A SCHEME OF SFDIA USING MULTILAYER PERCEPTRON AND EBP FOR UNMANNED AERIAL VEHICLE

Sudeesh , PD and Kashyap, SK and Kayalvizhi, R (2008) A SCHEME OF SFDIA USING MULTILAYER PERCEPTRON AND EBP FOR UNMANNED AERIAL VEHICLE. In: International Conference on Aerospace Science and Technology (INCAST 2008-094), 26-28 Jun 2008, Bangalore, India.

Full text available as:
[img] PDF - Published Version
Download (80Kb)

    Abstract

    An aircraft can be considered as a time-varying, non-linear system affected by process noise which could be due to external disturbances such as wind or gust. The aircraft states such as angle-of-attack, side-slip angle, angular rates, accelerations and velocity of aircraft are measured by sensors namely vanes, gyros,accelerometers and pitot-tube respectively. The measurements of these sensors are likely to be corrupted by measurement noise. These measured data are used directly/indirectly as feedback signal to control the aircraft to a desired flight condition. So, occurrence of fault in any of these sensors could affect the feedback signal which in turn can lead to instability of an aircraft. In order to get rid of the problem arising due to sensor faults, we need to accurately detect, isolate and accommodate the fault so that feedback signal is not affected. Sensor Fault Detection Identification and Accommodation (SFDIA) is an important part of safety critical systems used in aircraft for safer operations. SFDIA can be achieved either by hardware redundancy or analytical redundancy technique. The advantages like reduced complexity, cost and weight of analytical redundancy over hardware redundancy encourages the designers to follow the former technique. The analytical redundancy techniques could be one of model based or non-model based approaches. The observer based residual generation, parity based approach and parameter based approach are the different model based techniques. In non-model based approach, fuzzy decision-making and artificial neural networks is used for building analytical redundancy. Due to the learning and adaptation capability of Neural Network (NN), applicability to nonlinear and multivariable systems, parallel distributed processing and hardware implementation, Artificial NNs are very appealing for the purpose of providing fault tolerance capabilities in a flight control system following sensor failures. In this paper, the SFDIA is achieved by using a Main Neural Network (MNN) and n Decentralized Neural Networks (DNNs) for a system with n sensors. Here MNN is used to detect the fault and DNN is used for identifying the faulty sensor. The reconfiguration of faulty sensor can be achieved by feeding back the DNN estimate for the faulty sensor instead of sensor measurement to the flight control system. Thus this SFDIA scheme eliminates the physical redundancy. The SFDIA scheme is implemented using MATLAB for closed loop model of Unmanned Aerial Vehicle (UAV) (Aerosonde). The SFDIA algorithm is evaluated for different types of single and multiple faults. In each case SFDIA successfully detected, identified and accommodated for faulty sensor

    Item Type: Conference or Workshop Item (Poster)
    Subjects: AERONAUTICS > Aircraft Stability and Control
    Division/Department: Other, Flight Mechanics and Control Division, Other
    Depositing User: Dr Sudesh Kumar Kashyap
    Date Deposited: 30 Jun 2010 09:57
    Last Modified: 30 Jun 2010 09:57
    URI: http://nal-ir.nal.res.in/id/eprint/8462

    Actions (login required)

    View Item