RESEARCH ON FAULT DIAGNOSIS AND FAILURE RECONFIGURATION OF FLUSH AIR DATA SENSING

In 2018 and 2019, two Boeing 737 MAX belonging to PT Lion Mentari Airlines and Ethiopian Airlines crashed, resulting in 346 deaths. Afterward, the accident investigation revealed that the main reason was the failure of the traditional angle of attack sensor. Unlike the traditional probe-type atmospheric data sensor, flush air data sensing (FADS) employs the pressure values measured by pressure sensors that are embedded in pressure taps to deduce angle of attack α, angle of sideslip β, Mach number Ma, dynamic pressure qc, and static pressure P_∞, which effectively overcomes the former shortcomings and deficiencies. As a result, many countries, including the United States, France, and the United Kingdom, have carried out relevant research. The chief purpose of this paper is to propose a new fault diagnosis method for FADS and carry out further research on failure reconfiguration. First, the aerodynamic knowledge under subsonic and supersonic conditions is applied to establish a high-precision aerodynamic model of the sensor. Considering the severe working environment such as low temperature and low pressure, pressure taps in FADS will inevitably fail during the process of application. To address the problem, a new fault diagnosis method for the failure of single pressure tap and simultaneous failure of multiple pressure taps is proposed with the consideration of redundant signals. Besides, to reduce the false alarm rate and improve diagnosis accuracy, the threshold of alarm times is designed with statistical knowledge. Furthermore, after the fault diagnosis of FADS is realized by using the proposed method, the next step is to make use of the remaining normal pressure taps to continue measuring atmospheric data, i.e., failure reconfiguration. In this part of the study, the fault is divided into two cases: 1) single pressure tap fails; 2) multiple pressure taps fail. When a single pressure tap fails, we start from the derivation algorithm and employ the redundancy of FADS to obtain the final measurement result by reconstructing signals. However, when multiple pressure taps fail at the same time, all the measurement signals are wrong, and the above method is invalid. To address the problem, data fitting approach is firstly adopted in FADS to estimate α. To verify the effectiveness of the method, two representative examples are adopted in this paper.