Double-model adaptive fault detection and diagnosis applied to real flight data Peng Lu n , Laurens Van Eykeren, Erik-Jan van Kampen, Coen de Visser, Qiping Chu Delft University of Technology, Kluyverweg 1, 2629HS Delft, The Netherlands article info Article history: Received 12 August 2014 Accepted 4 December 2014 Keywords: Fault detection and diagnosis Air data sensors Double-model adaptive estimation Real flight test data Unscented Kalman filter abstract The existing multiple model-based estimation algorithms for Fault Detection and Diagnosis (FDD) require the design of a model set, which contains a number of models matching different fault scenarios. To cope with partial faults or simultaneous faults, the model set can be even larger. A large model set makes the computational load intensive and can lead to performance deterioration of the algorithms. In this paper, a novel Double-Model Adaptive Estimation (DMAE) approach for output FDD is proposed, which reduces the number of models to only two, even for the FDD of partial and simultaneous output faults. Two Selective-Reinitialization (SR) algorithms are proposed which can both guarantee the FDD performance of the DMAE. The performance is tested using a simulated aircraft model with the objective of Air Data Sensors (ADS) FDD. Another contribution is that the ADS FDD using real flight data is addressed. Issues related to the FDD using real flight test data are identified. The proposed approaches are validated using real flight data of the Cessna Citation II aircraft, which verified their effectiveness in practice. & 2014 Elsevier Ltd. All rights reserved. 1. Introduction Presently, Fault Detection and Diagnosis (FDD) is important to achieve fault-tolerance (Patton, 1997). For flight control systems, sensor or actuator faults may cause serious problems. Thus, quick detection and isolation of these faults is highly desirable (Goupil, 2011). During the last few decades, many approaches have been proposed for aircraft actuator and sensor Fault Detection and Isolation (FDI) (Chen & Patton, 1999; Isermann, 2005; Marzat, Piet-Lahanier, Damongeot, & Walter, 2012). Some recent advances and trends can be found in Zolghadri (2012) and Goupil (2011). One recent European project, Advanced Fault Diagnosis for Sus- tainable Flight Guidance and Control (ADDSAFE), aims to develop FDI methods for aircraft flight control systems (Goupil & Marcos, 2014). Within this project, a number of model-based FDI methods were tested and evaluated, refer to Varga and Ossmann (2014), Van Eykeren and Chu (2014), Henry, Cieslak, Zolghadri, and Efimov (2014), Alwi and Edwards (2014), Chen, Patton, and Goupil (2012), Vanek, Edelmayer, Szabó, and Bokor (2014), Hecker and Pfifer (2014) and Marcos (2012). However, few of these papers (Van Eykeren & Chu, 2014) consider the FDD of the Air Data Sensors (ADS). The ADS measure the air data information which is critical to the pilot and to the flight control system. They are usually mounted to the outside of the fuselage. Therefore, they can be affected by the environment in which the aircraft is flying. Faults of the ADS are contributing factors which have led to several aircraft accidents. For civil aircraft, the final report of the Air France Flight 447 accident stated that erroneous airspeed mea- surements from the pitot probes were a contributing factor (Lombaerts, 2010). An example for military aircraft is the cause of the crash of a B-2 Bomber; it was found that moisture in the port transducer units caused a large bias to the ADS (Lombaerts, 2010). These are only two examples of recent air disasters caused by failures of the ADS system. Therefore, the FDD of the ADS is important. Recently, Freeman, Seiler, and Balas (2013) model the faults of the ADS using the physical air data relationships and experimental wind tunnel data. The present paper deals with the detection and diagnosis of the ADS faults. One of the most effective approaches for the FDD is the multiple-model-based approach (Zhang & Li, 1998). The basic idea of performing FDD using the multiple-model (MM) approach is: a Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/conengprac Control Engineering Practice http://dx.doi.org/10.1016/j.conengprac.2014.12.007 0967-0661/& 2014 Elsevier Ltd. All rights reserved. Abbreviation: ADDSAFE, advanced fault diagnosis for sustainable flight guidance and control; ADS, air data sensors; DMAE, double-model adaptive estimation; DMAE-NSR, double-model adaptive estimation-no selective reinitialization; FDD, fault detection and diagnosis; FDI, fault detection and isolation; GPS, global positioning systems; IMM, interacting multiple-model; IMU, inertial measurement unit; MM, multiple-model; MMAE, multiple-model adaptive estimation; SRMMAE, selective-reinitialization multiple-model adaptive estimation; SR, selective- reinitialization; UKF, unscented Kalman filter n Corresponding author. E-mail addresses: P.Lu-1@tudelft.nl (P. Lu), L.VanEykeren@tudelft.nl (L. Van Eykeren), E.vanKampen@tudelft.nl (E. van Kampen), c.c.devisser@tudelft.nl (C. de Visser), q.p.chu@tudelft.nl (Q. Chu). Control Engineering Practice 36 (2015) 39–57