The aim of this study was to develop a method for airplane ground roll distance measurement. This method is based on measurements of aircraft accelerations, magnetic field vector, as well as Global Positioning System (GPS) position. A convolution neural network was developed in this study and it recognizes characteristic moments of flight during takeoff: longitudinal acceleration, nose wheel up (rotation), and liftoff, as well as during landing: approach, touchdown, rollout, and stop. The developed neural network includes two filters that enable collection of input data from both the total dataset as well as from data portions for information about the occurrence of a peak at the time of touchdown and the change of acceleration at the time of takeoff.The distance traveled by the plane on the ground during takeoff or landing is then calculated using the GPS coordinates, measured simultaneously. The method uses a sensor system consisting of accelerometers, magnetometers, and a GPS receiver; a prototype hardware was based on a single-board miniature computer. The hardware is small in size and fits anywhere in the aircraft cockpit. The method has been practically examined by means of flight tests and it was concluded that it performs well: the mean difference in the measured take-off and landing distances from the results obtained with the reference method was 2,92 % and 6,31 % respectively. The method was not sensitive to airplane design type: low wing vs. high wing. Also, the method has been tested for two different pilots and no significant difference in the results of measurement was observed.
Read full abstract