Strapdown inertial navigation systems are widely used in surface ships and warships. Although high-precision optical fibre inertial navigation systems are available, they have high cost and limited practicality. Therefore, they cannot replace the traditional platform inertial navigation systems in all ships. Hence, microelectromechanical system (MEMS)-based inertial sensors are widely used for robust navigation. Accurate and timely identification of sensor faults while ensuring stable navigation is a challenging task. This paper proposes a robust fault detection (FD) approach for a low-cost system that loosely integrates a strapdown inertial navigation system and the global navigation satellite system, where the integrated navigation state estimation provides high-accuracy output. A cubature Hi/H∞-optimised FD filter was designed for a nonlinear discrete time-varying system considering sensitivity to faults and robustness to disturbances. Furthermore, a threshold for FD was derived considering a compromise between the false alarm rate and fault diagnosis accuracy. Finally, the proposed method was validated through simulations using multiple noise distribution sensor data generated by a ship-manoeuvring simulator.

The aim morphing of this study is to detect and reconstruct a fault in angle-of-attack sensor and pitot probes that are components in commercial aircrafts, without false alarm and no need for additional measurements. Real flight data collected from a local airline was used to design the relevant system. Correlation analysis was performed to select the data related to the angle-of-attack and airspeed. Fault detection and reconstruction were carried out by using Adaptive Neural Fuzzy Inference System (ANFIS) and Artificial Neural Networks (ANN), which are machine-learning methods. No false alarm was detected when the fault test following the fault modeling was carried out at 0–1 s range by filtering the residual signal. When the fault was detected, fault reconstruction process was initiated so that system output could be achieved according to estimated sensor data. Instead of using the methods based on hardware redundancy, we designed a new system within the scope of this study.

The inertial navigation system/Doppler velocity logger (INS/DVL) plays an important role in ocean navigation. Any DVL malfunction poses serious risks to navigation. A precise detection system is required to detect the initial moments of DVL signal malfunctions; moreover, with loss of DVL, a fault-tolerant scheme (FTS) is necessary for DVL signal reconstruction. In this paper, an evolutionary knowledge-based method, namely improved evolutionary TS-fuzzy (I-eTS), is adopted to build an artificial intelligence (AI)-based pseudo DVL to deal with long-term outage of DVL. By employing Gaussian process regression (GPR) models for fault detection, a new FTS is constructed. To verify the effectiveness of the new fault detection and fault tolerance system, navigation data is gathered by a test setup and algorithms are performed in the laboratory. In the tests, it is demonstrated that the proposed FTS leads to rapid detection of both gradual and abrupt faults, which leads to less interaction between fault detection and FTS.

The variable stiffness actuator (VSA) is helpful to realize the post-collision safety strategies for safe human–robot interaction.1 The stiffness of the robot will be reduced to protect the user from injury when the collision between the robot and human occurs. However, The VSA has a mechanism limit constraint that can cause harm to users even if the stiffness is minimized. Accordingly, in this article, a concept combining danger index and robust fault detection and isolation is presented and applied to active–passive variable stiffness elastic actuator (APVSEA). APVSEA can actively change joint stiffness with the change of danger index. Experimental results show that this concept can effectively confirm the fault mode and provide additional protection measures to ensure the safety of users when the joint stiffness has been adjusted to the minimum.

Real-time and accurate fault detection and isolation is very important to ensure the reliability and precision of integrated inertial navigation and global navigation satellite systems. In this paper, the detection performance of a residual chi-square method is analysed, and on this basis an improved method of fault detection is proposed. The local test based on a standardised residual is introduced to detect and identify faulty measurements directly. Differing from the traditional method, two appropriate thresholds are selected to calculate the weight factor of each measurement, and the gain matrix is adjusted adaptively to reduce the influence of the undetected faulty measurement. The sliding window test, which uses past measurements, is also added to further improve the fault detection performance for small faults when the local test based on current measurements cannot judge whether a fault has occurred or not. Several simulations are conducted to evaluate the proposed method. The results show that the improved method has better fault detection performance than the traditional detection method, especially for small faults, and can improve the reliability and precision of the navigation system effectively.

In life-critical applications, the real-time detection of faults is very important in Global Positioning System/Inertial Navigation System (GPS/INS) integrated navigation systems. A new fault detection method for soft fault detection is developed in this paper with the purpose of improving real-time performance. In general, the innovation information obtained from a Kalman filter is used for test statistic calculations in Autonomous Integrity Monitored Extrapolation (AIME). However, the innovation of the Kalman filter is degraded by error tracking and closed-loop correction effects, leading to time delays in soft fault detection. Therefore, the key issue of improving real-time performance is providing accurate innovation to AIME. In this paper, the proposed algorithm incorporates Least Squares-Support Vector Machine (LS-SVM) regression theory into AIME. Because the LS-SVM has a good regression and prediction performance, the proposed method provides replaced innovation obtained from the LS-SVM driven by real-time observation data. Based on the replaced innovation, the test statistics can follow fault amplitudes more accurately; finally, the real-time performance of soft fault detection can be improved. Theoretical analysis and physical simulations demonstrate that the proposed method can effectively improve the detection instantaneity.

During the past decade, robotics for cochlear implant electrode array insertion has been limited to manipulation assistance. Going beyond manipulation assistance, this paper presents the new concept of perception augmentation to detect and warn against the onset of intracochlear electrode array tip folding. This online failure detection method uses a combination of intraoperative electrode insertion force data and a predictive model of insertion force profile progression as a function of insertion depth. The predictive model uses statistical characterization of insertion force profiles during normal robotic electrode array insertions as well as the history of intra-operative insertion forces. Online detection of onset of tip folding is achieved using the predictive model as an input into a support vector machine classifier. Results show that the detection of tip folding onset can be achieved with an accuracy of 88% despite the use of intra-operative insertion force data representing incomplete insertion. This result is significant because it allows the surgeon or robot to choose a corrective action for preventing intra-cochlear complications.

For the integrated navigation system, the correctness and the rapidity of fault detection for each sensor subsystem affects the accuracy of navigation. In this paper, a novel fault detection method for navigation systems is proposed based on Gaussian Process Regression (GPR). A GPR model is first used to predict the innovation of a Kalman filter. To avoid local optimisation, particle swarm optimisation is adopted to find the optimal hyper-parameters for the GPR model. The Fault Detection Function (FDF), which has an obvious jump in value when a fault occurs, is composed of the predicted innovation, the actual innovation of the Kalman filter and their variance. The fault can be detected by comparing the FDF value with a predefined threshold. In order to verify its validity, the proposed method is used in a SINS/GPS/Odometer integrated navigation system. The comparison experiments confirm that the proposed method can detect a gradual fault more quickly compared with the residual chi-squared test. Thus the navigation system with the proposed method gives more accurate outputs and its reliability is greatly improved.

In Global Navigation Satellite System (GNSS) positioning, it is standard practice to apply the Fault Detection and Exclusion (FDE) procedure iteratively, in order to exclude all faulty measurements and then ensure reliable positioning results. Since it is often only necessary to consider a single fault in a Receiver Autonomous Integrity Monitoring (RAIM) procedure, it would be ideal if a fault could be correctly identified. Thus, fault detection does not need to be applied in an iterative sense. One way of evaluating whether fault detection needs to be reapplied is to determine the probability of a wrong exclusion. To date, however, limited progress has been made in evaluating such probabilities. In this paper the relationships between different parameters are analysed in terms of the probability of correct and incorrect identification. Using this knowledge, a practical strategy for incorporating the probability of a wrong exclusion into the FDE procedure is developed. The theoretical findings are then demonstrated using a GPS single point positioning example.

This paper presents the application of new time frequency method, ensemble empirical modedecomposition (EEMD), in purpose to detect localized faults of damage at an early stage.EEMD is a self adaptive analysis method for non-linear and non-stationary signals and itwas recently proposed by Huang and Wu to overcome the drawbacks of the traditionalempirical mode decomposition (EMD). The vibration signal is usually noisy. To detect thefault at an early stage of its development, generally the residual signal is used. Thereexist different methods in literature to calculate the residual signal, in this paper wemention some of them and we propose a new method which is based on EEMD. The results givenby the different methods are compared by using simulated and experimental signals.

Le travail que nous présentons dans cet article a pour but de montrerles avantages liés à la mise en place d'une maintenance prédictive,relative aux outils de coupe pour les usinages à enlèvement de matière.Nous décrivons la méthode de détection des fautes basée sur l'utilisationdu modèle analytique de la rugosité des pièces usinées. Ce modèleest établi sous des conditions de coupe données, calculées en fonctionde la rugosité maximale souhaitée. Deux types de fautes, d'une partl'usure prématurée de l'outil de coupe et d'autre part le bris del'outil sont injectées lors de l'usinage des pièces et permettentainsi de valider la méthode de détection proposée. Le principe decette méthode est de comparer la rugosité mesurée à la rugosité calculéeà partir du modèle analytique. En l'absence de faute, les deux valeurstraduisant ces rugosités sont identiques, tandis que l'occurrence d'unefaute se traduit par une différence entre ces deux signaux, cettedifférence étant appelée résidu. Une maintenance systématique estégalement proposée pour pallier à l'usure normale de l'outil. La duréede vie de l'outil de coupe est calculée en utilisant le modèle dela rugosité couplé à la rugosité maximale spécifiée.

In this paper a discrete-time observer-based approach to Fault Detection and Isolation (FDI) for industrial robotic manipulators is presented and experimentally tested. In order to counteract the effects of unmodeled dynamics and disturbances, a time-delayed estimate of such effects is adopted. Remarkably, the observer is designed directly in the discrete-time domain. The performance of the proposed approach are experimentally verified on a six-degrees-of-freedom industrial robot.

Sophisticated robotic applications require systems to be reconfigurable at the system level. Aiming at this requirement, this paper presents the design and implementation of a software architecture for a reconfigurable real-time multi-processing system for multi-robot control. The system is partitioned into loosely coupled function units and the data modules manipulated by the function units. Modularized and unified structures of the sub-controllers and controller processes are designed and constructed. All the controller processes run autonomously and intra-sub-controller information exchange is realized by shared data modules that serve as a data repository in the sub-controller. The dynamic data-management processes are responsible for data exchange among sub-controllers and across the computer network. Among sub-controllers there is no explicit temporal synchronization and the data dependencies are maintained by using datum-based synchronization. The hardware driver is constructed as a two-layered system to facilitate adaptation to various robotic hardware systems. A series of effective schemes for software fault detection, fault anticipation and fault termination are accomplished to improve run-time safety. The system is implemented cost-effectively on a QNX real-time operating system (RTOS) based system with a complete PC architecture, and experimentally validated successfully on an experimental dual-arm test-bed. The results indicate that the architectural design and implementation are well suited for advanced application tasks.

In this note, fault detection techniques based on finite dimensional results are extended and applied to a class of infinite dimensional dynamical systems. This special class of systems assumes linear plant dynamics having an abrupt additive perturbation as the fault. This fault is assumed to be linear in the (unknown) constant (and possibly functional) parameters. An observer-based model estimate is proposed which serves to monitor the system's dynamics for unanticipated failures, and its well posedness is summarized. Using a Lyapunov synthesis approach extended and applied to infinite dimensional systems, a stable adaptive fault diagnosis (fault parameter learning) scheme is developed. The resulting parameter adaptation rule is able to “sense” the instance of the fault occurrence. In addition, it identifies the fault parameters using the additional assumption of persistence of excitation. Extension of the adaptive monitoring scheme to incipient faults (time varying faults) is summarized. Simulations studies are used to illustrate the applicability of the theoretical results.