Roger Xu

An integrated approach to bearing fault diagnostics and prognostics

This paper presents an integrated fault diagnostic and prognostic approach for bearing health monitoring and condition-based maintenance. The proposed scheme consists of three main components including principal component analysis (PCA), hidden Markov model (HMM), and an adaptive stochastic fault prediction model. The principal signal features extracted by PCA are utilized by HMM to generate a component health/degradation index, which is the input to an adaptive prognostics component for on-line remaining useful life prediction. The effectiveness of the scheme is shown by simulation studies using experimental vibration data obtained from a bearing health monitoring testbed.

A novel approach to fault diagnostics and prognostics

A novel fault diagnostics and prognostics algorithm based on hidden Markov model (HMM) is proposed. The algorithm combines fault diagnostics and prognostics in a unified framework. The algorithm has been fully tested by using experimental data from a rotating shift testbed in our laboratory.

An Enhanced Prognostic Model for Intermittent Failures in Digital Electronics

This paper presents an enhanced prognostic model to predict remaining useful life. The model utilizes environmental loads and in-situ performance measurements in conjunction with two baseline prediction algorithms: life consumption monitoring (LCM) and uncertainty adjusted prognostics (UAP). Fusion techniques are then utilized to integrate the two prognostic algorithms. A key and unique value of this combined prognostic model is its ability to assess intermittent as well as hard failures. In the paper we show how it has been validated for intermittent and hard solder joint interconnect failures under temperature cycling loads.

A neural network based approach to adaptive fault tolerant flight control

This work presents a neural network based approach to fault tolerant control of nonlinear flight control systems. The proposed scheme consists of three main components. First, a fault diagnosis scheme is designed to detect and identify automatically the occurrence of any faults. Second, a controller suite comprises a nominal controller and a neural network based adaptive fault tolerant controller designed to compensate for the effects of faults. Third, a reconfiguration supervisor makes decisions regarding controller reconfiguration. A simulation study using a nonlinear Beaver flight simulator illustrates the effectiveness of the proposed scheme and shows that flight safety is significantly improved by using the controller reconfiguration.

An intelligent hierarchical approach to actuator fault diagnosis and accommodation

This paper presents a novel intelligent hierarchical approach to automatically detecting, isolating, and accommodating faults in flight control systems. The proposed architecture has three main components. First, a new nonlinear fault diagnosis scheme is used to detect the occurrence of any faults and to determine the particular component that has failed. Second, a controller module consists of a primary nominal controller and a secondary adaptive fault-tolerant controller. While the nominal controller can be any existing conventional flight control system, the secondary neural network (NN) based nonlinear adaptive controller is designed to maintain acceptable control performance after the detection of fault occurrence. Third, a reconfiguration supervisor makes decision regarding controller reconfiguration and control reallocation by using on-line diagnostic information. Following failures of primary aerodynamic actuators, flight safety can be maintained by utilizing alternative actuation systems for critical stability and control augmentation tasks. The effectiveness of the proposed integrated fault diagnosis and accommodation approach has been illustrated by using the research civil aircraft model (RCAM) developed by the group for aeronautical research and technology in Europe (GARTEUR). Extensive simulation studies have clearly shown the benefits of the proposed adaptive fault-tolerant control scheme using on-line diagnostic information

Fault tolerant formation flight control of UAVs

In this paper, we summarise our recent results in fault tolerant formation control of Unmanned Air Vehicles (UAV). A fault tolerant control scheme to deal with both GPS sensor failure and wireless communication packet losses is presented. Moreover, some extensions to the formation control algorithms developed by UC Berkeley are made to support time-varying heading and curved flight trajectories. The effectiveness of the presented fault tolerant control scheme is illustrated by real-time formation flight simulations conducted in a wireless network environment. Both rotary and fixed wing UAVs, mesh and triangular formations, and straight and curved trajectories are considered in our real-time simulations.

IDDQ trending as a precursor to semiconductor failure

Airborne electronic systems have been used virtually everywhere on board military and commercial aircraft. Since Field Effect Transistors (FETs) are building blocks for the electronic systems and their components, the diagnosis and prognosis of potential FET system failures are critical to the flight and ground crew. In this paper, we developed an advanced prognostic methodology based on the Direct Drain Quiescent Current (IDDQ) testing technique for potential Field Effect Transistors (FETs) failures. To predict the Remaining useful life (RUL) of the FET-based devices, a thorough failure mechanism study for FETs was performed in order to select a subset of failure mechanisms that cause progressive degradation and relate with IDDQ signals. With the selected failure mechanisms, we utilized the symbolic dynamics-based method to perform the fault degradation status estimation and a novel Uncertainty Adjusted Prognostics (UAP) method to predict the RUL with uncertainty management. Finally, the prognostic methodology was verified using developed 2-D/3-D simulation models.

Toxic Vapor Classification and Concentration Estimation for Space Shuttle and International Space Station

Abstract. During space walks, the space suits of astronauts may be contaminated by toxic vapors such as hydrazine, which are used for attitude control. Here we present some initial results on vapor classification and concentration estimation by using Support Vector Machine (SVM). The vapor was collected by electronic nose. By collaborating closely with NASA KCS, we achieved great results. For example, for Kam15f (90-second) data set, the classification success rate was 97.5% using SVM as compared to 87% using the linear discriminant method in [1]. Comparative studies were conducted between the SVM classifier and other classifiers such as Back Propagation (BP) Neural Network, Probability Neural Network (PNN), and Learning Vector Quantization (LVQ). In all cases, the SVM classifier showed superior performance over other classifiers. In the concentration estimation part by using SVM, we achieved more than 99% correct estimation of concentration by using the 90 th second data samples.

Fault Detection and Identification in Aircraft Hydraulic Pumps Using MCA

Abstract Early detection of aircraft hydraulic pump failures is critical for safety of flights. Hydraulic pump failures are usually indicated by increasing noise-to-signal ratio in the case drain flow. In this paper, we present a robust scheme for detecting and identifying pump failures using minor component analysis. First, a residual model is generated off-line using sensor data collected under normal system operation conditions. Then the residual generation model can be used on-line to process new sensor data and detect any abnormal pump behaviors. Moreover, a novel Cault identification algorithm is derived to estimate the Cault size after Cault detection, which could be very useful for component remaining life prediction and fault isolation.

Sensor validation using nonlinear minor component analysis

In this paper, we present a unified framework for sensor validation, which is an extremely important module in the engine health management system. Our approach consists of several key ideas. First, we applied nonlinear minor component analysis (NLMCA) to capture the analytical redundancy between sensors. The obtained NLMCA model is data driven, does not require faulty data, and only utilizes sensor measurements during normal operations. Second, practical fault detection and isolation indices based on Squared Weighted Residuals (SWR) are employed to detect and classify the sensor failures. The SWR yields more accurate and robust detection and isolation results as compared to the conventional Squared Prediction Error (SPE). Third, an accurate fault size estimation method based on reverse scanning of the residuals is proposed. Extensive simulations based on a nonlinear prototype non-augmented turbofan engine model have been performed to validate the excellent performance of our approach.