Digital Thread Analysis and Visualization Tool (DT-AVT)
This repository contains the analysis and visualization tool used in the following articles and thesis:
- Singh V, Willcox K (2021) Decision Making Under Uncertainty for a Digital Thread Enabled Design Process, J. Mech. Des. 143 (9):091707, DOI 10.1115/1.4050108
- Singh V (2019) Towards a Feedback Design Process Using Digital Thread. PhD Thesis, MIT
- Singh V, Willcox K (2018) Engineering Design with Digital Thread. AIAA Journal 56 (11):4515–4528, DOI 10.2514/1.J057255
Digital Threads, Digital Twins, and the design process at various stages can be analyzed and visualized at the component level within this tool. Different policies can be compared (including designs) and their performances can be quantified. The tool is written in the context of the example design problem presented in the above works, though it can be modified to suite other design scenarios. All code is natively written in MATLAB R2018b on a Windows 10 64-bit machine.
A demo ribweb has been setup for running and viewing.
- Launch
visualizer.m
. From there, right-click in the right pane and select load. From the loading screen, loadDemos/RibWebSeed.mat
. - Next, initialize policies by right-clicking in the right pane and selecting
Policies > Initialize
. - Once all policies are initialized, evaluate the policies by right-clicking in the right pane and selecting
Policies > Evaluate
. - After evaluation of the policies, you can navigate Digital Threads and Digital Twins for each policy using the search tree on the left. Additional viewing options for different tree levels can be accessed by right-clicking on figure axes that pop up.
- To optimize policies, right-click in the right pane and select
Policies > Optimize
. To view the results of the optimization, re-evaluate the policies by selecting againPolicies > Evaluate
. Currently, optimization is done for one iteration across all stages and policies. This is controlled in the functionsbellmanBackup.m\bellmanBackup()
(for iterations) andvisualizer.m\optimizePolicies()
(for which policies to be run). For good results, it is recommended to run for at least 100 iterations (~approx 10-15 hrs on a Windows 10 64-bit laptop). The results in the works above have been run for over 2000 iterations.
If you need more detailed instructions, assistance for navigation (including all features of the tool not described in Quick Start or Documentation), or setting up and running your own design problem, please email:
victorsingh at alum dot mit dot edu