Fault Detection Index Research Articles

A novel multimode process monitoring method integrating LDRSKM with Bayesian inference

A local discriminant regularized soft k-means (LDRSKM) method with Bayesian inference is proposed for multimode process monitoring. LDRSKM extends the regularized soft k-means algorithm by exploiting the local and non-local geometric information of the data and generalized linear discriminant analysis to provide a better and more meaningful data partition. LDRSKM can perform clustering and subspace selection simultaneously, enhancing the separability of data residing in different clusters. With the data partition obtained, kernel support vector data description (KSVDD) is used to establish the monitoring statistics and control limits. Two Bayesian inference based global fault detection indicators are then developed using the local monitoring results associated with principal and residual subspaces. Based on clustering analysis, Bayesian inference and manifold learning methods, the within and cross-mode correlations, and local geometric information can be exploited to enhance monitoring performances for nonlinear and non-Gaussian processes. The effectiveness and efficiency of the proposed method are evaluated using the Tennessee Eastman benchmark process.

Reconstruction-based contribution approaches for improved fault diagnosis using principal component analysis

This paper provides two new proposed data-based fault diagnosis approaches using the principal component analysis (PCA). Since faults are really complex and may be in multidimensional directions, the first proposal is a generalized RBC method. The theoretical diagnosability analysis using this one guarantees correct fault diagnosis in the case of complex faults that have large magnitudes. Nevertheless, the common assumption resulting from the use of the contribution methods in general is that variables with largest contributions to the fault detection indices are more likely to be the root causes of the fault occurrence. To handle the more complex faults and remedy the defective of the RBC method, the second proposal presents an alternative approach called RBC ratio (RBCR). A theoretical diagnosability analysis testifies to its strong performance in identifying the detectable faults. Indeed, the isolation of fault is guaranteed once its magnitude has satisfied a sufficient condition for isolability. These proposed approaches have been successfully applied to a numerical example as well as the CSTR process.

Analyzing Threefold Schemes for Enhancing Communication Channel Efficiencies Using IP Multimedia Server–Client Systems for LTE Networks

In 3G, the server platform of IP multimedia subsystem (IMS) offers multimedia service to the multimedia clients through the exchange of authenticated key agreement (AKA) scheme. Since the IMS service is transmitted over public networks, the server and client of IMS shares a common session-key via the execution of AKA scheme to secure the communication service over a public network. Though there are some identical steps involved in the key authentication Scheme of IMS, the communication efficiencies of IMS server and client are still conciliating. Besides, it is having the issues of insecurity and session compatibility dealt with the real time transport protocol (RTP). To resolve the issues of IMS, this paper proposes three novel schemes: (1) The first scheme is multimedia server---client authentication scheme and its purpose is to mitigate the packet congestion on the networks; (2) The second scheme is traffic on background and its purpose is to analyze the signal congestion, bandwidth consumption and network utilization of the system; and (3) The last one is fault detection index and its purpose is to prevent the SIP vulnerabilities, like bye and invite. In addition, we integrate the proposed and existing authentication schemes, such as EPS-AKA, AP-AKA, UMTS-AKA, X-AKA, S-AKA and so on., in the IMS server and client environments to testify the metrics, namely round trip time, call response time, packet generation, flow analysis, RTP session, network utilization, throughput analysis, bandwidth utilization and flooding attack detection realistically.

Quality-related prediction and monitoring of multi-mode processes using multiple PLS with application to an industrial hot strip mill

This paper is focused on the development of the quality prediction and fault detection schemes for the industrial hot strip mill process (HSMP). Considering that the pure partial least squares (PLS) model is based on the assumption of a single operating mode, in this paper, a multiple PLS based method is proposed. The new method address the multi-mode problem in HSMP with the Gaussian mixture model (GMM), then the advantage of original PLS is subsequently followed to achieve the quality prediction and monitoring goals. Meanwhile, a new probabilistic fault detection index called quality-related fault probability index is also developed for the fault detection purpose. Finally, the whole proposed scheme is exercised with the real industrial data, and performances are evaluated by comparing with other existing methods.

Multimode Process Monitoring Based on Sparse Principal Component Selection and Bayesian Inference-Based Probability

According to the demand for diversified products, modern industrial processes typically have multiple operating modes. At the same time, variables within the same mode often follow a mixture of Gaussian distributions. In this paper, a novel algorithm based on sparse principal component selection (SPCS) and Bayesian inference-based probability (BIP) is proposed for multimode process monitoring. SPCS can be formulated as a just-in-time regression between all PCs and each sample. SPCS selects PCs according to the nonzero regression coefficients which indicate the compact expression of the sample. This expression is necessarilydiscriminative: amongst all subset of PCs, SPCS selects the PCs which most compactly express the sample and rejects all other possible but less compact expressions. BIP is utilized to compute the posterior probabilities of each monitored sample belonging to the multiple components and derive an integrated global probabilistic index for fault detection of multimode processes. Finally, to verify its superiority, the SPCS-BIP algorithm is applied to the Tennessee Eastman (TE) benchmark process and a continuous stirred-tank reactor (CSTR) process.

A wavelet-based index for fault detection and its application in condition monitoring of helicopter drive-train components

This paper presents a new condition indicator using wavelet analysis for the purpose of fault detection in an AH–64 gearbox. Historically, vibration–based condition indicators from employed component monitoring equipment are derived from both temporal and spectral domain analysis. However, these indicators failed to accurately capture high order correlations for the gearbox study addressed in this paper. An improved approach is necessary to overcome limitations of traditional vibrational monitoring techniques. The proposed condition indicator is derived from the Morlet continuous wavelet .transform The power spectra obtained from the wavelet transform coefficients at a certain scale or frequency are added together and then are normalised to one composite signal, denoted by a numeric index. Concepts of the wavelet index are discussed. This index is applied using real–world vibration data from a tail rotor gearbox with an output seal leak as part of condition–based maintenance practices. Results demonstrate potential of the proposed wavelet index to more effectively capture the fault when compared to gearbox condition indicators. [Received 24 January 2014; Revised 28 August 2014; Accepted 3 October 2014]

Observer‐based open transistor fault diagnosis and fault‐tolerant control of five‐phase permanent magnet motor drive for application in electric vehicles

To meet increasing demand for higher reliability in power electronics converters applicable in electric vehicles, fault detection (FD) is an important part of the control algorithm. In this study, a model-based open transistor fault diagnsosis method is presented for a voltage-source inverter (VSI) supplying a five-phase permanent magnet motor drive. To realise this goal, a model-based observer is designed to estimate model parameters. After that, the estimated parameters are used to design a sliding mode observer in order to estimate the phase current in an ideal model. Subsequently, the proposed FD technique measures the similarity between the estimated current and real current using cross-correlation factor. This factor is used for the first time in this study to define a FD index in VSI. The presented FD scheme is simple and fast; also, it is able to detect multiple open switch or open phase faults in contrast to conventional methods. On the other side, in order to track reference current of the motor, the estimated parameters are used to design a proportional resonant controller. The FD technique is used to operate a multiphase fault-tolerant brushless direct current (BLDC) motor drive. Experimental results on a five-phase BLDC motor with in-wheel outer rotor applicable in electrical vehicles are conducted to validate the theory.

ASCS online fault detection and isolation based on an improved MPCA

Multi-way principal component analysis (MPCA) has received considerable attention and been widely used in process monitoring. A traditional MPCA algorithm unfolds multiple batches of historical data into a two-dimensional matrix and cut the matrix along the time axis to form subspaces. However, low efficiency of subspaces and difficult fault isolation are the common disadvantages for the principal component model. This paper presents a new subspace construction method based on kernel density estimation function that can effectively reduce the storage amount of the subspace information. The MPCA model and the knowledge base are built based on the new subspace. Then, fault detection and isolation with the squared prediction error (SPE) statistic and the Hotelling (T 2) statistic are also realized in process monitoring. When a fault occurs, fault isolation based on the SPE statistic is achieved by residual contribution analysis of different variables. For fault isolation of subspace based on the T 2 statistic, the relationship between the statistic indicator and state variables is constructed, and the constraint conditions are presented to check the validity of fault isolation. Then, to improve the robustness of fault isolation to unexpected disturbances, the statistic method is adopted to set the relation between single subspace and multiple subspaces to increase the corrective rate of fault isolation. Finally fault detection and isolation based on the improved MPCA is used to monitor the automatic shift control system (ASCS) to prove the correctness and effectiveness of the algorithm. The research proposes a new subspace construction method to reduce the required storage capacity and to prove the robustness of the principal component model, and sets the relationship between the state variables and fault detection indicators for fault isolation.

Method for computing efficient electrical indicators for offshore wind turbine monitoring

Offshore wind turbines availability is an important issue if such wind farms are to be considered a reliable source of renewable energy for the future. Environmental conditions and the low accessibility of such wind farms have contributed to the decrease of the availability of the wind turbines, compared to the onshore ones. In order to improve the reliability, condition monitoring systems and the implementation of scheduled maintenance strategies are a must for offshore power plants. This paper proposes a method of computing efficient electrical indicators using the available three-phase electrical quantities. These indicators are then to be used to obtain fault indicators for fault detection and diagnosis. The electrical indicators are obtained by using the instantaneous symmetrical components decomposition, a well proven method in power networks design and diagnosis. The new quantities are able to fully describe the whole electrical system and provide an effective mean to quantify the balance and unbalance in the system. The method uses the electrical three-phase quantities measured at the output of the generator in a wind turbine to obtain the indicators. The performance of this method is illustrated using both synthetic and experimental data.

A probabilistic approach for data-driven fault isolation in multimode processes

This paper addresses the problem of fault isolation in processes which are working in different operating points. Due to nonlinearities and set-point changes, the statistical model which is obtained from data is different from one operating point to another. Therefore the classical multivariate statistical process monitoring approaches may not be suitable for monitoring and diagnosis purposes. For that, a data-driven fault isolation method is proposed which splits the process into several local models. Based on the local models, a probabilistic approach is proposed to determine the contribution of each variable to the fault detection index and find the risky variables which are responsible for the fault. Finally, the proposed method is demonstrated through its application on a laboratory setup of continuous stirred tank heater.

Relationships between PCA and PLS-regression

This work aims at comparing several features of Principal Component Analysis (PCA) and Partial Least Squares Regression (PLSR), as techniques typically utilized for modeling, output prediction, and monitoring of multivariate processes. First, geometric properties of the decomposition induced by PLSR are described in relation to the PCA of the separated input and output data (X-PCA and Y-PCA, respectively). Then, analogies between the models derived with PLSR and YX-PCA (i.e., PCA of the joint input–output variables) are presented; and regarding to process monitoring applications, the specific PLSR and YX-PCA fault detection indices are compared. Numerical examples are used to illustrate the relationships between latent models, output prediction models, and fault detection indices. The three alternative approaches (PLSR, YX-PCA and Y-PCA plus X-PCA) are compared with regard to their use for statistical modeling. In particular, a case study is simulated and the results are used for enhancing the comprehension of the PLSR properties and for evaluating the discriminatory capacity of the fault detection indices based on the PLSR and YX-PCA modeling alternatives. Some recommendations are given in order to choose the more appropriate approach for a specific application: 1) PLSR and YX-PCA have similar capacity for fault detection, but PLSR is recommended for process monitoring because it presents a better diagnosing capability; 2) PLSR is more reliable for output prediction purposes (e.g., for soft sensor development); and 3) YX-PCA is recommended for the analysis of latent patterns imbedded in datasets.

Research on PCA Based Sensor Fault Detection and its Application

This paper deals with problem related to fault detection in complex process with various sorts of sensors. Not model based but data-driven multivariate statistical process monitoring approach PCA is proposed, fault detection indices statistic and square prediction error are discussed and their complementary relationship is also presented. Availability and reliability of the method proposed in this paper is verified by experimental example.

Hybrid modelling of doubly fed induction generators with inter‐turn stator fault and its detection method using wavelet analysis

Stator faults contribute to 38% of all electrical machine failures. The detection of inter-turn stator fault has therefore been the focus of research for many years. The aim of this research is to provide new condition monitoring technique for doubly fed induction generator during inter-turn fault. The contribution of this research is to formulate a method of doubly-fed induction generator (DFIG) inter-turn stator fault detection that is applicable when this machine operates during steady-state condition. New proposed hybrid model (ABC/dq) has been presented for modelling of DFIG with inter-turn stator fault. Finally a new index has been defined for the detection of inter-turn stator fault in DFIG that employs wavelet analysis of the stator currents and energy of a signal which has been called fault detection index. Proposed index is robust to variation of operating point, percentage of turn fault, stator and rotor resistance and sampling frequency.

Improving mesh stiffness calculation of cracked gears for the purpose of vibration-based fault analysis

For the purpose of vibration-based condition monitoring and to prevent occurrence of catastrophic gear failures it is important to improve the simulated dynamic response of the studied gear model. The time-varying gear mesh stiffness will contribute to the dynamic response of a geared system. Some previously applied methods for stiffness calculation, described in the literature, show good agreement with the results obtained with FEM simulation for smaller crack sizes. However, when larger crack sizes are reached, these methods show an increasing deviation from FEM simulation results. A reduction in the gear mesh stiffness can be considered to assess the status of tooth damage and, therefore, by increasing the accuracy of the calculated mesh stiffness, dynamic simulations of a gear can be improved. In this paper a new method is presented for calculating the gear mesh stiffness for a propagating crack in the tooth root. The influence of gear mesh stiffness on the vibration-based fault detection indicators, the RMS, kurtosis and the crest factor, is investigated. Different crack sizes are examined by using this new method for sizes up to around 50% of the total tooth root thickness. When compared to FEM simulations, the presented method shows more accurate results for calculations of the gear mesh stiffness (for the studied model) than the previously suggested methods.

Vibration signal analysis for gear fault diagnosis with various crack progression scenarios

There are different analytical scenarios assumed for crack propagation in the gear tooth root. This paper presents an investigation of the performance of statistical fault detection indicators (the RMS and kurtosis) for three different series of crack propagation scenarios, to compare these scenarios from a fault diagnostics point of view. These scenarios imply different forms of cracks with propagation by a certain step of crack depth. The first scenario assumes a crack being extended through the whole tooth width with a uniform crack depth distribution, while the second scenario assumes the crack being extended through the whole tooth width with a parabolic crack depth distribution, and finally in the third scenario the crack is assumed to be propagating in both the depth and the length directions simultaneously. The time-varying gear mesh stiffness has been investigated using the programme code developed in the present research, and the crack propagation can be modelled with any of the presented crack propagation scenarios. Dynamic simulation has been performed to obtain the residual signals of all the studied cases for each crack propagation scenario. The comparison of the statistical indicators applied to the residual signals shows that in the first scenario the faults are most easily detectable, since in this scenario there is a change in the indicators implying a dramatic decrease in the gear mesh stiffness. The fault detection in the 2nd scenario is more difficult, as the crack propagates with no significant reflection on the mesh stiffness loss. The 3rd proposed scenario should receive more attention in research because it could occur in reality in case of non-uniform load distribution. However, with this scenario it is difficult to perform early fault detection, since there is a very slight change in the statistical indicators at the beginning of the crack propagation. After which, these indicators show a significant change when the crack grows deeper which implies a serious crack propagation condition.

An adaptive threshold estimation scheme for abrupt changes detection algorithm in a cement rotary kiln

This work deals with a major problem that arises when searching for a reliable, accurate and easily exploitable adaptive threshold based fault detection technique to indicate with great accuracy the presence of a fault in a cement rotary kiln system. The adaptive threshold estimation scheme for detection of faults is developed through the mean and variance of the measured signals. Squared radii, which represent the sum of squares of these statistical parameters, are obtained. This fault detection index is calculated using several repeated experiments under the same operation conditions. At each sample time and for all the experiments, the confidential interval of instantaneous squared radii is closely related to its estimated probability density function. Jarque–Bera hypothesis testing is performed at a typical error risk level and confirms the obtained probability distribution law. Several significance levels are considered where the limitations of fixed thresholding and the performance of the proposed adaptive thresholding procedure are demonstrated respectively through the rate of false alarms. The proposed adaptive threshold is compared with the fixed threshold by evaluating the detection performance across various types of faults in a cement rotary kiln system.

Quality‐relevant and process‐relevant fault monitoring with concurrent projection to latent structures

This paper proposes a new concurrent projection to latent structures is proposed in this paper for the monitoring of output‐relevant faults that affect the quality and input‐relevant process faults. The input and output data spaces are concurrently projected to five subspaces, a joint input‐output subspace that captures covariations between input and output, an output‐principal subspace, an output‐residual subspace, an input‐principal subspace, and an input‐residual subspace. Fault detection indices are developed based on these subspaces for various fault detection alarms. The proposed monitoring method offers complete monitoring of faults that happen in the predictable output subspace and the unpredictable output residual subspace, as well as faults that affect the input spaces only. Numerical simulation examples and the Tennessee Eastman challenge problem are used to illustrate the effectiveness of the proposed method. © 2012 American Institute of Chemical Engineers AIChE J, 59: 496–504, 2013

Sensor Fault Detection, Diagnosis and Validation - A Survey

This paper provides an overview and analysis of data-driven sensor fault detection, diagnosis and validation from the application viewpoint. The typical sensor fault detection indices in the literature and the fundamental issues of necessary and sufficient conditions for detectability, reconstructability, identifiability and isolatability are analyzed. The main objective is to study the essential and important algorithms and techniques for single or multiple sensor fault diagnosis and validation. The issues of optimal principal components, sensor validity index, maximized sensitivity, as well as robust sensor fault diagnosis, etc. are discussed. Additional focuses are summarized at the end of the paper for future investigation.

A New and Effective Method of Bearing Fault Diagnosis Using Wavelet Packet Transform Combined with Support Vector Machine

After briefly analyzing past research, by wavelet packet transform with support vector machine (SVM), a new method of bearing fault diagnosis is presented. Wavelet packets have greater decor relation properties than standard wavelets in that they induce a finer partitioning of the frequency domain of the process generating the data. we analyze the vibration features of testing signals of a bearing system in different running conditions by wavelet de-noising with thresholds; we decompose the feature signals into different frequency bands with the wavelet packet transform (WPT) and then calculate the energy percentage of every frequency band component to obtain its fault detection index used for fault diagnosis by the support vector machine (SVM). We analyze the vibration features of testing signals of a bearing system in different running conditions by wavelet de-noising with thresholds; and to decompose the feature signals into different frequency bands with the wavelet packet transform (WPT), through wavelet packet transform to obtain wavelet coefficients and then Energy eigenvector of frequency domain are extracted by using Shannon entropy principle. Subsequently, the extracted Energy eigenvector of frequency domain are applied as inputs to support vector machine（SVM）for bearing from internal fault. Fault state of bearing is identified by using radial basis function genetic-support vector machine. What is worth mentioning in particular is that our method can also effectively diagnose compound faults.

Dynamic latent variable modeling for statistical process monitoring

AbstractDynamic principal component analysis (DPCA) has been widely used in the monitoring of dynamic multivariate processes. In traditional DPCA, the dynamic relationship between process variables are implicit and hard to interpret. To extract explicit latent factors that are dynamically correlated, a new dynamic latent variable model is proposed. The new structure can improve modeling of dynamic data and enhance the process monitoring performance. Fault detection indices are developed based on the proposed model. A case study is given to illustrate the effectiveness of the proposed new dynamic factor model.