def preprocess(self):
# self.data = pd.DataFrame(columns=['rpm', 'amp_ratio', 'rms', 'std', 'mean', 'mode', 'range', 'label'])
self.data_dict = {
'rpm': [],
'amp_ratio': [],
'rms': [],
'std': [],
'mean': [],
'mode': [],
'range': [],
'label': [],
}
for label in [0, 1]:
self.__db.query_dict = {'label': {'$eq': label}}
file_list = self.__db.query()
for f in file_list:
file_uri = os.path.join(f['a_uri'], f['filename'])
print(file_uri)
dp = preprocess.DataPreprocess(filename=file_uri)
data_pre = dp.origin_data().data_split()
for i, df in enumerate(data_pre):
df = dp.data_drop(df)
df = dp.data_drop(df, drop_type='tail', pct=0.2)
self.data_dict['rpm'].append(i)
signal0 = signalProcess.SignalProcess(df=df)
self.data_dict['amp_ratio'].append(signal0.f_amp_ratio())
self.data_dict['rms'].append(signal0.t_v_rms())
self.data_dict['std'].append(signal0.t_a_std())
self.data_dict['mean'].append(signal0.t_a_mean())
self.data_dict['mode'].append(signal0.t_a_mode())
self.data_dict['range'].append(signal0.t_a_range())
self.data_dict['label'].append(label)
self.data = pd.DataFrame().from_dict(self.data_dict)
def abnormal_zero_anasys():
figure_no = 0
db = mongodb_op.MongodbOp()
db._query_dict = {'label': {'$eq': 1}}
file_dict = db.query()
temp = []
for f_dict in file_dict:
figure_no += 1
plt.figure(figure_no)
file = os.path.join(f_dict.get('a_uri'), f_dict.get('filename'))
print(file)
dp = preprocess.DataPreprocess(filename=file)
dp.origin_data(filter_zero=False)
dp.df.loc[dp.df.all(axis=1), 'z'] = 1
f, p = signal.welch(dp.df.z, 4000, scaling='spectrum')
plt.plot(f, p)
return f, p
def origin_distribute():
db = mongodb_op.MongodbOp()
db._query_dict = {'label': {'$eq': 0}}
file_dict = db.query()
temp = []
for f_dict in file_dict:
file = os.path.join(f_dict.get('a_uri'), f_dict.get('filename'))
print(file)
dp = preprocess.DataPreprocess(filename=file)
dp.origin_data()
dp.df = dp.df.drop(['x', 'y'], axis=1)
df_rpm_1, _, _ = dp.data_split()
df_rpm_1 = dp.data_drop(df_rpm_1, )
df_rpm_1 = dp.data_drop(df_rpm_1, drop_type='tail')
df_rpm_1.columns = [file]
temp.append(df_rpm_1)
return pd.concat(temp, sort=False, axis=1)
def helper(file_dict, out_dir, label=1):
gc.collect()
file_uri = os.path.join(file_dict['a_uri'], file_dict['filename'])
print(file_uri)
dp = preprocess.DataPreprocess(filename=file_uri)
data_pre = dp.origin_data().data_split()
temp_lst = []
for i, df in enumerate(data_pre):
df = dp.data_drop(df)
df = dp.data_drop(df, drop_type='tail', pct=0.2)
dp.df = df
df_sub = dp.data_split(n=20)
for df_temp in df_sub:
# df_temp = df.sample(frac=0.4)
signal0 = signalProcess.SignalProcess(df=df_temp)
print('{} signal process'.format(file_uri))
data_dict = dict()
data_dict['rpm'] = [i]
data_dict['amp_ratio'] = [signal0.f_amp_ratio()]
data_dict['rms'] = [signal0.t_v_rms()]
data_dict['std'] = [signal0.t_a_std()]
data_dict['mean'] = [signal0.t_a_mean()]
data_dict['mode'] = [signal0.t_a_mode()]
data_dict['range'] = [signal0.t_a_range()]
data_dict['label'] = label
temp_lst.append(pd.DataFrame().from_dict(data_dict))
del df_temp
# all_df.concat(temp, sort=False, ignore_index=True, copy=False)
# all_df.to_pickle('{}.pkl'.format(os.path.basename(file_uri)))
# del temp
del df_sub
del df
del data_pre
pd.concat(temp_lst, ignore_index=True).to_pickle(
os.path.join(out_dir, '{}.pkl'.format(file_dict['filename'])))
del temp_lst
print('--process finished')
return 1
def f_envelope_spectrum():
db = mongodb_op.MongodbOp()
fig_num = 0
for label in [0, 1]:
db.query_dict = {'label': {'$eq': label}}
file_dict = list(db.query())[:2]
for f_dict in file_dict:
file_uri = os.path.join(f_dict['a_uri'], f_dict['filename'])
print(file_uri)
dp = preprocess.DataPreprocess(filename=file_uri)
data_pre = dp.origin_data().data_split()
for i, df in enumerate(data_pre):
df = dp.data_drop(df)
df = dp.data_drop(df, drop_type='tail', pct=0.2)
dp.df = df
df_sub = dp.data_split(n=20)
for df_temp in df_sub:
fig_num += 1
plt.figure(fig_num)
signal0 = signalProcess.SignalProcess(df=df_temp)
f, p = signal0.f_envelope_spectrum()
plt.plot(f, p)
plt.savefig('fig/{}_{}.jpg'.format(label, fig_num))
del df_sub
def init_param(self, args):
self.sig = signalProcess.SignalProcess(df=pd.DataFrame())
self.dp = preprocess.DataPreprocess()
self.mdb = mongodb_op.MongodbOp()
self.segment = 'train'