def current_rms(current, period_length=PERIOD_LENGTH):
"""Calculates the Current RMS (CRMS).
Let \\(I_{W(k)}\\) denote the \\(k\\)th of \\(N\\) periods of current
measurements, \\(\\langle t_1, t_2, ..., t_n \\rangle\\) the
concatenation of tuples \\(t_1, ..., t_n\\), and \\(\\nu = N \\mod 10\\).
Then:
\\[CRMS = \\begin{pmatrix}\\text{rms}(\\langle I_{W(1)}, I_{W(2)}, ...,
I_{W(10)}\\rangle) \\\\ \\text{rms}(\\langle I_{W(11)}, I_{W(12)}, ...,
I_{W(20)}\\rangle) \\\\ ... \\\\ \\text{rms}(\\langle I_{W(N - \\nu + 1)},
I_{W(N - \\nu + 2)}, ..., I_{W(N)}\\rangle)\\end{pmatrix}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
n_periods (int): Length of transient and steady state in periods
Returns:
numpy.ndarray: Transient steady states ratio as a (n_samples, 1)-dimensional array.
"""
n = int(math.floor(current.shape[1] / period_length / 10))
cutoff = n * 10 * period_length
first = rms(current[:, :cutoff].reshape(-1, n, 10 * period_length), axis=2)
# Handle window_size is not multiple of 10*period_length
if cutoff != current.shape[1]:
return np.hstack((first, rms(current[:, cutoff:])))
return first
def positive_negative_half_cycle_ratio(current, period_length=PERIOD_LENGTH):
"""Calculates Positive-negative half cycle ratio (PNR).
Let \\(I_{P_\\text{pos}}\\) and \\(I_{P_\\text{neg}}\\) be the RMS of 10
averaged positive and negative current half cycles. Then
\\[PNR = \\frac{\\min\\{I_{P_\\text{pos}}, I_{P_\\text{neg}}\\}}
{\\max\\{I_{P_\\text{pos}}, I_{P_\\text{neg}}\\}}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: PNR as a (n_samples, 1)-dimensional array.
"""
# Create indices for positive and negative half cycles of sin wave
idcs = np.arange(0, period_length * 10)
p_idcs = idcs[idcs % period_length < period_length / 2]
n_idcs = idcs[idcs % period_length >= period_length / 2]
p_current = current[:, p_idcs].reshape(-1, 10, period_length // 2)
n_current = current[:, n_idcs].reshape(-1, 10, period_length // 2) * -1
# Calculate RMS of averaged half cycles
p_n = np.hstack((rms(np.mean(p_current,
axis=1)), rms(np.mean(n_current, axis=1))))
return (np.min(p_n, axis=1) / np.max(p_n, axis=1)).reshape(-1, 1)
def apparent_power(voltage, current):
"""Calculates Apparent power (S).
\\[S = \\text{rms}(V) \\times \\text{rms}(I)\\]
Args:
voltage (numpy.ndarray): (n_samples, window_size)-dimensional array of voltage measurements.
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: Apparent power as a (n_samples, 1)-dimensional array.
"""
return rms(voltage) * rms(current)
def resistance_mean(voltage, current):
"""Calculates Resistance \\(R_\\text{mean}\\) (mean version) from
voltage and current arrays.
\\[R_\\text{mean} = \\frac{\\sqrt{\\text{mean}(V^2)}}{\\sqrt{\\text{mean}(I^2)}}\\]
Args:
voltage (numpy.ndarray): (n_samples, window_size)-dimensional array of voltage measurements.
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: Resistance as a (n_samples, 1)-dimensional array.
"""
return rms(voltage) / rms(current)
def inrush_current_ratio(current, period_length=PERIOD_LENGTH):
"""Calculates the Inrush current ratio (ICR).
Let \\(I_{W(k)}\\) denote the \\(k\\)th of \\(N\\) periods of current
measurements. Then \\(I_{P(k)} = \\text{rms}(I_{W(k)})\\) and
\\[ICR = \\frac{I_{P(1)}}{I_{P(N)}}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: Inrush current ratio as a (n_samples, 1)-dimensional array.
"""
return rms(current[:, :period_length]) / rms(current[:, -period_length:])
def temporal_centroid(current,
mains_frequency=POWER_FREQUENCY,
period_length=PERIOD_LENGTH):
"""Calculates the Temporal centroid \\(C_t\\).
Let \\(I_{W(k)}\\) denote the \\(k\\)th of \\(N\\) periods of current
measurements. Then \\(I_{P(k)} = \\text{rms}(I_{W(k)})\\) and
\\[C_t = \\frac{1}{f_0} \\cdot \\frac{\\sum_{k=1}^N I_{P(k)} \\cdot k}
{\\sum_{k=1}^N I_{P(k)}} \\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
mains_frequency (int): Mains frequency, defaults to power frequency.
Returns:
numpy.ndarray: Temporal centroid as a (n_samples, 1)-dimensional array.
"""
# Calculate RMS of each period
# TODO: Use apply_to_periods?
ip = rms(current.reshape(current.shape[0], -1, period_length), axis=2)
# Calculate numerator and denominator and put together
numerator = np.sum(ip * np.arange(1, ip.shape[1] + 1), axis=1)
denominator = np.sum(ip, axis=1)
return mains_frequency * (numerator / denominator).reshape(-1, 1)
def crest_factor(current):
"""Calculates Crest factor (CF).
\\[CF = \\frac{\\max(|I|)}{\\text{rms}(I)}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: Crest factor as a (n_samples, 1)-dimensional array.
"""
return np.max(np.abs(current), axis=1).reshape(-1, 1) / rms(current)
def form_factor(current):
"""Calculates Form factor (FF).
\\[FF = \\frac{\\text{rms}(I)}{\\text{mean}(|I|)}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: Form factor as a (n_samples, 1)-dimensional array.
"""
return rms(current) / np.mean(np.abs(current), axis=1).reshape(-1, 1)
def transient_steady_states_ratio(current,
n_periods=5,
period_length=PERIOD_LENGTH):
"""Calculates the Transient steady states ratio (TSSR).
Let \\(I_{W(k)}\\) denote the \\(k\\)th of \\(N\\) periods of current
measurements and \\(\\langle t_1, t_2, ..., t_n \\rangle\\) the
concatenation of tuples \\(t_1, ..., t_n\\). Then:
\\[TSSR = \\frac{\\text{rms}(\\langle I_{W(1)}, I_{W(2)}, ...,
I_{W(\\text{n_periods})}\\rangle)}
{\\text{rms}(\\langle I_{W(N - \\text{n_periods})},
I_{W(N - \\text{n_periods} + 1)}, ..., I_{W(N)}\\rangle)}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
n_periods (int): Length of transient and steady state in periods
Returns:
numpy.ndarray: Transient steady states ratio as a (n_samples, 1)-dimensional array.
"""
return (rms(current[:, :n_periods * period_length]) /
rms(current[:, -n_periods * period_length:])).reshape(-1, 1)
def reactive_power(voltage,
current,
phase_shift=None,
period_length=PERIOD_LENGTH):
"""Calculates Reactive power (Q).
If phase shift is None, it is calculated.
\\[Q = \\text{rms}(V) \\times \\text{rms}(I) \\times \\sin(\\phi)\\]
Args:
voltage (numpy.ndarray): (n_samples, window_size)-dimensional array of voltage measurements.
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
phase_shift (numpy.ndarray): (n_samples, 1)-dimensional array of phase shifts.
Returns:
numpy.ndarray: Reactive power as a (n_samples, 1)-dimensional array.
"""
if phase_shift is None:
phase_shift = phase_shift(voltage,
current,
period_length=period_length)
return np.abs(rms(voltage) * rms(current) * np.sin(phase_shift))
def max_inrush_ratio(current, period_length=PERIOD_LENGTH):
"""Calculates the Max inrush ratio (MIR).
Let \\(I_{W(k)}\\) be the current measurements of the \\(k\\)th period and
\\(I_{P(k)} = \\text{rms}(I_{W(k)})\\). Then
\\[MIR = \\frac{I_{P(1)}}{\\max(|I_{W(1)}|)}\\]
Args:
current (numpy.ndarray): (n_samples, window_size)-dimensional array of current measurements.
Returns:
numpy.ndarray: Max inrush ratio as a (n_samples, 1)-dimensional array.
"""
first_period_rms = rms(current[:, :period_length])
first_period_max = np.max(np.abs(current[:, :period_length]), axis=1)
return first_period_rms / first_period_max.reshape(-1, 1)
def total_harmonic_distortion(harmonics_amp, spectrum_amp):
"""Calculates the Total harmonic distortion (THD).
Let \\(x_{f_1}, ..., x_{f_{20}}\\) be the amplitudes of the first 20
harmonics of the current and \\(x_{f_0}\\) the mains frequency amplitude.
Then:
\\[THD = \\frac{\\text{rms}([x_{f_1}, x_{f_2}, x_{f_3}, ..., x_{f_{20}}])}
{x_{f_0}}\\]
Args:
harmonics_amp (numpy.ndarray): Harmonic amplitudes as a (n_samples, n)-dimensional array.
spectrum_amp (numpy.ndarray): Spectral amplitudes as a (n_samples, window)-dimensional array.
Returns:
numpy.ndarray: Total harmonic distortion as a (n_samples, 1)-dimensional array.
"""
return (rms(harmonics_amp) / _mains_frequency_amplitude(spectrum_amp))
