Olaf Milbredt

Optimizing intermodal traffic management involving airports and railway using forecast and what-if capability

Forecasting and what-if capabilities are common to modern airport management systems. Using such a system often requires the operator to have a clear idea of how the optimal solution will look like. In this paper we consider the case of delayed trains to the airport. The question is whether and to what extend to delay flights to wait for passengers who took the delayed trains. We present an approach for assessing possible actions based on gain and loss. The definitions of gain and loss are based on the specific flight carta of affected flights and costs for delaying a flight. The methodology is based on assessing a large number of what-if runs. Costs are included as parameters and can be customized by operators.

Model-based airport security analysis in case of blackouts or cyber-attacks

With the increasing exposure of critical infrastructures to external factors, critical infrastructure protection gained attention within security research. To develop protection methods the effect of external factors on critical infrastructures has to be known. In this paper we present an event-based, dynamic modeling approach using power flow analysis and network flow analysis to simulate this effect on the critical airport infrastructure airport with object-based programming. Network component objectification is used to treat the highly interconnected cyber-physical system as a two-layered system with separated cyber and physical properties. Network stability and cyber infection status is checked with two control feedback loops checking grid stability and infection status. Therefore with given object load powers and network interconnections the working capability of infrastructure parts can be predicted for various scenarios and used as input for passenger flow simulations or economical prognosis.

Parameter weighting for multi-dimensional fuzzy inference systems

Exponential growth of the number of rules of a multi-dimensional Fuzzy Inference System with growing number of input parameters makes it impractical for humans to populate the description of such a system via the Fuzzy Control Language by hand. In this paper we present a methodology for automatically generating rules. The generation process is influenced by a given weighting of the input parameters. The weighting is derived from a parameter weighting matrix used in the context of Analytic Hierarchy Process. The resulting Fuzzy Inference Systems are intended to be the starting point of an Adaptive Neuro-Fuzzy Inference System. As application of the methodology we present the security management at an airport. Here, Fuzzy Logic is used to evaluate the overall security situation.

A Neural Network Approach to Airport Management

An adaptive method for Performance Based Airport Management using a Neural Network for evaluation of the change of KPIs and appropriate action is presented. It assists the operators by evaluating past, current, and expected values of the KPIs and identifying the most promising reaction, if it is required. A first implementation of this concept covers the control of border control checkpoints which are crucial for airports serving as a hub. Based on the KPI “boarding quota“ a Neural Network is designed making suggestions about the number of open checkpoints, if there exist delays of the arriving flights. It is trained by a posteriori optimized configurations using a Genetic Algorithm.