Incipient Fault Detection and Identification in Process Systems Using Accelerated Neural Network Learning

This paper presents an approach to classifying power system faults using rough set methods. A knowledge-based fault detection and identification (FDI) system for power system faults has been introduced. The FDI system has the ability to detect and classify power system faults by combining conventional signal analysis methods (e.g., FFT, IFFT and wavelets) with granular computing and rough set methods. In granular computing, experimental data is partitioned into collections of data (called information granules) that are in some way similar. Rough set methods are based on set approximation, partition of each finite universe using an indiscernibility relation, attribute reduction, decision-rule derivation, and many useful measures such as approximation accuracy and rough inclusion. Traditional fuzzy set theory is also as part of fault signal feature extraction. The FDI system derives an indication of the type of faults that have occurred and also generates classification rules for the fault classification. This system has resulted from a study of fault files recorded by the Transcan Recording System (TRS) at the Manitoba Hydro Dorsey Station over several years. The contribution of this paper is the introduction of an approach to classifying power system faults using a combination of traditional signal analysis methods and a number of computational intelligence methods (granular computing, and rough set theory).

A fuzzy-based gas turbine fault detection and identification system for full and part-load performance deterioration

PurposeThis paper aims to analyze and investigate the performance of an improved fault detection and identification (FDI) method based on multiple criteria, applied to six-switch three-phase inverter (SSTPI)-fed induction motor (IM) drives under both single and multiple open insulated-gate bipolar transistors(IGBT) faults.Design/methodology/approachThis paper proposes an advanced diagnostic method for both single and multiple open IGBT faults dedicated to SSTPI-fed IM drives considering five distinct faulty operating conditions as follows: a single IGBT open-circuit fault, a single-phase open-circuit fault, a non-crossed double fault in two different legs, a crossed double fault in two different legs and a three-IGBT open-circuit fault. This is achieved because of the introduction of a new diagnosis variable provided using the information of the slope of the current vector in (α-β) frame. The proposed FDI method is based on the synthesis and the analysis, under both healthy and faulty operations, of the behaviors of the introduced diagnosis variable, the three motor phase currents and their normalized average values. Doing so, the developed FDI method allows a best compromise of fast detection and precision localization of IGBT open-circuit fault of the inverter.FindingsSimulation works, carried out considering the implementation of the direct rotor flux oriented control in an IM fed by the conventional SSTPI, have proved the high performance of the advanced FDI method in terms of fast fault detection associated with a high robustness against false alarms, against speed and load torque fast variations and against the oscillations of the DC-bus voltage in the case of both healthy and faulty operations.Research limitations/implicationsThis work should be extended considering the validation of the obtained simulation results through experiments.Originality/valueDifferent from other FDI methods, which suffer from a low diagnostic effectiveness for low load levels and false alarms during transient operation, this method offers the potentialities to overcome these drawbacks because of the introduction of the new diagnosis variable. This latter, combined with the information provided from the three motor phase currents and their normalized average values allow a more efficient detection and identification of IGBT open-circuit fault.

Next generation aircraft require the development and integration of a deal of innovative technologies to meet the ambitious sustainability goals set for aviation. This transformational effort is associated with a tremendous increase of the complexity of the onboard systems and their multiphysics coupled behaviours. A critical aspect relates to the characterization of the coupled fault modes resulting from the integration of these novel technologies, which introduce identifiability issues and demand new approaches for the efficient identification of (multimodal) non-nominal conditions. Model-based fault detection and identification (FDI) has been proven essential to infer onboard systems' health from signal acquisitions, but existing methods are too expensive and fail to capture incipient faults in presence of multimodality which prevent scaling to complex multiphysics systems. This work introduces a multifidelity framework to accelerate the identification of fault modes affecting complex systems. An original two-stage compression computes an optimally informative and highly reduced representation of the monitoring signals for the minimum demand of onboard resources. A multifidelity scheme for Bayesian inversion is developed to infer multidomain fault parameters from the compressed signals: variable cost and fidelity models are optimally queried for a major reduction of the overall computational expense. The framework is demonstrated and validated for aerospace electromechanical actuators affected by incipient multimodal faults. Remarkable accelerations of the FDI procedure are observed and the exact identification of the incipient fault condition achieved one order of magnitude faster than with standard algorithms.

This paper presents a method to increase the reliability of Unmanned Aerial Vehicle (UAV) sensor Fault Detection and Identification (FDI) in a multi-UAV context. Differential Global Positioning System (DGPS) and inertial sensors are used for sensor FDI in each UAV. The method uses additional position estimations that augment individual UAV FDI system. These additional estimations are obtained using images from the same planar scene taken from two different UAVs. Since accuracy and noise level of the estimation depends on several factors, dynamic replanning of the multi-UAV team can be used to obtain a better estimation in case of faults caused by slow growing errors of absolute position estimation that cannot be detected by using local FDI in the UAVs. Experimental results with data from two real UAVs are also presented.

Incipient fault detection in transformers can provide early warning of electrical failure and could prevent catastrophic losses. To develop transformer incipient fault detection technique, a transformer model to simulate internal incipient faults is required. This paper presents a methodology to model internal incipient winding faults in distribution transformers. These models were implemented by combining deteriorating insulation models with an internal short circuit fault model. The internal short circuit fault model was developed using finite element analysis. The deteriorating insulation model, including an aging model and an arcing model connected in parallel, was developed based on the physical behavior of aging insulation and the arcing phenomena occurring when the insulation was severely damaged. The characteristic of the incipient faults from the simulation were compared with those from some potential experimental incipient fault cases. The comparison showed the experimentally obtained characteristic's of terminal behavior of the faulted transformer were similar to the simulation results from the incipient fault models.

Fault Detection and Identification Using Neural Network and Fuzzy Logic

This article presents a sufficient comparison of two types of advanced non-parametric classifiers implemented in remote sensing for land cover classification. A SPOT-5 HRG image of Yanqing County, Beijing, China, was used, in which agriculture and forest dominate land use. Artificial neural networks (ANNs), including the adaptive backpropagation (ABP) algorithm, Levenberg–Marquardt (LM) algorithm, Quasi-Newton (QN) algorithm and radial basis function (RBF) were carefully tested. The LM–ANN and RBF–ANN, which outperform the other two, were selected to make a detailed comparison with support vector machines (SVMs). The experiments show that those well-trained ANNs and SVMs have no significant difference in classification accuracy, but the SVM usually performs slightly better. Analysis of the effect of the training set size highlights that the SVM classifier has great tolerance on a small training set and avoids the problem of insufficient training of ANN classifiers. The testing also illustrates that the ANNs and SVMs can vary greatly with regard to training time. The LM–ANN can converge very quickly but not in a stable manner. By contrast, the training of RBF–ANN and SVM classifiers is fast and can be repeatable.

Next-generation aircraft require the development and integration of a deal of innovative green technologies to meet the ambitious sustainability goals set for aviation. Those transformational efforts are associated with a tremendous increase in the complexity of the onboard systems and their multiphysics-coupled behaviors and dynamics. A critical aspect relates to the identification of the coupled faults resulting from the integration of those green technologies, which introduce damage identifiability issues and demand new approaches for the efficient and accurate identification of non-nominal fault conditions. Model-based fault detection and identification (FDI) methodologies have been proven essential to identify onboard the damages affecting the systems from physical signal acquisitions, but existing methods are typically computationally expensive and fail to capture incipient coupled faults, which prevents their adoption and scaling with the increasing complexity of novel multiphysics systems. This work introduces a multifidelity framework to accelerate the identification of fault modes affecting complex systems. An original two-stage compression computes an optimally informative and highly reduced representation of the monitoring signals for the minimum demand of onboard resources. A multifidelity scheme for Bayesian inversion is developed to infer multidomain fault parameters from the compressed signals: variable cost and fidelity models are optimally queried for a major reduction of the overall computational expense. The framework is demonstrated and validated for aerospace electromechanical actuators affected by incipient multimodal faults. Remarkable accelerations of the FDI procedure are observed, and the exact identification of the incipient fault condition was achieved one order of magnitude faster than with standard algorithms.

This paper addresses the problem of incipient fault detection and isolation for a class of Lipschitz nonlinear systems with multiple sensor biases and explores further results of total measurable fault information residual (ToMFIR). Firstly, state and output transformations are introduced to transform the original system into two subsystems. Since the effects of system disturbances on the residuals are completely decoupled, the incipient faults in the second subsystem can be detected by constructing ToMFIR. Further, sliding-mode observers (SMOs) based fault isolation scheme is designed to guarantee the isolation of multiple sensor faults even if they occur at the same time, which is the main contribution of this paper. Finally, application results conducted on a CRH2 high-speed railway traction motor are given to illustrate the effectiveness of the proposed approach.

An advanced distance relay strategy for incipient fault detection and location in underground distribution systems

The incipient fault detection of a complex industrial process is a challenging problem for traditional dynamic detection methods. Traditional dynamic detection methods usually decouple the correlations among the variables and dynamic correlations simultaneously, which makes the two types of correlations mixed and may lead to performance deterioration in long-sequence dynamic detection. Some incipient faults may not change the amplitudes of process variables but change the long-sequence dynamic features. Based on the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T^{2}$</tex-math> </inline-formula> statistic and matrix multiplication transformation ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T^{2}$</tex-math> </inline-formula> S-MMT), traditional dynamic detection methods can detect many faults effectively. However, the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T^{2}$</tex-math> </inline-formula> S-MMT can not effectively detect some incipient faults due to the above two types of correlations mixed. In order to overcome the shortcomings of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T^{2}$</tex-math> </inline-formula> S-MMT and improve the detection ability of some incipient faults, this paper proposes a fault detection method based on the dynamic k-nearest neighbor model and Dual Control Chart (DKNN-DCC), which can improve the incipient fault detection performance by using long-sequence dynamic detection. The proposed method is verified by the Tennessee Eastman (TE) process and the continuously stirred tank reactor (CSTR) process. The experimental results show the effectiveness of the proposed method in incipient fault detection compared with traditional dynamic detection methods. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This paper presents a novel incipient fault detection method, which directly mines the long-sequence dynamic abnormal information from the process variable and overcomes the problem of some abnormal information being submerged in the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$T^{2}$</tex-math> </inline-formula> statistic calculated based on the matrix multiplication transformation. The proposed method can detect incipient faults that are not easily detected by some traditional methods and can help operators find the abnormal and avoid more serious losses. The structure of the proposed method jumps out of the frameworks of traditional dynamic detection methods, which is feasible to apply to different stable industrial processes.

Coventional control schemes are developed under the assumption that the sensors and actuators are free from faults. However, the occurrence of faults will cause degradation in the closed-loop performance and also will have an impact on safety, productivity, and plant economy. In the present work, we have proposed a fault-tolerant control scheme (FTCS) by integrating a fault detection and identification (FDI) technique with conventional control. The principal component of our proposed FTCS is a compensation strategy (supervisory system) which uses the information provided by the FDI to appropriately modify the controller as well as the model used in FDI. This allows online application of the FTCS without causing significant degradation in the closed-loop performance due to the occurrences of biases in sensors and actuators or due to changes in unmeasured disturbance variables and due to moderate change in process parameters. Through stochastic simulation studies of a continuous stirred tank reactor process, we demonstrate that our proposed FTCS leads to significant improvement in the closed-loop performance in comparison to a conventional control scheme, especially as the fault magnitude increases. The proposed compensation scheme also allows identification of multiple faults that occur sequentially in time and is also found to be robust with respect to moderate plant−model mismatch.

An approach to sensor fault detection and identification (FDI) in the dead reckoning of mobile robots is proposed. It is based on the interacting multiple-model (IMM) algorithm. Changes of sensor normal/failure modes are explicitly modeled as switching from one mode to another in a probabilistic manner, and mode probabilities and estimates from single-model-based filters, each of which is based on a model matching to a particular system mode, interact with each other effectively. To provide better fault decision, a simple decision rule is incorporated into the IMM based algorithm. The proposed FDI algorithm is implemented on our mobile robot. Sixteen system modes (one normal mode and fifteen 'hard' sensor-failure modes) of four internal sensors (two wheel-encoders, a steering potentiometer and a yaw-rate gyro) are handled. Experimental results illustrate that the dead reckoning with the proposed FDI system allows robust motions of the robot subject to sensor failures.

In chemical processes, there are both nonlinearity and slow dynamics among process variables. Meanwhile, at the beginning of faults, slight variation may not prompt the significant parametric changes. Traditional fault detection methods, based on mean and variation shifts, often fail to achieve satisfactory performance. To improve the detection of incipient quality-related faults, a novel orthonormal slow feature-based quality-related data decomposition network with Jensen-Shannon divergence (OSF-JS) fault detection method is proposed. The orthonormal subspace analysis (OSA) is embedded with Siamese to extract orthonormal slow features (OSFs) from decomposed subspaces. Enhancing the detection of incipient faults, JS divergence combined with a sliding window algorithm is introduced to construct statistics. The effectiveness of OSF-JS is validated for a wastewater treatment process and a real-world chemical process. Results show that the proposed OSF-JS method not only improves the fault detection rate (FDR) and reduces false alarm rate (FAR) but also identifies whether faults occur in the quality-related component.