Optimizing for Cost and Performance: Trade Space Exploration of System-of-Systems Architectures
Overview
Complex tasks and missions often involve numerous parameters and design options, resulting in large permutations of candidate solutions when evaluating cost and performance. The nuanced interactions among constituent systems can lead to disproportionate outcomes from seemingly minor changes, making the system difficult to characterize for formal optimization. MATLAB and Simulink provide the ability to frame such studies more easily as traditional optimization problems—or, when appropriate, as machine learning challenges. This analytical approach enables engineers, integrators, and mission analysts to more effectively evaluate and navigate large trade spaces of candidate solutions.
In this talk, MathWorks will demonstrate exploration of large trade spaces using multi-object optimization techniques. These techniques will identify optimal architectures for a representative mission based on cost and performance objectives. Additionally, the MathWorks team will highlight how model-based systems engineering (MBSE) methodologies and model-based design can be used to perform dynamic mission simulations.
Using a search and rescue mission as a case study, we will highlight several key areas where MBSE complements mission engineering analysis. We will illustrate how to model a satellite communications system at multiple levels of fidelity with architecture variants. A trade space analysis will be performed by assembling a dynamic simulation of platforms executing a defined mission. We will also show how to import pre-existing platform models to improve confidence in simulation accuracy and analysis results. End-to-end mission modeling will be further enabled through scenario visualization in Unreal, orchestrated by executable architecture models.
Highlights
- Leverage mission engineering and simulation for requirements development
- Explore large trade spaces through multi-objective optimization
- Orchestrate mission simulations through activity diagrams
- Model RF systems at multiple levels of fidelity
- Visualize scenarios with Unreal and Cesium
About the Presenter
Russell Graves is an Application Engineer at MathWorks. He holds a PhD in Mechanical Engineering from The University of Tennessee. His expertise is in multi-agent machine learning, complex systems, and 3d visualizations. He has worked for 4 years prior to joining MathWorks applying these to transportation and autonomous vehicles in a research capacity.
Sam Lehman is an Application Engineer at MathWorks. Sam graduated from the Georgia Institute of Technology with a BS in Physics and an MS in Electrical Engineering with a concentration in Digital Signal Processing. Prior to MathWorks, Sam worked as a software engineer developing signals intelligence and geo-location systems for the Boeing Corporation.
Steve Ajemian is a Technical Account Manager for the MathWorks supporting Aerospace and Defense customers in the United States. Prior to joining MathWorks in 2020, Steve has supported the DoD for over 18 years, most recently for the MITRE Corporation. At MITRE, he provided subject-matter expertise in wireless communications and networking to DoD customers and lead projects supporting technology acquisition of tactical datalinks for the US Navy. Steve holds a BS and MS in EE from the Johns Hopkins University and an MS in Engineering Management from MIT.
Product Focus
This event is part of a series of related topics. View the full list of events in this series.
