UoM administered thesis: Phd

  • Authors:
  • Mengying Zhang


With the rapid development of modern aviation industry, how to make a balance between gradually increased aviation activities and the life quality of residents living within the vicinities of the airports becomes an increasingly prominent challenge. Research efforts have been made on single or multiple objective trajectory optimisations using gradient-based methods. Despite the effectiveness of the gradient-based methods, their applicability is limited by high standard requirements for problem formulation with smooth differentiable dynamics models and explicit expressions of optimisation objectives. An alternative option is using gradient-free algorithms capable of solving optimisation problems with discrete models, facilitating the integration of ''black box'' models that lack gradient information. However, the goals are generally achieved at the price of extensive computational burden. Moreover, adding further environmental parameters or different noise attributes, an optimal solution cannot be selected from the solution set obtained in the multi-objective optimisation problems, unless further algorithms are introduced. The present work focuses on the development of a multi-objective optimisation framework for departure and arrival aircraft to minimise multiple environmental impacts (noise and exhaust emissions). This framework includes a user-defined input module, a optimisation module, and flight perforemance module. The trajectory optimistaion module includes a set of of nonlinear models: aircraft dynamics, trajectory constraints and objective functions. A method to parameterise movement in the lateral plane based on a B\'{e}zier curve has been proposed to decrease the number of free parameters. The environmental objectives are modeled by a comprehensive flight mechanics programe FLIGHT and the ANP database from EUROCONTROL. Two posterior selection strategies based on a preference function and monetisation approaches are used to evaluate the resulting Pareto solution set. A simple single flight trajectory optimisation problem is identified and formulated as a multi-objective optimal control problem with a discontinuous problem formulation, solved by non-gradient algorithms. Among the three different non-gradient algorithms used, a non-dominated sorting genetic algorithm is identified as the most widely used method. However, two PSO-based multi-objective optimisers are explored to overcome some of the drawbacks of the GAs technique. Complex multiple flight events trajecotry optimisation problem is identified and formulated as a mixed-integer non-linear programming problem. A time-based separation rule is applied to approach the real flight assignment scenario. Finally, this application is conducted with a simple scenario to demonstrate its functionality. We demonstrate that this simulation framework is capable of solving trajectory optimisation problems with multiple simultaneous environmental objectives.


Original languageEnglish
Awarding Institution
Award date1 Aug 2019