Fault Diagnosis in a Flight Actuator using Extended Kalman Filter Parameter Estimator

Paper number

IAC-07-D1.2.07

Author

Mr. M. Jayakumar, Indian Space Research Organisation (ISRO), VSSC, India

Coauthor

Dr. Bijan B Das, Indian Space Research Organisation (ISRO), VSSC, India

Year

2007

Abstract

Expeditious diagnosis and prognosis of faults in mission critical systems are identified as enabling technologies for future space transportation systems. Such techniques initiate proactive steps like mission salvage to prevent loss of vehicle in presence of impending failures and allow condition based maintenance scheduling. Conventional fault detection methods based on hardware redundancy, limit checking, etc., suffer from an inherent limitation – inability for fault localization. An alternate method that permits deep diagnosis of faults and which offers advantages in cost, weight, power consumption and reliability is the analytical redundancy technique. In this, the inherent redundancy contained in the static and dynamic relationships between the system inputs and measured outputs is exploited for fault diagnosis. This paper presents an analytical redundancy scheme for the diagnosis of process faults in a DC motor-based electromechanical flight control actuation system. 

In this paper, model based parameter estimation technique that allows deep diagnosis is used for identifying faults. Extended Kalman Filter (EKF), that facilitates simultaneous state and parameter estimation is used for estimating the model parameters of the system. The model parameters to be estimated are considered as extra states in an augmented state vector. From the identified model parameters the physical parameters of the actuation system like motor resistance, inductance, torque constant, etc. are extracted. By estimating the deviation of the physical parameters from nominal values, analytic symptoms indicative of impending electrical and mechanical faults in the system are generated. The symptoms are then represented as fuzzy sets with appropriate membership functions. A-priori knowledge of basic relationships between faults and symptoms combined with fuzzy inference is used for fault diagnosis. Fuzzy logic reasoning is done adopting the principles of fuzzy IF-THEN rule systems with the steps: fuzzification, inference, accumulation and defuzzification. 

Majority of the work reported in this area have used recursive least square parameter identification technique and have relied on simplified models of actuators. This paper uses EKF along with complete model of the flight actuator, which includes mounting structure stiffness, load dynamics, compensators and power amplifier with current loop for process fault diagnosis under realistic flight conditions.

The system was modeled and simulations were carried out to assess the efficiency of the proposed fault diagnosis methodology. From the simulation results it is found that the parameter estimation and fuzzy inference scheme formulated is capable of efficiently diagnosing impending process faults like winding/commutator failures, mechanical degradation, etc., in flight control actuators.

Abstract document

IAC-07-D1.2.07.pdf

Manuscript document

IAC-07-D1.2.07.pdf (🔒 authorized access only).

To get the manuscript, please contact IAF Secretariat.