PEN-Method: Predictor model and Validation Data

Published: 3 September 2021| Version 4 | DOI: 10.17632/459f33wxf6.4
Contributor:
Alex Halle

Description

This Data contains the PEN-Predictor-Keras-Model as well as the 100 validation data sets.

Files

Steps to reproduce

With the following steps, the prediction of data (using the predictor and the validation data sets as input) can be reproduced. 1. Extract the zip-files. 2. Run the python code. ```python # Using python 3.9 # Using tensorflow 2.5 import tensorflow as tf import numpy as np # loading the model model_predictor = tf.keras.models.load_model('model_predictor_64x64') # loading input validation datasets d_inp = 8 inputrows = np.load("Validation Data/inputrows_{0}x{0}.npz".format(d_inp))['arr_0'] R_k_val = inputrows[:, :2 * (d_inp + 1)**2] R_s_val = inputrows[:, 2 * (d_inp + 1)**2:4 * (d_inp + 1)**2] M_tar_val = inputrows[:, 4 * (d_inp + 1)**2] # loading output validation dataset (index 0) d = 64 index = 0 np_load = np.load("Validation Data/outputrows_{0}_{1}x{1}.npz".format(index, d)) X_TO_val = np.reshape(np_load['min_Material'], [d, d]) c_TO_val = np.asscalar(np_load['Nachgiebigkeit']) M_TO_val = np.asscalar(np_load['Materialabweichung']) F_TO_val = np.asscalar(np_load['Filter']) P_TO_val = np.asscalar(np_load['Unbestimmtheit']) t_TO_val = np.asscalar(np_load['time_fe']) # predicting outputs (8x8, 16x16, 32x32, 64x64) for all 100 validation input data sets X_pred_val = model_predictor.predict([R_k_val, R_s_val, M_tar_val]) ```

Institutions

Technische Universitat Chemnitz Fakultat fur Maschinenbau

Categories

Artificial Neural Networks, Data Validation

Licence