K. Khorasani

Dynamic neural network-based fault diagnosis of gas turbine engines

In this paper, a neural network-based fault detection and isolation (FDI) scheme is presented to detect and isolate faults in a highly nonlinear dynamics of an aircraft jet engine. Towards this end, dynamic neural networks (DNN) are first developed to learn the input-output map of the jet engine. The DNN is constructed based on a multi-layer perceptron network which uses an IIR (infinite impulse response) filter to generate dynamics between the input and output of a neuron, and consequently of the entire neural network. The trained dynamic neural network is then utilized to detect and isolate component faults that may occur in a dual spool turbo fan engine. The fault detection and isolation schemes consist of multiple DNNs or parallel bank of filters, corresponding to various operating modes of the healthy and faulty engine conditions. Using the residuals that are generated by measuring the difference of each network output and the measured engine output various criteria are established for accomplishing the fault diagnosis task, that is addressing the problem of fault detection and isolation of the system components. A number of simulation studies are carried out to demonstrate and illustrate the advantages, capabilities, and performance of our proposed fault diagnosis scheme.

An inverse dynamics control strategy for tip position tracking of flexible multi‐link manipulators

In this article, we present an inverse dynamics control strategy to achieve small tracking errors for a class of multi-link structurally flexible manipulators. This is done by defining new outputs near the end points of the arms as well as by augmenting the control inputs by terms that ensure stable operation of the closed loop system under specific conditions. The controller is designed in a two-step process. First, a new output is defined such that the zero dynamics of the original system are stabilized. Next, to ensure stable asymptotic tracking, the control input is modified such that stable asymptotic tracking of the new output or approximate tracking of the actual output may be achieved. This is illustrated for the case of single- and two-link flexible manipulators. ©1997 John Wiley & Sons, Inc.

Fault detection and isolation of a dual spool gas turbine engine using dynamic neural networks and multiple model approach

In this paper, a fault detection and isolation (FDI) scheme for an aircraft jet engine is developed. The proposed FDI system is based on the multiple model approach and utilizes dynamic neural networks (DNNs) to accomplish this goal. Towards this end, multiple DNNs are constructed to learn the nonlinear dynamics of the aircraft jet engine. Each DNN corresponds to a specific operating mode of the healthy engine or the faulty condition of the jet engine. Using residuals obtained by comparing each network output with the measured jet engine output and by invoking a properly selected threshold for each network, reliable criteria are established for detecting and isolating faults in the jet engine components. The fault diagnosis task consists of determining the time as well as the location of a fault occurrence subject to presence of unmodeled dynamics, disturbances, and measurement noise. Simulation results presented demonstrate and illustrate the effectiveness of our proposed dynamic neural network-based FDI strategy.

Adaptive friction compensation based on the Lyapunov scheme

A simple method for constructing a nonlinear estimator for adaptive friction compensation is developed. The design is based on the Lyapunov technique and an attempt is made to compensate for frictional force by estimating the unknown Coulomb friction coefficient. The contribution of this paper is to provide a systematic procedure for selecting a nonlinear function in the estimator. It is shown that asymptotic stability of the error dynamics is guaranteed without imposing a constraint on the velocity. Simulation results confirm the advantages of the proposed method for a single-mass nonlinear system as well as a complicated nonlinear system such as a two-link planar rigid robot manipulator.

Adaptive formation control of UAVs in the presence of unknown vortex forces and leader commands

In this paper, stable adaptive formation control algorithms for two-dimensional models of aircraft are developed in the presence of unknown leader commands and disturbances due to vortex effects. The analysis is presented in the leader-follower frame for three cases of unknown leader commands and vortex forces in the velocity and heading angle dynamics. The algorithms are applied to the formation flight control of two UAVs. Simulation results show that the follower follows precisely the leader in the presence of uncertainties in the aerodynamic coefficients and leader commands

Dynamic neural network-based estimator for fault diagnosis in reaction wheel actuator of satellite attitude control system

This paper presents an approach to simultaneous fault detection and isolation in the reaction wheel actuator of the satellite attitude control system. A model-based adaptive nonlinear parameter estimation technique is used based on a highly accurate reaction wheel dynamical model while each parameter is an indication of a specific type of fault in the system. The estimation is based on the nonlinear finite-memory filtering strategy that is solved for optimal estimation functions. To make the optimization feasible for on-line application, the optimal estimation functions are approximated by MLP neural networks thus reducing the functional optimization problem to a nonlinear programming problem, namely, the optimization of the neural weights. The well-known standard back-propagation algorithm and backpropagation through-time algorithm were employed inside the neural adaptation algorithms to obtain the required gradients. Simulation results show the effectiveness of the methodology for the proposed application.

Interactive Bank of Unscented Kalman Filters for Fault Detection and Isolation in Reaction Wheel Actuators of Satellite Attitude Control System

The main objective of the research investigated in this paper is the detection and isolation of partial (soft) and total (hard) failures in the reaction wheel (RW) actuators of the satellite attitude control system (ACS) during its mission operation. The fault detection and isolation (FDI) is accomplished using the interactive multiple models (IMM) scheme developed based on the unscented Kalman filter (UKF) algorithm. Towards this objective, the healthy mode of the ACS system under different operating conditions as well as a number of different fault scenarios including changes and anomalies in the temperature, power supply bus voltage, and unexpected current variations in the actuators of each axis of the satellite are considered. We describe and develop a bank of interacting multiple model unscented Kalman filters (IMM-UKF) to detect and isolate the above mentioned reaction wheel failures in the ACS system. Also, it should be emphasized that the proposed IMM-UKF technique is implemented based on a high-fidelity highly nonlinear model of a commercial RW. Compared to other fault detection and isolation (FDI) strategies developed in the control systems literature, the proposed FDI scheme is shown, through extensive numerical simulations, to be more accurate, less computationally demanding, and more robust with the potential of extending to a number of other engineering applications

A robust adaptive congestion control strategy for large scale networks with differentiated services traffic

In this paper, a robust decentralized congestion control strategy is developed for a large scale network with Differentiated Services (Diff-Serv) traffic. The network is modeled by a nonlinear fluid flow model corresponding to two classes of traffic, namely the premium traffic and the ordinary traffic. The proposed congestion controller does take into account the associated physical network resource limitations and is shown to be robust to the unknown and time-varying delays. Our proposed decentralized congestion control strategy is developed on the basis of Diff-Serv architecture by utilizing a robust adaptive technique. A Linear Matrix Inequality (LMI) condition is obtained to guarantee the ultimate boundedness of the closed-loop system. Numerical simulation implementations are presented by utilizing the QualNet and Matlab software tools to illustrate the effectiveness and capabilities of our proposed decentralized congestion control strategy.

Detection of actuator faults using a dynamic neural network for the attitude control subsystem of a satellite

The main objective of this paper is to develop a neural network-based residual generator for fault detection (FD) in the attitude control subsystem (ACS) of a satellite. Towards this end, a dynamic multilayer perceptron (DMLP) network with dynamic neurons is considered. The neuron model consists of a second order linear IIR filter and a nonlinear activation function with adjustable parameters. Based on a given set of input-output data pairs collected from the attitude control subsystem, the network parameters are adjusted to minimize a performance index specified by the output estimation error. The proposed dynamic neural network structure is applied for detecting faults in a reaction wheel (RW) that is often used as an actuator in the ACS of a satellite. The performance and capabilities of the proposed dynamic neural network is investigated and compared to a model-based observer residual generator design that is to detect various fault scenarios.

An inverse dynamics sliding control technique for flexible multi-link manipulators

In this paper, a control strategy based on nonlinear inversion is considered that results in small tip-position tracking errors while maintaining robust closed-loop performance for a class of multi-link structurally flexible manipulators. This is achieved by defining new outputs near the end points of the arms as well as by augmenting the nominal control inputs by terms that ensure stable operation of the closed loop system in the presence of considerable parametric uncertainties. Motivated by the concept of a sliding surface in variable structure control (VSC) a robust control term is developed to drive the nonlinear plants error dynamics onto a sliding surface. The discontinuous functions normally used in classical VSC are replaced by saturation nonlinearities at the outset. This also facilitates analysis by the standard Lyapunov techniques. The controller performance is demonstrated by simulation on a two-link flexible manipulator with a considerable amount of parametric uncertainty.