A set of artificial neural network (ANN) models are developed ...

haremboingAI and Robotics

Oct 20, 2013 (3 years and 9 months ago)

94 views

A set of artificial neural network (ANN) models are developed following Baysian learning
approach for the estimation of several key process characteristics as function of welding
conditions in friction stir welding (FSW). The training and the testing data
to develop the ANN
models are generated using a well tested three dimensions numerical heat transfer and visco
-
plastic material flow model. The focus of the work is to provide a methodology and an openly
available tool that can help in successful use of FS
W without premature tool failure. This page
contains a link to a zipped file that has four directories: (i) final_input_data_files, (ii)
final_map_data_files, (iii) final_nn_code_for_upload, (iv) final_tif_map_figures. The directory
"input_files" contains
the input files that are used to develop the optimized neural nets. The
directory "results_map" includes the text files which are generated using the optimized models.
These files are used to construct maps of safe working zone for a given combination of w
elding
conditions. The directory "code_manual" has the aritifical neural network source codes and
manual. The directory "process_maps" includes several examples of process maps which are
constructed using the corresponding optimized models. Please download

the files in your local
computer first and then try to open them. In the case, you have any difficulty to use or realize
any information published in the link mentioned above, contact
amit@iitb.ac.in

or
debroy@psu.edu
.