示例#1
0
def main():
    data_dir_path = 'data'
    model_dir_path = 'model'
    dateparser = lambda x: pd.datetime.strptime(x, '%Y-%m-%d %H:%M:%S')
    satellite_data1 = pd.read_csv(data_dir_path + '/data_std.csv',
                                  sep=',',
                                  index_col=0,
                                  encoding='utf-8',
                                  parse_dates=True,
                                  date_parser=dateparser)
    column = [
        'INA1_PCU输出母线电流', 'INA4_A电池组充电电流', 'INA2_A电池组放电电流', 'TNZ1PCU分流模块温度1',
        'INZ6_-Y太阳电池阵电流', 'VNA2_A蓄电池整组电压', 'VNC1_蓄电池A单体1电压', 'VNZ2MEA电压(S3R)',
        'VNZ4A组蓄电池BEA信号'
    ]

    # column = ['INA1_PCU输出母线电流']#,'INA4_A电池组充电电流','INA2_A电池组放电电流','TNZ1PCU分流模块温度1','INZ6_-Y太阳电池阵电流','VNA2_A蓄电池整组电压','VNC1_蓄电池A单体1电压','VNZ2MEA电压(S3R)','VNZ4A组蓄电池BEA信号']
    satellite_data = satellite_data1.loc[:, column].iloc[0:80]  #96700
    print(satellite_data.head())
    satellite_np_data = satellite_data.as_matrix()
    scaler = MinMaxScaler()
    satellite_np_data = scaler.fit_transform(satellite_np_data)
    print(satellite_np_data.shape)
    index = satellite_data.index
    columns = satellite_data.columns
    time_window_size = 8
    # data_std = pd.DataFrame(satellite_np_data, index=index, columns=columns)
    # data_std.to_csv('data/data_scaler.csv', encoding='utf-8')
    input_dataset = np.reshape(
        satellite_np_data,
        ((int)(satellite_np_data.shape[0] / time_window_size),
         time_window_size, satellite_np_data.shape[1]))
    ae = LstmAutoEncoder6(index, columns)

    # fit the data and save model into model_dir_path
    if DO_TRAINING:
        ae.fit(input_dataset,
               batch_size=10,
               model_dir_path=model_dir_path,
               time_window_size=time_window_size,
               estimated_negative_sample_ratio=0.9)

    # load back the model saved in model_dir_path detect anomaly
    ae.load_model(model_dir_path)
    anomaly_information = ae.anomaly(input_dataset)
    reconstruction_error = []
    for idx, (is_anomaly, dist) in enumerate(anomaly_information):
        print('# ' + str(idx) + ' is ' +
              ('abnormal' if is_anomaly else 'normal') + ' (dist: ' +
              str(dist) + ')')
        reconstruction_error.append(dist)

    visualize_reconstruction_error(reconstruction_error, ae.threshold)
def main():
    data_dir_path = 'data'
    model_dir_path = 'model'
    dateparser = lambda x: pd.datetime.strptime(x, '%Y-%m-%d %H:%M:%S')
    satellite_data1 = pd.read_csv(data_dir_path + '/data_std.csv',
                                  sep=',',
                                  index_col=0,
                                  encoding='utf-8',
                                  parse_dates=True,
                                  date_parser=dateparser)
    satellite_data = satellite_data1.iloc[0:96700]
    print(satellite_data.head())
    satellite_np_data = satellite_data.as_matrix()
    scaler = MinMaxScaler()
    satellite_np_data = scaler.fit_transform(satellite_np_data)
    print(satellite_np_data.shape)
    index = satellite_data.index
    columns = satellite_data.columns
    time_window_size = 1
    # data_std = pd.DataFrame(satellite_np_data, index=index, columns=columns)
    # data_std.to_csv('data/data_scaler.csv', encoding='utf-8')

    ae = LstmAutoEncoder5(index, columns)

    # fit the data and save model into model_dir_path
    if DO_TRAINING:
        ae.fit(satellite_np_data,
               model_dir_path=model_dir_path,
               time_window_size=time_window_size,
               estimated_negative_sample_ratio=0.9)

    # load back the model saved in model_dir_path detect anomaly
    ae.load_model(model_dir_path)
    anomaly_information = ae.anomaly(satellite_np_data[:96700, :])
    reconstruction_error = []
    for idx, (is_anomaly, dist) in enumerate(anomaly_information):
        print('# ' + str(idx) + ' is ' +
              ('abnormal' if is_anomaly else 'normal') + ' (dist: ' +
              str(dist) + ')')
        reconstruction_error.append(dist)

    visualize_reconstruction_error(reconstruction_error, ae.threshold)