Design and optimisation of an uncrewed aerial system service framework

Abstract

Creating an effective and competitive uncrewed aerial system (UAS) service requires a large array of decisions to be made. The solution has to encompass not only the choice of ucrewed aerial vehicle (UAV), but also the concept of operations, the location of the operating base(s), the personnel required to fly and maintain the platforms and the effect of the temporal weather to name a few. Currently, the decisions are made based on little evidence and previous knowledge of successes and failures. Here, the creation and use of a simulation tool to aid the decision making process of designing a UAS service is investigated. This thesis introduces a mission-based simulation tool that utilises discrete-event simulation techniques to replicate a real-world UAS service proposal. For the given service the tool models the UASs in terms of performance and reliability, and places them at operating bases with the required personnel. Missions and weather variables are dynamically generated from predefined probability-distribution functions set out in the service proposal. The simulation ultimately produces a score that signifies the effectiveness of the service design along with the cost. With these outputs and the data behind them, a design-value is produced from a value function. By running the simulation with different design candidates consisting of different combinations and numbers of UAV types at different operation bases it is possible to find an optimal service design. This is demonstrated in a case study applied to the simulation tool. The general lessons learnt while developing the computational tool are discussed and the model’s scalability and applicability was explored. The presented tool is capable of modelling a multitude of UAS service types and mission profiles. The framework is based on a modular, comprehensible, generic and realistic approach which benefits the applicability of the tool and the ability to update components. To perform large optimisation studies the addition of a combinatorial problem solver is recommended and discussed as future work.

More information

Identifiers

ORCID for Robert William Entwistle: orcid.org/0000-0002-1453-2820

Catalogue record

Export record