Title:	Simulation-based UAV SAR Mission Optimization
DNr:	SNIC 2019/3-372
Project Type:	SNIC Medium Compute
Principal Investigator:	Ingo Staack <ingo.staack@liu.se>
Affiliation:	Linköpings universitet
Duration:	2019-07-01 – 2019-11-01
Classification:	20302 20399
Keywords:

Abstract

Electric unmanned aerial vehicle (UAV) design is usually conducted without specific sensor and application cases in mind, in comparison to conventional transportation aircraft design, which mainly focus on payload and range requirements. For an optimized design of an UAV for search & rescue (SAR) mission however, the vehicle design work should be conducted together with the sensor design/choice, the mission requirements and the operational (behavioural) model. The solution is thus a trade-off study including four domains, each containing a vast number of design parameter to be optimized. For the simulations, a fast cyber-physical simulation (CPS) tool, Hopsan is used. This tool is an in-house simulation tool developed at LiU/Flumes for fast simulation models. Previously, Flumes researches have successfully utilized NSC resources for the Hopsan-simulation-based design optimization of a complex fluid power product. Experience with available optimization algorithms, the optimization framework and experimentation approaches from this are now to be transferred onto the aircraft domain. By exploring variable mission requirements, the study at hand aims to extend the existing general optimization framework conceptually. The optimisation problem contains four domains: aircraft system, sensor technologies, environment and behavioural mission model. Especially the latter is seldomly included within a design optimization. By including the operation scenario within the design space, a better design solution may be found. Sensor and aircraft co-design is important with respect to the large design influence and design choices imposed by the sensor selection: weight, drag increase and power consumption. Especially for active sensors, the sensor energy effect and flight altitude dependent losses may largely influence both the aircraft and the operational properties. A parametric UAV model has been created with a robust normalization of the parameters to ensure a robust and large valid design space for parameter sweeps. The sensor model is initially limited to few parameters to investigate the principal sensor-aircraft co-design relations, to be refined later to represent real-world sensors available. The behavioural model start as a fixed pattern model, later to include the aircraft dynamics (and related energy losses). An agent-based model reaches even higher model fidelity and enable realistic in-mission adaptions to the environment. Due to the high complexity and the high interdependence of the overall system, the optimization problem has probably to be staggered into different sub-modules to limit the overwhelmingly large overall design space. Scientifically, two different results are expected: First, results gained will yield a more solid scientific foundation for the comparison of optimization algorithm performance aspects for staggered, multi-domain system of systems problems. Second, the workflow and results can be applied to develop future SAR and fire-fighting systems in Sweden. The latter topic is highly backed by the Swedish government and the MSB to protect the society against threats as a result of the global warming. The developed optimization strategy with its multi-domain models may latter also being applied on electric civil transportation aircraft which might be a future scenario towards more environmentally friendly aviation industry.