Pilot in loop assessment of fault tolerant flight control schemes in a motion flight simulator

Abstract

Pilot in Loop Assessment of Fault Tolerant Flight Control Schemes in a Motion Flight Simulator Girish Kumar Sagoo This research presents the pilot in the loop tests carried out in a Six-Degree of Freedom (6-DOF) motion flight simulator to evaluate failure detection, isolation and identification (FDII) schemes for an advanced F-15 aircraft. The objective behind this study is to leverage the capability of the flight simulator at West Virginia University (WVU) to carry out a performance assessment of neurally augmented control algorithms developed on a Matlab/Simulink ® platform. The experimental setup features an interface setup of Gen-2 Simulink® ® schemes with MOTUS Flight Simulator (MFS). The set up is a close substitute to a real flight and thus is helpful in evaluation of the schemes in a realistic manner. The graphics in X-plane is used to obtain visual cues and the motion platform is used to obtain motion cues in the simulator cockpit. The whole setup enables the pilot to respond with a joystick in the advent of a failure as he would otherwise in a real flight. The pilot response in maintaining the mission profile is different for different neural network augmentations and thus an indication of performance comparison of these schemes. Secondly, FDII schemes are developed for a sensor and actuator failure using an adaptive threshold for cross-correlation coefficients of the angular rates of the aircraft. Failure detection, isolation and identification logic is formulated based on monitoring the crosscorrelation parameters with their Floating Limiter (FL) bounds. The FDII scheme developed shows a good performance with desktop simulation because of no pilot activity but with a pilot in the loop significant cross-correlation of the rates occur and hence the scheme become more susceptible to wrongs FDII. In addition, the pilot might induce some coupling of the crosscorrelation parameters between detection and identification time which may trigger false detections and may configure the controller differently based on incorrect detection. Thus it is necessary that FDII scheme accommodate real flight conditions. The performance of the FDII schemes is improved with a pilot in the loop by monitoring the cross-correlation parameters and fine tuning FDII algorithms for real situations. This study has set up an excellent example to effectively utilize the aural, visual and motion cues to create a higher level of simulation complexity in designing control algorithms.

Development and evaluation of a fault detection and identification scheme for the WVU YF-22 UAV using the artificial immune system approach

“…The FDII scheme as described in [80] is divided into three main steps: • Roll-pitch rate Cross-correlation parameter:…”

“…• Angular rate dominance parameter: [81]. The work presented in [80] that constitutes a more complete analysis of this FDII method is used in this analysis as the starting point.…”

“…All these parameters as well as the filter itself admit changes in order to make the FL more or less sensitive. For the implementation in [80] only an upper bound is used ( ).…”

“…The interface as described in [80] is still functional; however, very little documentation was generated and thus, to be able to modify it, efforts were conducted to obtain more flexibility. Figure 15 presented a schematic of the interaction between the different elements involved in this interface.…”

“…• Use Master Warning as visual mean to inform pilot of Failure Detection also triggering a sounding alarm (sim/cockpit/warnings/annunciators/master_warning). • Display an on screen message with identification results using a more elaborate text box than the one used in [80].…”