In [1]:

    

from neuralnilm import RealApplianceSource
import matplotlib.pyplot as plt
import numpy as np


    

    




/home/jack/workspace/python/nntools/lasagne/init.py:86: UserWarning: The uniform initializer no longer uses Glorot et al.'s approach to determine the bounds, but defaults to the range (-0.01, 0.01) instead. Please use the new GlorotUniform initializer to get the old behavior. GlorotUniform is now the default for all layers.
  warnings.warn("The uniform initializer no longer uses Glorot et al.'s "

In [2]:

    

source = RealApplianceSource(
    filename='/data/mine/vadeec/merged/ukdale.h5',
    appliances=[
        ['fridge freezer', 'fridge', 'freezer'],
        ['washer dryer', 'washing machine'],
        'hair straighteners',
        'television',
        'dish washer'
    ],
    max_appliance_powers=[300, 2400, 500, 200, 2500],
    on_power_thresholds=[5] * 5,
    min_on_durations=[60, 1800, 60, 60, 1800],
    min_off_durations=[12, 600, 12, 12, 1800],
    window=("2013-03-18", "2013-04-18"),
    seq_length=512,
    output_one_appliance=False,
    train_buildings=[1],
    validation_buildings=[1],
    n_seq_per_batch=8,
    standardise_input=True,
    independently_center_inputs=False,
    skip_probability=0.75,
    skip_probability_for_first_appliance=0.5,
    target_is_start_and_end_and_mean=True
)


    

    




Loading training activations...






    




INFO:neuralnilm.source:Loading training activations...






    




  Loading activations for fridge freezer from building 1...






    




INFO:neuralnilm.source:  Loading activations for fridge freezer from building 1...






    




Loaded 760 activations.






    




INFO:neuralnilm.source:Loaded 760 activations.






    




  Loading activations for washer dryer from building 1...






    




INFO:neuralnilm.source:  Loading activations for washer dryer from building 1...






    




Loaded 22 activations.






    




INFO:neuralnilm.source:Loaded 22 activations.






    




  Loading activations for hair straighteners from building 1...






    




INFO:neuralnilm.source:  Loading activations for hair straighteners from building 1...






    




Loaded 16 activations.






    




INFO:neuralnilm.source:Loaded 16 activations.






    




  Loading activations for television from building 1...






    




INFO:neuralnilm.source:  Loading activations for television from building 1...






    




Loaded 52 activations.






    




INFO:neuralnilm.source:Loaded 52 activations.






    




  Loading activations for dish washer from building 1...






    




INFO:neuralnilm.source:  Loading activations for dish washer from building 1...






    




Loaded 11 activations.






    




INFO:neuralnilm.source:Loaded 11 activations.






    




Done loading activations.






    




INFO:neuralnilm.source:Done loading activations.

In [3]:

    

X, y = source._gen_data()
X, y = source._process_data(X, y)


    

In [9]:

    

y[0]


    

    
Out[9]:





array([[ 0.47070312,  0.        ,  0.        ,  0.0078125 ,  0.        ],
       [ 0.66796875,  0.        ,  0.        ,  0.99804688,  0.        ],
       [ 0.28838283,  0.        ,  0.        ,  0.56630176,  0.        ]])

In [13]:

    

len(source.train_activations['fridge freezer'])


    

    
Out[13]:





823

In [ ]:

    




    

In [14]:

    

plt.plot(X[3,:,:]); plt.show()


    

In [12]:

    

source.target_stats


    

    
Out[12]:





{'mean': array([ 0.], dtype=float32), 'std': array([ 0.], dtype=float32)}

In [3]:

    

source.start()


    

In [5]:

    

X.shape


    

    
Out[5]:





(8, 64, 1)

In [7]:

    

#source.standardise_input = True
X, y = source.queue.get()


    

In [16]:

    

#source.input_stats = {'std': np.array([0.01], dtype=np.float32)}
X, y = source._gen_data()
X, y = source._process_data(X, y)


    

In [17]:

    

SEQ_I = 7

plt.plot(X[SEQ_I,:,:])
plt.show()


    

In [6]:

    

X[SEQ_I,:,:].mean()


    

    
Out[6]:





0.20007956572024463

In [7]:

    

X[SEQ_I,:,:].std()


    

    
Out[7]:





1.2653309337904646

In [8]:

    

source.input_stats


    

    
Out[8]:





{'mean': array([ 0.04142749], dtype=float32),
 'std': array([ 0.10218309], dtype=float32)}

In [33]:

    

plt.plot(y[0,:,:])


    

    
Out[33]:





[<matplotlib.lines.Line2D at 0x7fbf612135d0>]

In [9]:

    

lines = plt.plot(y[4,:,:])
plt.legend([label for label in source.get_labels()], framealpha=.5)
plt.show()


    

In [8]:

    

y[0]


    

    
Out[8]:





array([[ 0.        ,  0.        ,  2.14326191,  0.1450543 ,  0.48895144]])

In [28]:

    

source.target_stats


    

    
Out[28]:





{'mean': array([ 0.04066789,  0.01881946,  0.24639061,  0.17608672,  0.10273963], dtype=float32),
 'std': array([ 0.11449792,  0.07338708,  0.26608968,  0.33463112,  0.21250485], dtype=float32)}

In [9]:

    

y.min()


    

    
Out[9]:





0.0

In [8]:

    

import numpy as np
fig, ax = plt.subplots()
ax.plot(np.sum(y[2,:,:], axis=1), label='appliances')
ax.plot(X[2,:,:], label='mains')
plt.legend()
plt.show()


    

In [14]:

    

y.mean(axis=0).mean(axis=0)


    

    
Out[14]:





array([ -2.48352684e-12,   1.46666716e-05,   5.46665751e-06,
         4.00013960e-07,   6.26781002e-07], dtype=float32)

In [7]:

    

plt.plot(y.reshape(10*1500, 5))
plt.show()


    

    




---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-7-52a6256b067f> in <module>()
----> 1 plt.plot(y.reshape(10*1500, 5))
      2 plt.show()

ValueError: total size of new array must be unchanged

In [15]:

    

y.reshape(10*1500, 5).std(axis=0)


    

    
Out[15]:





array([ 0.0084313 ,  0.00956285,  0.00271722,  0.03271817,  0.07027333], dtype=float32)

In [7]:

    

X.reshape(10*1500, 1).std(axis=0)[0]


    

    
Out[7]:





0.13455896

In [49]:

    

import numpy as np
any(np.isnan(X.flatten()))


    

    
Out[49]:





False

In [50]:

    

any(np.isnan(y.flatten()))


    

    
Out[50]:





False

In [1]:

    

from neuralnilm.net import SubsampleLayer


    

    




Couldn't import dot_parser, loading of dot files will not be possible.

In [3]:

    

sub = SubsampleLayer(None, 5)


    

In [7]:

    

sub.input_shape = X.shape


    

In [9]:

    

out = sub.get_output_for(X)


    

In [10]:

    

out.shape


    

    
Out[10]:





(5, 200, 1)

In [15]:

    

plt.plot(out[0,:,:])
plt.show()


    

In [11]:

    

from nilmtk import DataSet
ds = DataSet('/data/mine/vadeec/merged/ukdale.h5')


    

In [12]:

    

elec = ds.buildings[1].elec
meter = elec['washer dryer']


    

In [13]:

    

activations = meter.activation_series()


    

In [18]:

    

activations[3].plot()
plt.show()


    

In [27]:

    

try:
    a = b
except:
    print(1)
    raise
finally:
    print(2)


    

    




1
2






    




---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-27-a5790709ee0a> in <module>()
      1 try:
----> 2     a = b
      3 except:
      4     print(1)
      5     raise

NameError: name 'b' is not defined

In [8]:

    

y.shape


    

    
Out[8]:





(10, 300, 5)

In [15]:

    

b = y > 0


    

In [17]:

    

inactive = b.sum(axis=1) == 0


    

In [18]:

    

inactive.shape


    

    
Out[18]:





(10, 5)

In [20]:

    

inactive


    

    
Out[20]:





array([[False,  True, False,  True,  True],
       [False,  True,  True, False, False],
       [False,  True,  True,  True,  True],
       [False,  True,  True,  True,  True],
       [False,  True, False,  True, False],
       [ True,  True,  True, False,  True],
       [ True,  True,  True,  True, False],
       [False,  True,  True,  True,  True],
       [False,  True, False,  True, False],
       [False,  True,  True,  True, False]], dtype=bool)

In [35]:

    




    

    
Out[35]:





(1, 3000, 5)

In [22]:

    

inactive[0]


    

    
Out[22]:





array([False,  True, False,  True,  True], dtype=bool)

In [46]:

    

np.swapaxes(y,2,1)[inactive].shape


    

    
Out[46]:





(32, 300)

In [48]:

    

import theano
import theano.tensor as T


    

In [50]:

    

yt = theano.shared(y, name='y')


    

In [51]:

    

bt = yt > 0


    

In [96]:

    

inactivet = bt.sum(axis=1) <= 0


    

In [97]:

    

inactivet.eval()


    

    
Out[97]:





array([[0, 1, 0, 1, 1],
       [0, 1, 1, 0, 0],
       [0, 1, 1, 1, 1],
       [0, 1, 1, 1, 1],
       [0, 1, 0, 1, 0],
       [1, 1, 1, 0, 1],
       [1, 1, 1, 1, 0],
       [0, 1, 1, 1, 1],
       [0, 1, 0, 1, 0],
       [0, 1, 1, 1, 0]], dtype=int8)

In [98]:

    

yt.dimshuffle(0,2,1)[inactivet.nonzero()].shape.eval()


    

    
Out[98]:





array([ 32, 300])

In [99]:

    

data = yt.dimshuffle(0,2,1)[inactivet.nonzero()].eval()


    

In [101]:

    

plt.plot(data[0])


    

    
Out[101]:





[<matplotlib.lines.Line2D at 0x7f2ca82b3dd0>]

In [ ]: