Input train_FD001, output a list

Generator function for creating random batches of training-data for many to many models

1D convolutional layer with or without dropout or batch normalization :param batch_norm: bool, enable batch normalization :param is

Helper function to create a fully connected layer with or without batch normalization or dropout regularization :param x: previous layer

Enter the data of an engine_id of train_FD001 and output the corresponding RUL (remaining life) of these data. type is list

:param save: switch to load the already preprocessed data or begin preprocessing of raw data :param save_training_data: same functionality

Function is to load PHM 2008 challenge dataset

:param error_arr: a list of errors for each training trajectory :return: standered score value for RUL

The architecture is a Meny-to-meny model combining CNN and LSTM models :param dataset: select the specific dataset between PHM08 or CMAPSS

Generate pre-processed data according to the given dataset :param dataset: choose between "phm" for PHM 2008 dataset or "cmapss" for CMAPSS

This helper function only augments the training data to look like testing data. Training data always run to a failure. But testing data is

Helper function to get a different types of RNN cells with or without dropout wrapper :param cell_types: cell_type can be 'GRU' or 'LSTM' o

Piecewise linear function with zero gradient and unit gradient ^ | MAXLIFE |----------- |

Extract training trjectories one by one test_engine_id = [11111111...,22222222....,...]

Plot every plot on top of each other