Robust Data-Driven Fault Detection: An Application to Aircraft Air Data Sensors

Table 3

Training and testing data specifics in this paper.

	Aircraft and condition	Duration (min)	Altitude (km)	Airspeed (m/s)	AOA (°)	Sideslip angle (°)	Case distribution in {05} (%)
Training	AP cruise	327	[9.63, 10.7, 0.35]	[227, 252, 9]	[-1.3, 0.6, 0.4]	[-1.9, 0.4, 0.6]	{27, 13, 16, 16, 14, 14}
	manual LTO	295	[0.03, 0.66, 0.11]	[93, 167, 14]	[-2.7, 7.7, 1.2]	[-2.2, 1.6, 0.3]	{28, 14, 13, 13, 18, 14}
Testing	manual LTO	67		[75, 151, 12]	[0.8, 6.7, 0.8]	[-1.5, 0.5, 0.3]	{28, 20, 14, 11, 16, 11
	manual flight	30		[80, 141, 21]	[10, 11, 45]	[-3.8, 9.6, 2.5]	{10, 10, 17, 19, 15, 29}
	AP cruise	162	[3.50, 4.00, 0.19]	[68, 71, 0.38]	[0.5, 3.2, 0.37]	[-7.3, 2.7, 2.2]	{23, 16, 17, 14, 17, 13}
	manual flight	151	[0.01, 1.61, 0.26]	[47, 276, 31]	[-14, 19, 4]	[-8, 5, 1]	{29, 16, 11, 14, 14, 16}

Note: [minimum, maximum, stand deviation] of clean altitude and ADS states are characterized; altitude was not recorded in and real flight.