4. Application to aviation safety
4.3 Demonstrating an On-board Navigation Decision Support System
4.3.1 BADA Database Overview
BADA is a collection of text files that lays down operation performance parameters, airline procedure parameters and performance summary tables for more than 300 aircraft types [97]. It was developed and is maintained by the European Organization for the Safety of Air Navigation (EUROCONTROL). The information contained in these files was obtained using the mass-varying kinetic approach to aircraft modelling. It models an aircraft as a point along with the underlying forces acting upon it which causes its motion. Figure 31 [155] shows the structure of the BADA Aircraft Performance Model (APM).
Application to aviation safety 180
The model is organized into five sub-models, namely: characteristics, actions, motion, operation, and limitations. The arrows represent the dependencies between the sub-models. The actions sub-model is used to calculate various forces acting on the aircraft, whereas the motion sub-model accounts for geometrical, kinematics, and kinetic parts of motion using the Total Energy Model (TEM) method. The operations sub-model is used to simulate the different operation modes of an aircraft, such as flying with a constant Mach number. The limitation sub-model mimics the operational limit of the aircraft such as the maximum altitude, throttle limit, and maximum airspeed. Finally, the characteristics sub-model contains coefficients that characterize an aircraft such as the wing span [152]. Each modelled aircraft is parameterized into three text files. First, an Operations Procedure File (OPF) holds aerodynamic constants such as thrust, fuel, and drag coefficients.
Second, an Airlines Procedures file (APF) contains parametric information about the recommended speed procedures during different flying phases, and third, a Procedure Table File (PTF) represents the recommended operation procedures in the form of look-up tables [152]. This demonstration utilizes the
Limitations Aircraft Characteristics
Actions
Motion
Operations
Figure 31. Structure of BADA APM
Application to aviation safety 181
look-up table of a specific aircraft that is contained within a PTF file as it gives the user direct access to performance data without the need to implement the complete TEM [97], which in turn reduces the complexity of the developed network.
The PTF file contains the recommended operating producers for airspeed, rate of climb/descent (ROCD), and fuel flow at different flight levels of a specific aircraft. (An example of a PTF file can be found in [156].) The header section of the PTF file specifies general information about the type of the aircraft, creation date, speeds, temperature data, maximum altitude and mass levels. This is followed by the table of performance data, where the operation information is organized into three sections: cruise, climb and descent [152].
In this demonstration, a script was written to subtract the performance table of a PTF file of a specific aircraft and organize the data in a look-up table that is more suitable for analysis by MATLAB. In addition, the script verifies the validity of the PTF file by checking for the presence of some permanent text within the header section of the file.
4.3.2 Assumptions and Proposed Design
We propose a framework that facilitates the base of aircraft data (BADA) as a navigation planning decision support system for pilots to make informed decision about navigation planning. The decision support system is implemented as a software tool to extract performance data of an aircraft type from BADA database, integrate with other on-board fault detection and isolation systems, and estimate the beneficial value of these recommendations. The designed network presented in this section is an
Application to aviation safety 182
extension of the diagnostic decision trees described in section 4.2. Figure 32 shows the structure of the proposed design.
To ensure that the BADA unit receives navigation readings of high accuracy, a check for equipment faults is added to the network. The fault detection algorithm follows a model-based approach (see section 4.2) whereby the measured aircraft states are verified with a reading that has been calculated using a simulated aircraft running in parallel to the actual aircraft.
The simulated aircraft uses the high-fidelity 6 Degrees of Freedom (6DoF) model to simulate displacement and rotation of an aircraft in three-dimensional space. The 6DoF model is contained within the Math Engine Unit (MEU). The states from the aircraft and the math engine block are applied to the Fault Detection Unit (FDU), where a state of no fault is assumed if the two data sources are identical. Otherwise, the FDU will start a diagnostic procedure to isolate the malfunctioning equipment using Bayesian diagnostic decision trees. As a result, the BADA Unit (BU) can select the most reliable source of
Figure 32. Structure of the proposed design
Application to aviation safety 183
data to be compared to the recommended operation records, which were in turn obtained from the BADA database unit throughout the converter script.
The BU recommendation algorithm begins by checking for sources of reliable readings and the utility associated with making the recommendation. If no reliable information could be obtained, the BU will display a warning message informing the pilot about the situation. Otherwise, the BU begins by detecting the flight phase of the aircraft (i.e. cruise, climb, or descent). BU uses the value of the angle of attack, landing gear position and ROCD to detect the flight phase. Low ROCD and low angle of attack along with landing gear at the up position would indicate cruising phase. Otherwise, the aircraft is either climbing or descending. In order to detect which of these the aircraft is in, BU uses the ROCD values of the climb and descent from the PTF file as a feature search space. The nearest five neighbours to the current ROCD of the aircraft is calculated using the k-nearest neighbour algorithm, then the flight phase is determined based on basic majority vote. To reduce the amount of calculation, BU can be programmed to treat negative ROCD as descending indicator and positive ROCD as climbing indicator, which might be beneficial in situations in which no PTF file is available. When the flight phase is known, the corresponding look-up table for that specific phase is selected to obtain the recommended procedure data. If the measured aircraft states are not within the tolerated limit of the values recommended by BADA, the BU will inform the pilot about the situation and recommend changing his/her navigation parameters accordingly.