourrier Transform of a periodic signal, that signal being build by periodizing the frame. Nevertheless, the actual frame is never a period of a periodic signal and repeating the period creates artificial discontinuities at the end/begin of the frame. The effects strong discontinuities in the spectral domain, deemded spectral leakage, could be reduced by windowing the frame. Computing the real-to-complex DFT result in a power distribution, featuring peaks at particular frequencies.
In addition the frequency of a given peak is better estimated as the mean frequency with respect to power density, as shown in Why are frequency values rounded in signal using FFT?
Another tool to estimate fundamental frequencies is to compute the autocorrelation of the signal: it is higher near the periods of the signal. Since the signal is a vector of 3 components, an autocorelation matrix can be built. It is a 3x3 Hermitian matrix for each time and therefore features real eigenvalues. The maxima of the higher eigen value can be picture as the magnitude of vaibrations while the correponding eigenvector is a complex direction, somewhat similar to the direction of vibrations combined to angular offsets. The angular offset may signal an ellipsoidal vibration.
Here is a fake signal, build by adding a guassian noise and sine waves:
Here is the power density spectrum for a given frame overlapping on sine wave:
Here is the resulting eigenvalues of the autocorrelation of the same frame, where the period of the 50Hz sine wave is visible. Vertical scaling is wrong:
Here goes a sample code:
import matplotlib.pyplot as plt
import numpy as np
import scipy.signal
n=2000
t=np.linspace(0.,n/200,num=n,endpoint=False)
# an artificial signal, just for tests
ax=0.3*np.random.normal(0,1.,n)
ay=0.3*np.random.normal(0,1.,n)
az=0.3*np.random.normal(0,1.,n)
ay[633:733]=ay[633:733]+np.sin(2*np.pi*30*t[633:733])
az[433:533]=az[433:533]+np.sin(2*np.pi*50*t[433:533])
#ax=np.sin(2*np.pi*10*t)
#ay=np.sin(2*np.pi*30*t)
#az=np.sin(2*np.pi*50*t)
plt.plot(t,ax, label='x')
plt.plot(t,ay, label='y')
plt.plot(t,az, label='z')
plt.xlabel('t, s')
plt.ylabel('acc, m.s^-2')
plt.legend()
plt.show()
#splitting the sgnal into frames of 0.5s
noiseheight=0.
for i in range(2*(n/200)):
print 'frame', i,' time ', i*0.5, ' s'
framea=np.zeros((100,3))
framea[:,0]=ax[i*100:i*100+100]
framea[:,1]=ay[i*100:i*100+100]
framea[:,2]=az[i*100:i*100+100]
#for that frame, apply window. Factor 2 so that average remains 1.
window = np.hanning(100)
framea[:,0]=framea[:,0]*window*2
framea[:,1]=framea[:,1]*window*2
framea[:,2]=framea[:,2]*window*2
#DFT transform.
hatacc=np.fft.rfft(framea,axis=0, norm=None)
# scaling by length of frame.
hatacc=hatacc/100.
#computing the magnitude : all non-zero frequency are doubled to merge energy in bin N-k exp(-2ik/n) to bin k
accmag=2*(np.abs(hatacc[:,0])*np.abs(hatacc[:,0])+np.abs(hatacc[:,1])*np.abs(hatacc[:,1])+np.abs(hatacc[:,2])*np.abs(hatacc[:,2]))
accmag[0]=accmag[0]*0.5
#first frame says something about noise
if i==0:
noiseheight=2.*np.max(accmag)
if np.max(accmag)>noiseheight:
peaks, peaksdat=scipy.signal.find_peaks(accmag, height=noiseheight)
timestep=0.005
freq= np.fft.fftfreq(100, d=timestep)
#see https://stackoverflow.com/questions/54714169/why-are-frequency-values-rounded-in-signal-using-fft/54775867#54775867
# frequencies of peaks are better estimated as mean frequency of peak, with respect to power density
for ind in peaks:
totalweight=accmag[ind-2]+accmag[ind-1]+accmag[ind]+accmag[ind+1]+accmag[ind+2]
totalweightedfreq=accmag[ind-2]*freq[ind-2]+accmag[ind-1]*freq[ind-1]+accmag[ind]*freq[ind]+accmag[ind+1]*freq[ind+1]+accmag[ind+2]*freq[ind+2]
print 'found peak at frequency' , totalweightedfreq/totalweight, ' of height', accmag[ind]
#ploting
plt.plot(freq[0:50],accmag[0:50], label='||acc||^2')
plt.xlabel('frequency, Hz')
plt.ylabel('||acc||^2, m^2.s^-4')
plt.legend()
plt.show()
#another approach to find fundamental frequencies: computing the autocorrelation of the windowed signal and searching for maximums.
#building the autocorellation matrix
autocorr=np.zeros((100,3,3), dtype=complex)
acxfft=np.fft.fft(framea[:,0],axis=0, norm=None)
acyfft=np.fft.fft(framea[:,1],axis=0, norm=None)
aczfft=np.fft.fft(framea[:,2],axis=0, norm=None)
acxfft[0]=0.
acyfft[0]=0.
aczfft[0]=0.
autocorr[:,0,0]=np.fft.ifft(acxfft*np.conj(acxfft),axis=0, norm=None)
autocorr[:,0,1]=np.fft.ifft(acxfft*np.conj(acyfft),axis=0, norm=None)
autocorr[:,0,2]=np.fft.ifft(acxfft*np.conj(aczfft),axis=0, norm=None)
autocorr[:,1,0]=np.fft.ifft(acyfft*np.conj(acxfft),axis=0, norm=None)
autocorr[:,1,1]=np.fft.ifft(acyfft*np.conj(acyfft),axis=0, norm=None)
autocorr[:,1,2]=np.fft.ifft(acyfft*np.conj(aczfft),axis=0, norm=None)
autocorr[:,2,0]=np.fft.ifft(aczfft*np.conj(acxfft),axis=0, norm=None)
autocorr[:,2,1]=np.fft.ifft(aczfft*np.conj(acyfft),axis=0, norm=None)
autocorr[:,2,2]=np.fft.ifft(aczfft*np.conj(aczfft),axis=0, norm=None)
# at a given time, the 3x3 matrix autocorr is Hermitian.
#Its eigenvalues are real, its unitary eigenvectors signals directions of vibrations and phase between components.
autocorreigval=np.zeros((100,3))
autocorreigvec=np.zeros((100,3,3), dtype=complex)
for j in range(100):
autocorreigval[j,:], autocorreigvec[j,:,:]=np.linalg.eigh(autocorr[j,:,:],UPLO='L')
peaks, peaksdat=scipy.signal.find_peaks(autocorreigval[:50,2], 0.3*autocorreigval[0,2])
cleared=np.zeros(len(peaks))
peakperiod=np.zeros(len(peaks))
for j in range(len(peaks)):
totalweight=autocorreigval[peaks[j]-1,2]+autocorreigval[peaks[j],2]+autocorreigval[peaks[j]+1,2]
totalweightedperiod=0.005*(autocorreigval[peaks[j]-1,2]*(peaks[j]-1)+autocorreigval[peaks[j],2]*(peaks[j])+autocorreigval[peaks[j]+1,2]*(peaks[j]+1))
peakperiod[j]=totalweightedperiod/totalweight
#cleared[0]=1.
fundfreq=1
for j in range(len(peaks)):
if cleared[j]==0:
print "found fundamental frequency :", 1.0/(peakperiod[j]), 'eigenvalue', autocorreigval[peaks[j],2],' dir vibration ', autocorreigvec[peaks[j],:,2]
for k in range(j,len(peaks),1):
mm=np.zeros(1)
np.floor_divide(peakperiod[k],peakperiod[j],out=mm)
if ( np.abs(peakperiod[k]-peakperiod[j]*mm[0])< 0.2*peakperiod[j] or np.abs(peakperiod[k]-(peakperiod[j])*(mm[0]+1))< 0.2*peakperiod[j]) :
cleared[k]=fundfreq
#else :
# print k,j,mm[0]
# print peakperiod[k], peakperiod[j]*mm[0], peakperiod[j]*(mm[0]+1) , peakperiod[j]
fundfreq=fundfreq+1
plt.plot(t[i*100:i*100+100],autocorreigval[:,2], label='autocorrelation, large eigenvalue')
plt.plot(t[i*100:i*100+100],autocorreigval[:,1], label='autocorrelation, medium eigenvalue')
plt.plot(t[i*100:i*100+100],autocorreigval[:,0], label='autocorrelation, small eigenvalue')
plt.xlabel('t, s')
plt.ylabel('acc^2, m^2.s^-4')
plt.legend()
plt.show()
The output is:
frame 0 time 0.0 s
frame 1 time 0.5 s
frame 2 time 1.0 s
frame 3 time 1.5 s
frame 4 time 2.0 s
found peak at frequency 50.11249238149811 of height 0.2437842149351196
found fundamental frequency : 50.31467771196368 eigenvalue 47.03344783764712 dir vibration [-0.11441502+0.00000000e+00j 0.0216911 +2.98101624e-18j
-0.9931962 -5.95276353e-17j]
frame 5 time 2.5 s
frame 6 time 3.0 s
found peak at frequency 30.027895460975156 of height 0.3252387031089667
found fundamental frequency : 29.60690406120401 eigenvalue 61.51059682797539 dir vibration [ 0.11384195+0.00000000e+00j -0.98335779-4.34688198e-17j
-0.14158908+3.87566125e-18j]
frame 7 time 3.5 s
found peak at frequency 26.39622018109896 of height 0.042081187689137545
found fundamental frequency : 67.65844834016518 eigenvalue 6.875616417422696 dir vibration [0.8102307 +0.00000000e+00j 0.32697001-8.83058693e-18j
0.48643275-4.76094302e-17j]
frame 8 time 4.0 s
frame 9 time 4.5 s
Frequencies 50Hz and 30Hz got caught as 50.11/50.31Hz and 30.02/29.60Hz and directions are quite accurate as well. The last feature at 26.39Hz/67.65Hz is likely garbage, as it features different frequencies for the two methods and lower magnitude/eigenvalue.
Regarding monitoring of road surface to improve maintenance, I know of a project at my compagny, called Aigle3D. A laser fitted at the back of a van scans the road at highway speed at milimetric accuracy. The van is also fitted with a server, cameras and other sensors, thus providing a huge amount of data on road geometry and defects, presently covering hundreds of km of the french national road network. Detecting and repairing small early defects and cracks may extend the life expectancy of the road at limited cost. If useful, data from accelerometers of daily users could indeed complete the monitoring system, allowing a faster reaction whenether a large pothole appears.

How to implement/perform DFT on a segment in python?

I am trying to write a simple program in python that will calculate and display DFT output of 1 segment.
My signal is 3 seconds long, I want to calculate DFT for every 10ms long segment. Sampling rate is 44100. So one segment is 441 samples long.
Since I am in the phase of testing this and original program is much larger(speech recognition) here is an isolated part for testing purposes that unfortunately behaves odd. Either that or my lack of knowledge on the subject.
I read somewhere that DFT input should be rounded to power of 2 so I arranged my array to 512 instead 441. Is this true?
If I am sampling at a rate of 44100, at most I can reach frequency of 22050Hz and for sample of length 512(~441) at least 100Hz ?
If 2. is true, then I can have all frequencies between 100hz and 22050hz in that 10ms segments, but the length of segment is 512(441) samples only, output of fft returns array of 256(220) values, they cannot contain all 21950 frequencies in there, can they?
My first guess is that the values in output of fft should be multiplied by 100, since 10ms is 100th of a second. Is this good reasoning?
The following program for two given frequencies 1000 and 2000 returns two spikes on graph at positions 24 and 48 in the output array and ~2071 and ~4156 on the graph. Since ratio of numbers is okay (2000:1000 = 48:24) I wonder if I should ignore some starting part of the fft output?
import matplotlib.pyplot as plt
import numpy as np
t = np.arange(0, 1, 1/512.0) # We create 512 long array
# We calculate here two sinusoids together at 1000hz and 2000hz
y = np.sin(2*np.pi*1000*t) + np.sin(2*np.pi*2000*t)
n = len(y)
k = np.arange(n)
# Problematic part is around here, I am not quite sure what
# should be on the horizontal line
T = n/44100.0
frq = k/T
frq = frq[range(n/2)]
Y = fft(y)
Y = Y[range(n/2)]
# Convert from complex numbers to magnitudes
iY = []
for f in Y:
iY.append(np.sqrt(f.imag * f.imag + f.real * f.real))
plt.plot(frq, iY, 'r')
plt.xlabel('freq (HZ)')
plt.show()

I read somewhere that the DFT input should be rounded to power of 2 so I arranged my array to 512 instead 441. Is this true?
The DFT is defined for all sizes. However, implementations of the DFT such as the FFT are generally much more efficient for sizes which can be factored in small primes. Some library implementations have limitations and do not support sizes other than powers of 2, but that isn't the case with numpy.
If I am sampling at a rate of 44100, at most I can reach frequency of 22050Hz and for sample of length 512(~441) at least 100Hz?
The highest frequency for even sized DFT will be 44100/2 = 22050Hz as you've correctly pointed out. Note that for odd sized DFT the highest frequency bin will correspond to a frequency slightly less than the Nyquist frequency. As for the minimum frequency, it will always be 0Hz. The next non-zero frequency will be 44100.0/N where N is the DFT length in samples (which gives 100Hz if you are using a DFT length of 441 samples and ~86Hz with a DFT length of 512 samples).
If 2) is true, then I can have all frequencies between 100Hz and 22050Hz in that 10ms segments, but the length of segment is 512(441) samples only, output of fft returns array of 256(220) values, they cannot contain all 21950 frequencies in there, can they?
First there aren't 21950 frequencies between 100Hz and 22050Hz since frequencies are continuous and not limited to integer frequencies. That said, you are correct in your realization that the output of the DFT will be limited to a much smaller set of frequencies. More specifically the DFT represents the frequency spectrum at discrete frequency step: 0, 44100/N, 2*44100/N, ...
My first guess is that the values in output of FFT should be multiplied by 100, since 10ms is 100th of a second. Is this good reasoning?
There is no need to multiply the FFT output by 100. But if you meant multiples of 100Hz with a DFT of length 441 and a sampling rate of 44100Hz, then your guess would be correct.
The following program for two given frequencies 1000 and 2000 returns two spikes on graph at positions 24 and 48 in the output array and ~2071 and ~4156 on the graph. Since ratio of numbers is okay (2000:1000 = 48:24) I wonder if I should ignore some starting part of the fft output?
Here the problem is more significant. As you declare the array
t = np.arange(0, 1, 1/512.0) # We create 512 long array
you are in fact representing a signal with a sampling rate of 512Hz instead of 44100Hz. As a result the tones you are generating are severely aliased (to 24Hz and 48Hz respectively). This is further compounded by the fact that you then use a sampling rate of 44100Hz for the frequency axis conversion. This is why the peaks are not appearing at the expected 1000Hz and 2000Hz frequencies.
To represent 512 samples of a signal sampled at a rate of 44100Hz, you should instead use
t = np.arange(0, 511.0/44100, 1/44100.0)
at which point the formula you used for the frequency axis would be correct (since it is based of the same 44100Hz sampling rate). You should then be able to see peaks near the expected 1000Hz and 2000Hz (the closest frequency bins of the peaks being at ~1033Hz and 1981Hz).

1) I read somewhere that DFT input should be rounded to power of 2 so
I aranged my array to 512 instead 441. Is this true?
Yes, DFT length should be a power of two. Just pad the input with zero to match 512.
2) If I am sampling at a rate of 44100, at most I can reach frequency
of 22050hz and for sample of length 512(~441) at least 100hz ?
Yes, the highest frequency you can get is half the the sampling rate, It's called the Nyquist frequency.
No, the lowest frequency bin you get (the first bin of the DFT) is called the DC component and marks the average of the signal. The next lowest frequency bin in your case is 22050 / 256 = 86Hz, and then 172Hz, 258Hz, and so on until 22050Hz.
You can get this freqs with the numpy.fftfreq() function.
3) If 2) is true, then I can have all frequencies between 100hz and
22050hz in that 10ms segments, but the length of segment is 512(441)
samples only, output of fft returns array of 256(220) values, they
cannot contain all 21950 frequencies in there, can they?
DFT doesn't lose the original signal's data, but it lacks accuracy when the DFT size is small. You may zero-pad it to make the DFT size larger, such as 1024 or 2048.
The DFT bin refers to a frequency range centered at each of the N output
points. The width of the bin is sample rate/2,
and it extends from: center frequency -(sample rate/N)/2 to center
frequency +(sample rate/N)/2. In other words, half of the bin extends
below each of the N output points, and half above it.
4) My first guess is that the values in output of fft should be
multiplied by 100, since 10ms is 100th of a second. Is this good
reasoning?
No, The value should not be multiplied if you want to preserve the magnitude.
The following program for two given frequencies 1000 and 2000 returns
two spikes on graph at positions 24 and 48 in the output array and
~2071 and ~4156 on the graph. Since ratio of numbers is okay
(2000:1000 = 48:24) I wonder if I should ignore some starting part of
the fft output?
The DFT result is mirrored in real input. In other words, your frequencies will be like this:
n 0 1 2 3 4 ... 255 256 257 ... 511 512
Hz DC 86 172 258 344 ... 21964 22050 21964 ... 86 0

How do I match lomb-scargle and FFT plots of same dataset?

I am doing some work, comparing the interpolated fft of the concentrations of some gases over a period, of which is unevenly sampled, with the lomb-scargle periodogram of the same data. I am using scipy's fft function to calculate the fourier transform and then squaring the modulus of this to give what I believe to be the power spectral density, in units of parts per billion(ppb) squared.
I can get the lomb-scargle plot to match almost the exact pattern as the FFT but never the same scale of magnitude, the FFT power spectral density always is higher, even though I thought the lomb-scargle power was power spectral density. Now the lomb code I am using:http://www.astropython.org/snippet/2010/9/Fast-Lomb-Scargle-algorithm, normalises the dataset taking away the average and dividing by 2 times the variance from the data, therefore I have normalised the FFT data in the same manner, but still the magnitudes do not match.
Therefore I did some more research and found that the normalised lomb-scargle power could unitless and therefore I cannot the plots match. This leads me to the 2 questions:
What units (if any) are the power spectral density of a normalised lim-scargle perioogram in?
How would I proceed to match my fft plot with my lomb-scargle plot, in terms of magnitude and pattern?
Thank you.

The squared modulus of the Fourier transform of a series is defined as the energy spectral density (ESD). You need to divide the ESD by the length of the series to convert to an estimate of power spectral density (PSD).
Units
The units of a PSD are [units]**2/[frequency] where [units] represents the units of your original series.
Normalization
To check for proper normalization, one can numerically integrate the PSD of a white noise (with known variance). If the integrated spectrum equals the variance of the series, the normalization is correct. A factor of 2 (too low) is not incorrect, though, and may indicate the PSD is normalized to be double-sided; in that case, just multiply by 2 and you have a properly normalized, single-sided PSD.
Using numpy, the randn function generates pseudo-random numbers that are Gaussian distributed. For example
10 * np.random.randn(1, 100)
produces a 1-by-100 array with mean=0 and variance=100. If the sampling frequency is, say, 1-Hz, the single-sided PSD will theoretically be flat at 200 units**2/Hz, from [0,0.5] Hz; the integrated spectrum would thus be 10, equaling the variance of the series.
Update
I modified the example included in the python code you linked to demonstrate the normalization for a normally distributed series of length 20, with variance 1, and sampling frequency 10:
import numpy
import lomb
numpy.random.seed(999)
nd = 20
fs = 10
x = numpy.arange(nd)
y = numpy.random.randn(nd)
fx, fy, nout, jmax, prob = lomb.fasper(x, y, 1., fs)
fNy = fx[-1]
fy = fy/fs
Si = numpy.mean(fy)*fNy
print fNy, Si, Si*2
This gives, for me:
5.26315789474 0.482185882163 0.964371764327
which shows you a few things:
The "Nyquist" frequency asked for is actually the sampling frequency.
The result needs to be divided by the sampling frequency.
The output is normalized for a double-sided PSD, so multiplying by 2 makes the integrated spectrum nearly 1.

In the time since this question was asked and answered, the AstroPy project has gained a Lomb-Scargle method, and this question is addressed in the documentation: http://docs.astropy.org/en/stable/stats/lombscargle.html#psd-normalization-unnormalized
In brief, you can compute a Fourier periodogram and compare it to the astropy Lomb-Scargle periodogram as follows
import numpy as np
from astropy.stats import LombScargle
def fourier_periodogram(t, y):
N = len(t)
frequency = np.fft.fftfreq(N, t[1] - t[0])
y_fft = np.fft.fft(y)
positive = (frequency > 0)
return frequency[positive], (1. / N) * abs(y_fft[positive]) ** 2
t = np.arange(100)
y = np.random.randn(100)
frequency, PSD_fourier = fourier_periodogram(t, y)
PSD_LS = LombScargle(t, y).power(frequency, normalization='psd')
np.allclose(PSD_fourier, PSD_LS)
# True
Since AstroPy is a common tool used in astronomy, I thought this might be more useful than an answer based on the code snippet mentioned above.