Hybrid deep fault detection and isolation: Combining deep neural networks and system performance models

SummaryThis paper proposes an integrated actuator and sensor active fault‐tolerant model predictive control scheme. In this scheme, fault detection is implemented by using a set‐valued observer, fault isolation (FI) is performed by set manipulations, and fault‐tolerant control is carried out through the design of a robust model predictive control law. In this paper, a set‐valued observer is used to passively complete the fault detection task, while FI is actively performed by making use of the constraint‐handling capability of robust model predictive control. The set‐valued observer is chosen to implement fault detection and isolation (FDI) because of its simple mathematical structure that is not affected by the type of faults such as sensor, actuator, and system‐structural faults. This means that only one set‐valued observer is needed to monitor all considered actuator and sensor statuses (health and fault) and to carry out the fault detection and isolation task instead of using a bank of observers (each observer matching a health/fault status). Furthermore, in the proposed scheme, the advantage of robust model predictive control is that it can effectively deal with system constraints, disturbances, and noises and allow to implement an active FI strategy, which can improve FI sensitivity when compared with the passive FI methods. Finally, a case study based on the well‐known two‐tank system is used to illustrate the effectiveness of the proposed fault‐tolerant model predictive control scheme. Copyright © 2016 John Wiley & Sons, Ltd.

This paper introduces a new nonlinear filter and deep learning-based Fault Detection and Isolation (FDI) approach for gas turbines. The system is first identified using a nonlinear feedforward network, including Finite Impulse Response (FIR) filters and Long Short-Term Memory (LSTM) networks. Following that, adaptive thresholds are devoted to generating residuals, and a Convolutional Neural Network (CNN) is employed for fault isolation. Utilizing two deep neural network frameworks ensures robust and accurate system identification under uncertainties and significantly improves fault detection and isolation processes compared to conventional methods. The proposed approach is validated on a heavy-duty gas turbine model. Test results show enhanced fault detection accuracy with reduced false alarms and improved fault isolation efficiency by correctly distinguishing fault effects in the system.

This paper covers model-based fault detection and isolation for linear and nonlinear distributed parameter systems (DPS). The first part mainly deals with actuator, sensor and state fault detection and isolation for a class of DPS represented by a set of coupled linear partial differential equations (PDE). A filter based observer is designed based on the linear PDE representation using which a detection residual is generated. A fault is detected when the magnitude of the detection residual exceeds a detection threshold. Upon detection, several isolation estimators are designed using filters whose output residuals are compared with predefined isolation thresholds. A fault on a linear DPS is declared to be of certain type if the corresponding isolation estimator output residual is below its isolation threshold while the other fault isolation estimator output residual is above its threshold. Next, the fault location is determined when a state fault is identified. The second part of this paper focuses on fault detection and isolation of nonlinear DPS by using a Luenberger type observer. Here fault isolation framework is introduced to isolate actuator, sensor and state faults with isolability condition by using additional boundary measurements and without filters. Finally, the effectiveness of the proposed fault detection and isolation schemes for both linear and nonlinear DPS are demonstrated through simulation.

The problem of designing and developing a hybrid fault detection and isolation (FDI) scheme for a non-linear system that is subject to large environmental disturbances is investigated. In the proposed FDI scheme, a hybrid architecture is introduced, which is composed of a bank of residual generators and a discrete-event system (DES) fault diagnoser. A novel set of residuals is generated so that a DES fault diagnoser will robustly detect and isolate actuator faults in the system by incorporating an appropriate combination of residuals and their sequential features. Necessary and sufficient conditions for the existence of a set of residuals that are used by the DES fault diagnoser are derived based on the non-linear geometric FDI approach. The proposed hybrid FDI scheme is then applied to actuator fault detection and isolation of an almost-lighter-than-air-vehicle. Simulation results presented demonstrate and validate the performance capabilities of the proposed hybrid FDI scheme.

In this paper, we consider the fault Detection and isolation (FDI) problem for faulty linear parameter-varying (LPV) systems subject to disturbances and propose a observer-based solution by using multiobjective optimization techniques. To simply the design process, a general faulty LPV system will be constructed from standard LPV description by converting actuator/system component faults into sensor faults at first. Then a bank of LPV FDI filters will be designed to identify each fault. Each FDI filter could generate a residual signal to track individual fault with minimum error and suppressing the effects of disturbances and other fault signals. The design of FDI filters will be formulated as multiobjective optimization problems in terms of linear matrix inequalities (LMIs) and can be solved efficiently. Two numerical examples are also presented to demonstrate the proposed fault detection and isolation approach on both LPV and LTI systems.

This work presents two fault detection and isolation (FDI) approaches for wind turbine systems (WTS). Firstly, a non-linear mathematical model for wind turbine (WT) dynamics is developed. Based on the developed WTS mathematical model, a robust fault detection observer is designed to estimate system faults, so as to generate residuals. The observer is designed to be robust to system disturbance and sensitive to system faults. A WT blade pitch system fault, a drive-train system gearbox fault and three sensor faults are simulated to the nominal system model, and the designed observer is then to detect these faults when the system is subjected to disturbance. The simulation results showed that the simulated faults are successfully detected. In addition, a neural network (NN) method is proposed for WTS fault detection and isolation. Two radial basis function (RBF) networks are employed in this method. The first NN is used to generate the residual from system input/output data. A second NN is used as a classifier to isolate the faults. The classifier is trained to achieve the following target: the output are all “0”s for no fault case; while the output is “1” if the corresponding fault occurs. The performance of the developed neural network FDI method was evaluated using the simulated three sensor faults. The simulation results demonstrated these faults are successfully detected and isolated by the NN classifier.

substantially from the prescribed or expected characteristics. To accommodate failures, the baseline inner-loop control is assisted by a fault detection and isolation algorithm tasked with monitoring and reporting control surface health. Similarly, the baseline outer-loop guidance is augmented with an adaptive neural network forming an adaptive outer-loop guidance system. These adaptive subsystems have been formulated for, and applied to, a transport aircraft model in a nonlinear, high-fidelity aircraft simulation environment to demonstrate the concept. Included results demonstrate that the fault detection and isolation and adaptive outer-loop functionality consistently improve tracking performance during both piloted maneuvers and autonomous waypoint path following. In some cases, these retrofit subsystems enable tasks to be completed that were not previously feasible. Adaptive enhancement of the robustness of existing non-adaptive systems has been an area of significant interest for many years. Although application in the aircraft community is of great benefit for maximizing performance and safety for unexpected changes in the dynamics due to flight control failures, damage, and adverse environmental conditions, certification of these technologies for safety-critical flight systems is challenging. The approach described here is developed under the constraint of maintaining a retrofit architecture thereby allowing the baseline, previously certified, software to remain. The augmented modules serve to assist systems already in place. This paper discusses the development and application of independent and modular adaptive subsystems that are retrofitted to existing guidance and control architectures. The baseline inner-loop control is assisted by a fault detection and isolation (FDI) algorithm tasked with monitoring and reporting control surface health. Similarly, the baseline outer-loop (BOL) guidance is augmented with an adaptive neural network collectively forming an adaptive outer-loop (AOL) guidance system. In general, an adequate inner-loop feedback control law is necessary to faithfully track guidance commands, and various inner-loop control design techniques can provide the requisite performance. Even in the case of degraded inner-loop tracking capability, the outer-loop guidance system may still meet its objectives by inflating inner-loop commands to compensate for inner-loop tracking errors. Nonetheless, faithful innerloop tracking is desirable to reduce the burden on other subsystems, and this is especially the case for piloted

A robust Principal Component Analysis (PCA) model-based approach to Fault Detection and Isolation (FDI) was developed for the Helical Coil Steam Generator (HCSG) systems of the International Reactor Innovative and Secure (IRIS) nuclear reactor. The PCA model was developed using the data generated from a simulation of the system dynamics. Because all the operation modes can be excited through well-designed simulations, a PCA model developed from simulation data is more robust to changes in operation conditions than a PCA model developed from historical data collected during routine operation conditions. In order to deal with model uncertainty that exists in a simulation model, the mismatch between PCA model predictions and online plant measurements was analyzed to characterize the model uncertainty. Based on the assumption that model uncertainty has a structured property, a complete algorithm for the PCA-based FDI was derived such that the FDI results are robust to model uncertainty. When this new method was applied to the IRIS HCSG systems, the results showed that this approach could avoid false alarms and fault misdiagnosis due to changes in operation conditions and model certainty. The approach would overcome the limitations of traditional data-driven model based FDI when routine operational data do not contain sufficient information, and the limitations of physical model based FDI when these models are too complicated for a direct use in FDI design and contain model uncertainty.

Modeling and fault diagnosis design for HVAC systems using recurrent neural networks

Linguistic OWA and two time-windows based fault identification in wide plants

To monitor the dynamics and non-stationarity inherent in industrial processes, we propose a novel incipient fault detection and isolation scheme grounded in a probabilistic perspective, using the Cauchy–Schwarz (CS) divergence. Our innovation lies in the utilization of marginal CS divergence for incipient fault detection and the conditional CS divergence for fault isolation. This approach neither require prior parametric assumptions about the underlying data distribution nor depend on historical fault data, while simultaneously providing explanatory diagnostics. Beyond this, we develop a change point detection-base diagnosis technique for practical engineering applications. This online process monitoring technique guarantees timely intervention to uphold process stability and safety. We demonstrate the compelling performance, higher detection rate and lower alarm rate, of the CS divergence over prevalent divergence-based approaches, such as Kullback–Leibler divergence, Wasserstein distance and Mahalanobis distance. We also illustrate the explanatory insights offered by conditional CS divergence in fault isolation on synthetic data, benchmarks of continuous stirred-tank reactor process and continuous stirred-tank heater process and even a real-world continuous catalytic reforming process. Code of this CS divergence based fault detection and isolation is available at https://github.com/Feiya-Lv/Incipient-Fault-Detection-and-Isolation-with-Cauchy--Schwarz-Divergence.

This paper describes a new method for detecting sensor faults using sliding mode observers. In general, required capabilities of FDI(Fault Detection and Isolation) filters are sensitivity of occurring faults and robustness against disturbances. A FDI filter that proposed in this paper can estimate faults even if disturbances exist. Moreover, we discretize continuous-time FDI filters into discrete-time FDI filters using difference equations and apply the FDI filter to a vibration experimental system. An example demonstrates the effectiveness of the proposed method in comparison with conventional methods.

The brake system is a key system for the safe operation and stopping of trains. As a core component of brake systems, the brake actuation unit (BAU) is essential for slowing down or stopping trains, and faults in the BAU will affect the safety and efficiency of train operation. In order to detect and locate faults in the BAU in time, a fault detection and isolation (FDI) strategy, based on mutual residuals (MRs), principal component analysis (PCA) and improved reconstruction-based contribution plots (IRBCP), was proposed. Firstly, the structural composition and working principle of the BAU were introduced, and its typical failure modes and effects were analyzed. Secondly, considering that the fault detection threshold is not easy to determine due to the variable operating conditions of the BAU, the steady-state fault feature based on MR was extracted. Thirdly, fault detection and isolation were realized based on PCA and the IRBCP algorithm. Finally, by using the fault injection method, case studies on test-rig experiment data of brake systems were conducted; the fault detection rate was 99% and the effectiveness of the proposed strategy was validated by the test data. The proposed strategy shows fast computing ability, and is suitable for systems with dynamic time-varying and nonlinear characteristics.

DC microgrids are becoming an attractive technology for future power grids. However, the fault detection and isolation (FDI) schemes for dc microgrids are still in the nascent state. This article presents an FDI method for dc microgrids using the regularity information of the second derivative of current (SDOC). As the major contribution of this article, the coherence between the singular feature in the SDOC and the short-circuit fault in dc lines is first proved and applied to FDI. Furthermore, a singularity detection approach using stationary wavelet transform (SWT) is introduced. With this FDI method, not only the fault isolation but also the fault-type classification can be achieved based on only local current measurements. Moreover, compared with other local current-based methods, this method has a distinct advantage in the robustness against the nonfault disturbances. The effectiveness of this FDI method was verified through hardware tests under the real-time (RT) simulation of various fault scenarios in a 5-kV three-terminal dc microgrid model based on dual-active-bridge (DAB) converters. This FDI method can be generalized to dc microgrids with different topologies and converters.

A combined data‐driven and observer‐design methodology for fault detection and isolation (FDI) in hybrid process systems with switching operating modes is proposed. The main contribution is to construct a unified framework for FDI by integrating Gaussian mixture models (GMM), subspace model identification (SMI), and results from unknown input observer (UIO) theory. Initially, a GMM is built to identify and describe the multimodality of hybrid systems using the recorded input/output process data. A state‐space model is then obtained for each specific operating mode based on SMI if the system matrices are unknown. An UIO is designed to estimate the system states robustly, based on which the fault detection is laid out through a multivariate analysis of the residuals. Finally, by designing a set of unknown input matrices for specific fault scenarios, fault isolation is performed through the disturbance‐decoupling principle from the UIO theory. A significant benefit of the developed framework is to overcome some of the limitations associated with individual model‐based and data‐based approaches in dealing with the problem of FDI in hybrid systems. Finally, the validity and effectiveness of the proposed monitoring framework are demonstrated using a numerical example, a simulated continuous stirred tank heater process, and the Tennessee Eastman benchmark process. © 2014 American Institute of Chemical Engineers AIChE J, 60: 2805–2814, 2014