OPTIMAL SEQUENCING IN ATM COMBINING GENETIC ALGORITHMS AND GRADIENT BASED METHODS TO A BILEVEL APPROACH

摘要

A bi-level algorithm is developed to determinethe optimal arrival sequence w.r.t a specificupper level objective function, e.g. minimiz-ing pollutant emissions or maximizing runwaythroughput. Within a sequence it is assumedthat all aircraft are operated optimally (lowerlevel optimization), minimizing individual fuelconsumption. However, all aircraft trajecto-ries adhere to constraints imposed by the up-per level, such as time and distance separation.The algorithm combines a direct optimal con-trol method for solving the lower level problemwith a genetic algorithm, which is applied tothe upper level combinatorial problem.The lower level problems are fully dis-cretized by applying a trapezoidal collocationscheme provided by FALCON.m [1] and subse-quently solved utilizing the interior point NLPsolver IPOPT [2]. The discretization is per-formed during initialization of the algorithmand reused during execution to minimize com-putational effort. For the upper level prob-lem an efficient mutation operator is intro-duced, which exploits the information con-tained within the Lagrange multiplier of thediscretized arrival time constraint.The algorithm is validated against a testcase scenario comprising five aircraft destinedfor runway 08L of Munich airport, which en-ter the terminal maneuvering area within fourminutes at multiple way points. Aircraft dy-namics are represented by a model derivedfrom BADA 3. The performance of the algo-rithm allows the application on a pre-tacticallevel for a limited number of arriving aircraft.