| 1. |
- Zhang, Liangwei, et al.
(författare)
-
A Review on Deep Learning Applications in Prognostics and Health Management
- 2019
-
Ingår i: IEEE Access. - : IEEE. - 2169-3536. ; 7, s. 162415-162438
-
Forskningsöversikt (refereegranskat)abstract
- Deep learning has attracted intense interest in Prognostics and Health Management (PHM), because of its enormous representing power, automated feature learning capability and best-in-class performance in solving complex problems. This paper surveys recent advancements in PHM methodologies using deep learning with the aim of identifying research gaps and suggesting further improvements. After a brief introduction to several deep learning models, we review and analyze applications of fault detection, diagnosis and prognosis using deep learning. The survey validates the universal applicability of deep learning to various types of input in PHM, including vibration, imagery, time-series and structured data. It also reveals that deep learning provides a one-fits-all framework for the primary PHM subfields: fault detection uses either reconstruction error or stacks a binary classifier on top of the network to detect anomalies; fault diagnosis typically adds a soft-max layer to perform multi-class classification; prognosis adds a continuous regression layer to predict remaining useful life. The general framework suggests the possibility of transfer learning across PHM applications. The survey reveals some common properties and identifies the research gaps in each PHM subfield. It concludes by summarizing some major challenges and potential opportunities in the domain.
|
|
| 2. |
- Zhang, Liangwei, et al.
(författare)
-
An Overview of Deep Learning in Prognostics and Health Management
- 2019
-
Ingår i: 2019 Annual Reliability and Maintainability Symposium (RAMS). - : IEEE.
-
Konferensbidrag (refereegranskat)abstract
- Deep learning has attracted intense interest recently in Prognostics and Health Management (PHM), due to its enormous representing power and capability in automated feature learning. This paper attempts to survey recent advancements of PHM methodologies associated with deep learning. After a brief introduction to several deep learning models, we reviewed and analyzed applications of fault detection, diagnosis and prognosis using deep learning, respectively. The survey reveals that most existing work utilized deep learning to conduct feature learning from unstructured raw data including vibration data, current signals, images and videos. Deep learning provides a general framework for PHM applications: fault detection uses either reconstruction error or stacks a binary classier on top of the network to detect anomalies; fault diagnosis typically adds a soft-max layer to perform multi-class classification; and prognosis adds a continuous regression layer to predict remaining useful life. We further pointed out some challenges and potential opportunities in the field.
|
|
| 3. |
- Zhang, Liangwei, et al.
(författare)
-
End-To-End Unsupervised Fault Detection Using A Flow-Based Model
- 2021
-
Ingår i: Reliability Engineering & System Safety. - : Elsevier. - 0951-8320 .- 1879-0836. ; 215
-
Tidskriftsartikel (refereegranskat)abstract
- Fault detection has been extensively studied in both academia and industry. The rareness of faulty samples in the real world restricts the use of many supervised models, and the reliance on domain expertise for feature engineering raises Other barriers. For this reason, this paper proposes an unsupervised, end-to-end approach to fault detection based on a flow-based model, the Nonlinear Independent Components Estimation (NICE) model. A NICE model models a target distribution via a sequence of invertible transformations to a prior distribution in the latent space. We prove that, under certain conditions, the L2-norm of normal samples’ latent codes in a trained NICE model is Chi-distributed. This facilitates the use of hypothesis testing for fault detection purpose. Concretely, we first apply Zero-phase Component Analysis to decorrelate the data of normal states. The whitened data are fed to a NICE model for training, in a maximum likelihood sense. At the testing stage, samples whose L2-norm of latent codes fail in the hypothesis testing are suspected of being generated by different mechanisms and hence regarded as potential faults. The proposed approach was validated on two datasets of vibration signals; it proved superior to several alternatives. We also show the use of NICE, a type of generative model, can produce real-like vibration signals because of the model's bijective nature.
|
|
| 4. |
- Zhang, Liangwei, et al.
(författare)
-
Intra-Domain Transfer Learning for Fault Diagnosis with Small Samples
- 2022
-
Ingår i: Applied Sciences. - : MDPI. - 2076-3417. ; 12:14
-
Tidskriftsartikel (refereegranskat)abstract
- The concept of deep transfer learning has spawned broad research into fault diagnosis with small samples. A considerable covariate shift between the source and target domains, however, could result in negative transfer and lower fault diagnosis task accuracy. To alleviate the adverse impacts of negative transfer, this research proposes an intra-domain transfer learning strategy that makes use of knowledge from a data-abundant source domain that is akin to the target domain. Concretely, a pre-trained model in the source domain is built via a vanilla transfer from an off-the-shelf inter-domain deep neural network. The model is then transferred to the target domain using shallow-layer freezing and finetuning with those small samples. In a case study involving rotating machinery, where we tested the proposed strategy, we saw improved performance in both training efficiency and prediction accuracy. To demystify the learned neural network, we propose a heat map visualization method using a channel-wise average over the final convolutional layer and up-sampling with interpolation. The findings revealed that the most active neurons coincide with the corresponding fault characteristics.
|
|
| 5. |
- Al-Douri, Yamur K., et al.
(författare)
-
Data clustering and imputing using a two-level multi-objective genetic algorithms (GA) : A case study of maintenance cost data for tunnel fans
- 2018
-
Ingår i: Cogent Engineering. - : Taylor & Francis. - 2331-1916. ; 5:1, s. 1-16
-
Tidskriftsartikel (refereegranskat)abstract
- Data clustering captures natural structures in data consisting of a set of objects and groups similar data together. The derived clusters can be used for scale analysis and to posit missing data values in objects, as missing data have a negative effect on the computational validity of models. This study develops a new two-level multi-objective genetic algorithm (GA) to optimize clustering in order to redact and impute missing cost data for fans used in road tunnels by the Swedish Transport Administration (Trafikverket). The first level uses a multi-objective GA based on fuzzy c-means to cluster cost data objects based on three main indices. The first is cluster centre outliers; the second is the compactness and separation ( ) of the data points and cluster centres; the third is the intensity of data points belonging to the derived clusters. Our clustering model is validated using k-means clustering. The second level uses a multi-objective GA to impute the missing cost redacted data in size using a valid data period. The optimal population has a low , 0.1%, and a high intensity, 99%. It has three cluster centres, with the highest data reduction of 27%. These three cluster centres have a suitable geometry, so the cost data can be partitioned into relevant contents to be redacted for imputing. Our model show better clustering detection and evaluation compared with k-means. The amount of missing data for the two cost objects are: labour 57%, materials 81%. The second level shows highly correlated data (R-squared 0.99) after imputing the missing data objects. Therefore, multi-objective GA can cluster and impute data to derive complete data that can be used for better estimation of forecasting.
|
|
| 6. |
|
|
| 7. |
- Famurewa, Stephen Mayowa, et al.
(författare)
-
Maintenance analytics for railway infrastructure decision support
- 2017
-
Ingår i: Journal of Quality in Maintenance Engineering. - : Emerald Group Publishing Limited. - 1355-2511 .- 1758-7832. ; 23:3, s. 310-325
-
Tidskriftsartikel (refereegranskat)abstract
- PurposeThis purpose of this article is to present a framework for maintenance analytics that is useful for the assessment of rail condition and for maintenance decision support. The framework covers three essential maintenance aspects: diagnostic, prediction and prescription. The article also presents principal component analysis (PCA) and local outlier factor (LOF) methods for detecting anomalous rail wear occurrences using field measurement data.Design/methodology/approachThe approach used in this paper includes a review of the concept of analytics and appropriate adaptation to railway infrastructure maintenance. The diagnotics aspect of the proposed framework is demonstrated with a case study using historical rail profile data collected between 2007 and 2016 for 9 sharp curves on the heavy haul line in Sweden.FindingsThe framework presented for maintenance analytics is suitable for extracting useful information from condition data as required for effective rail maintenance decision support. The findings of the case study include: combination of the two statistics from PCA model (T2 and Q) can help to identify systematic and random variations in rail wear pattern that are beyond normal: the visualisation approach is a better tool for anomaly detection as it categorises wear observations into normal, suspicious and anomalous observations.Practical implicationsA practical implication of this article is that the framework and the diagnostic tool can be considered as an integral part of eMaintenance solution. It can be easily adapted as online or onboard maintenance analytic tool with data from automated vehicle based measurement system.Originality/valueThis research adapts the concept of analytics to railway infrastructure maintenance for enhanced decision making. It proposes a graphical method for combining and visualising different outlier statistics as a reliable anomaly detection tool.
|
|
| 8. |
- Lin, Janet, et al.
(författare)
-
Data analysis of heavy haul wagon axle loads on Malmbanan line, Sweden : A case study for LKAB
- 2016
-
Rapport (övrigt vetenskapligt/konstnärligt)abstract
- The research presented in this report was carried out by Operation and Maintenance Engineering at Luleå University of Technology (LTU) from November 2015 to April 2016. LKAB initiated the research study and provided financial support. The purpose of this research was to support LKAB and Trafikverket in their operational strategy review and optimization of future axle load implementations. It developed five research questions and answered them by analyzing the data for the Malmbanan iron ore train axle loads for 2015.Data analysis comprises four parts. In the first part (section 2), the analysis focuses on axle loads of all loaded trains operating at three different terminals: Kiruna, Malmberget, and Svappavaara. In addition, it examines the differences of three weighing locations in Kiruna, five weighing locations in Malmberget and four weighing locations in Svappavaara (12 weighing locations). Based on these results, the analysis in the second part (section 3) focuses on the heavy haul wagon. Wagon loads are evaluated and predicted for different loading rules (31.0 and 32.5 tons separately). To optimize the current loading rules, the third part of the analysis (section 4) proposes a novel approach to optimize the wagon axle loads: “three sigma prediction”. Under this approach, Kiruna, Malmberget and Svappavaara can set new target loads based on various risk levels. In the fourth and final part of the data analysis (section 5), a comparison study is carried out by collecting axle load data for the test train (with a 32.5 ton axle load) using three different measurement systems in Malmberget, Sävast and Sunderbyn. Finally, sections 6 and 7 summarize the results and make some recommendations for future work. The work presented in this report should give LKAB and Trafikverket a good overview of the load distribution for the ore trains operating on Malmbanan line. It can serve as input into the process of evaluating possible changes in axle load limits. It also gives LKAB a base from which to identify and work with optimization of the various loading places to load trains more efficiently and save costs.
|
|
| 9. |
|
|
| 10. |
- Liu, Bin, et al.
(författare)
-
A Dynamic Maintenance Strategy for Prognostics and Health Management of Degrading Systems : Application in Locomotive Wheel-sets
- 2018
-
Konferensbidrag (refereegranskat)abstract
- This paper develops a dynamic maintenance strategy for prognostics and health management (PHM) of a degrading system. The system under investigation suffers a continuous degradation process, modeled as a Gamma process. In addition to the degradation process, the system is subject to aging, which contributes to the increase of failure rate. An additive model is employed to describe the impact of degradation level and aging on system failure rate. Inspection is implemented upon the system so as to effectively avoid failure. At inspection, the system will be repaired or replaced in terms of the degradation level. Different from previous studies which assume that repair will always lead to an improvement on system degradation, in our study, however, the effect of repair is twofold. It will reduce the system age to 0 but will increase the degradation level. System reliability is analyzed as a first step to serve for the maintenance decision making. Based on the reliability evolution, a maintenance model is formulated with respect to the inspection time. The optimal decision is achieved by minimizing the expected cost rate in one repair cycle. Finally, a case study of locomotive wheel-sets is adopted to illustrate the effectiveness of the proposed model. Our approach incorporates the joint influence of aging and degradation process, and determines the optimal inspection time dynamically, which exhibits the advantage of flexibility and can achieve better performance in field use.
|
|
| 11. |
- Liu, B., et al.
(författare)
-
A Dynamic Prescriptive Maintenance Model Considering System Aging and Degradation
- 2019
-
Ingår i: IEEE Access. - : IEEE. - 2169-3536. ; 7, s. 94941-94943
-
Tidskriftsartikel (refereegranskat)abstract
- This paper develops a dynamic maintenance strategy for a system subject to aging and degradation. The influence of degradation level and aging on system failure rate is modeled in an additive way. Based on the observed degradation level at the inspection, repair or replacement is carried out upon the system. Previous researches assume that repair will always lead to an improvement in the health condition of the system. However, in our study, repair reduces the system age but on the other hand, increases the degradation level. Considering the two-fold influence of maintenance actions, we perform reliability analysis on system reliability as a first step. The evolution of system reliability serves as a foundation for establishing the maintenance model. The optimal maintenance strategy is achieved by minimizing the long-run cost rate in terms of the repair cycle. At each inspection, the parameters of the degradation processes are updated with maximum a posteriori estimation when a new observation arrives. The effectiveness of the proposed model is illustrated through a case study of locomotive wheel-sets. The maintenance model considers the influence of degradation and aging on system failure and dynamically determines the optimal inspection time, which is more flexible than traditional stationary maintenance strategies and can provide better performance in the field.
|
|
| 12. |
- Liu, Biyu, et al.
(författare)
-
Maintenance service strategy for leased equipment : integrating lessor-preventive maintenance and lessee-careful protection efforts
- 2021
-
Ingår i: Computers & industrial engineering. - : Elsevier. - 0360-8352 .- 1879-0550. ; 156
-
Tidskriftsartikel (refereegranskat)abstract
- Lessees may abuse equipment during the lease period since lacking of ownership, thereby increasing lessors’ repair cost and lessees’ downtime losses. This study integrates lessees’ effort to protect leased equipment during the lease period with lessors’ preventive maintenance (PM) into maintenance service strategies. It is proved in a non-cooperative game, neither party achieves the cooperative game’s ideal revenue, but improvement in the lessee’s effort level and lessor’s PM degree can increase the other party’s revenue. A cost-sharing contract model is designed to achieve the maximum revenue as in a cooperative game and ensure Pareto improvement of the leasing parties. In the contract, the lessor grants the lessee a rental discount, and the lessor’s PM cost and lessee’s effort cost are shared with cost-sharing coefficients. Conditions under which the ideal revenue and Pareto improvement can be achieved are discussed. Numerical examples are provided to illustrate the effects of contract parameters, unit penalty on the effort level, and revenue. Managerial insights are finally proposed for leasing parties. The results show: the effect of the effort level and PM degree on equipment failures is marginally diminishing; proposed cost-sharing contract model can achieve the ideal revenue and Pareto improvement; the rental discount has a greater impact on the lessee, while the cost-sharing coefficients have a greater impact on the lessor; and increasing the unit penalty decreases (increases) the lessor’s (lessee’s) revenue but maintains the effort level at constant.
|
|
| 13. |
- Saari, Esi, et al.
(författare)
-
Novel Bayesian Approach to Assess System Availability using a Threshold to Censor Data
- 2019
-
Ingår i: International Journal of Performability Engineering. - : Totem Publisher, Inc.. - 0973-1318. ; 15:5, s. 1314-1325
-
Tidskriftsartikel (refereegranskat)abstract
- Assessment of system availability has been studied from the design stage to the operational stage in various system configurations using either analytic or simulation techniques. However, the former cannot handle complicated state changes, and the latter is computationally expensive. This study proposes a Bayesian approach to evaluate system availability. In this approach: 1) Mean Time to Failure (MTTF) and Mean Time to Repair (MTTR) are treated as distributions instead of being "averaged" to better describe real scenarios and overcome the limitations of data sample size; 2) Markov Chain Monte Carlo (MCMC) simulations are applied to take advantage of the analytical and simulation methods; and 3) a threshold is set up for Time to Failure (TTR) data and Time to Repair (TTR) data, and new datasets with right-censored data are created to reveal the connections between technical and "Soft" KPIs. To demonstrate the approach, the paper considers a case study of a balling drum system in a mining company. In this system, MTTF and MTTR are determined by a Bayesian Weibull model and a Bayesian lognormal model, respectively. The results show that the proposed approach can integrate the analytical and simulation methods to assess system availability and could be applied to other technical problems in asset management (e.g., other industries, other systems). By comparing the results with and without considering the threshold for censoring data, we show the threshold can be used as a monitoring line for continuous improvement in the investigated mining company.
|
|
| 14. |
- Saari, Esi, et al.
(författare)
-
System availability assessment using a parametric Bayesian approach : a case study of balling drums
- 2019
-
Ingår i: International Journal of Systems Assurance Engineering and Management. - : Springer. - 0975-6809 .- 0976-4348. ; 10:4, s. 739-745
-
Tidskriftsartikel (refereegranskat)abstract
- Assessment of system availability usually uses either an analytical (e.g., Markov/semi-Markov) or a simulation approach (e.g., Monte Carlo simulation-based). However, the former cannot handle complicated state changes and the latter is computationally expensive. Traditional Bayesian approaches may solve these problems; however, because of their computational difficulties, they are not widely applied. The recent proliferation of Markov Chain Monte Carlo (MCMC) approaches have led to the use of the Bayesian inference in a wide variety of fields. This study proposes a new approach to system availability assessment: a parametric Bayesian approach using MCMC, an approach that takes advantages of the analytical and simulation methods. By using this approach, mean time to failure (MTTF) and mean time to repair (MTTR) are treated as distributions instead of being “averaged”, which better reflects reality and compensates for the limitations of simulation data sample size. To demonstrate the approach, the paper considers a case study of a balling drum system in a mining company. In this system, MTTF and MTTR are determined in a Bayesian Weibull model and a Bayesian lognormal model respectively. The results show that the proposed approach can integrate the analytical and simulation methods to assess system availability and could be applied to other technical problems in asset management (e.g., other industries, other systems).
|
|
| 15. |
- Shao, Haidong, et al.
(författare)
-
A novel approach of multisensory fusion to collaborative fault diagnosis in maintenance
- 2021
-
Ingår i: Information Fusion. - : Elsevier. - 1566-2535 .- 1872-6305. ; 74, s. 65-76
-
Tidskriftsartikel (refereegranskat)abstract
- Collaborative fault diagnosis can be facilitated by multisensory fusion technologies, as these can give more reliable results with a more complete data set. Although deep learning approaches have been developed to overcome the problem of relying on subjective experience in conventional fault diagnosis, there are two remaining obstacles to collaborative efficiency: integration of multisensory data and fusion of maintenance strategies. To overcome these obstacles, we propose a novel two-part approach: a stacked wavelet auto-encoder structure with a Morlet wavelet function for multisensory data fusion and a flexible weighted assignment of fusion strategies. Taking a planetary gearbox as an example, we use noisy vibration signals from multisensors to test the diagnosis performance of the proposed approach. The results demonstrate that it can provide more accurate and reliable fault diagnosis results than other approaches.
|
|
| 16. |
- Shao, Haidong, et al.
(författare)
-
Compound fault diagnosis for a rolling bearing using adaptive DTCWPT with higher order spectra
- 2020
-
Ingår i: Quality Engineering. - : Taylor & Francis. - 0898-2112 .- 1532-4222. ; 32:3, s. 342-353
-
Tidskriftsartikel (refereegranskat)abstract
- Fault diagnosis plays a vital role in prognostics and health management. Researchers have devoted their efforts in enhancing the accuracy of fault diagnosis. However, diagnosis of compound faults in complex systems is still a challenging task. The problem lies in the coupling of multiple signals, which may conceal the characteristics of compound faults. Taking a rolling bearing as an example, this study aims to boost the accuracy of compound fault diagnosis through a novel feature extraction approach to making the fault characteristics more discriminative. The approach proposes an adaptive dual-tree complex wavelet packet transform (DTCWPT) with higher order spectra analysis. To flexibly and best match the characteristics of the measured vibration signals under analysis, DTCWPT is first adaptively determined by the minimum singular value decomposition entropy. Then, higher order spectra analysis is performed on the decomposed frequency sensitive band for feature extraction and enhancement. The proposed approach is used to analyze experimental signals of a bearing’s compound faults and found effective.
|
|
| 17. |
- Zhang, Liangwei, et al.
(författare)
-
Adaptive Kernel Density-based Anomaly Detection for Nonlinear Systems
- 2018
-
Ingår i: Knowledge-Based Systems. - : Elsevier. - 0950-7051 .- 1872-7409. ; 139:1, s. 50-63
-
Tidskriftsartikel (refereegranskat)abstract
- This paper presents an unsupervised, density-based approach to anomaly detection. The purpose is to define a smooth yet effective measure of outlierness that can be used to detect anomalies in nonlinear systems. The approach assigns each sample a local outlier score indicating how much one sample deviates from others in its locality. Specifically, the local outlier score is defined as a relative measure of local density between a sample and a set of its neighboring samples. To achieve smoothness in the measure, we adopt the Gaussian kernel function. Further, to enhance its discriminating power, we use adaptive kernel width: in high-density regions, we apply wide kernel widths to smooth out the discrepancy between normal samples; in low-density regions, we use narrow kernel widths to intensify the abnormality of potentially anomalous samples. The approach is extended to an online mode with the purpose of detecting anomalies in stationary data streams. To validate the proposed approach, we compare it with several alternatives using synthetic datasets; the approach is found superior in terms of smoothness, effectiveness and robustness. A further experiment on a real-world dataset demonstrated the applicability of the proposed approach in fault detection tasks.
|
|
| 18. |
- Zhang, Liangwei, et al.
(författare)
-
An Angle-based Subspace Anomaly Detection Approach to High-dimensional Data : With an Application to Industrial Fault Detection
- 2015
-
Ingår i: Reliability Engineering & System Safety. - : Elsevier BV. - 0951-8320 .- 1879-0836. ; 142, s. 482-497
-
Tidskriftsartikel (refereegranskat)abstract
- The accuracy of traditional anomaly detection techniques implemented on full-dimensional spaces degrades significantly as dimensionality increases, thereby hampering many real-world applications. This work proposes an approach to selecting meaningful feature subspace and conducting anomaly detection in the corresponding subspace projection. The aim is to maintain the detection accuracy in high-dimensional circumstances. The suggested approach assesses the angle between all pairs of two lines for one specific anomaly candidate: the first line is connected by the relevant data point and the center of its adjacent points; the other line is one of the axis-parallel lines. Those dimensions which have a relatively small angle with the first line are then chosen to constitute the axis-parallel subspace for the candidate. Next, a normalized Mahalanobis distance is introduced to measure the local outlier-ness of an object in the subspace projection. To comprehensively compare the proposed algorithm with several existing anomaly detection techniques, we constructed artificial datasets with various high-dimensional settings and found the algorithm displayed superior accuracy. A further experiment on an industrial dataset demonstrated the applicability of the proposed algorithm in fault detection tasks and highlighted another of its merits, namely, to provide preliminary interpretation of abnormality through feature ordering in relevant subspaces.
|
|
| 19. |
|
|
| 20. |
- Zhang, Liangwei, et al.
(författare)
-
An unsupervised end-to-end approach to fault detection in delta 3D printers using deep support vector data description
- 2024
-
Ingår i: Journal of manufacturing systems. - : Elsevier. - 0278-6125 .- 1878-6642. ; 72, s. 214-228
-
Tidskriftsartikel (refereegranskat)abstract
- Fault detection in 3D printers is crucial for safety and quality assurance, emphasizing proactive prediction over reactive rectification based on manufacturing factors. Presently, most detection techniques rely on shallow models with limited representational capabilities, necessitating manual feature extraction from the captured signals. This manual process is not only cumbersome and potentially costly but often requires intricate domain-specific knowledge. Additionally, these handcrafted features might not optimally distinguish between normal and faulty samples, potentially reducing prediction accuracy. In this study, we introduce an end-to-end approach using the Deep Support Vector Data Description model for fault detection in 3D printers. This design inherently facilitates automatic feature learning, where the features are synergistically optimized for fault detection. Our experiments leverage magnetic field signals for fault detection in 3D printers, using 1D convolutional layers to discern temporal signal patterns and wide kernels in the initial layer to mitigate high-frequency noise. Furthermore, our model can be easily adapted to integrate multi-channel signals for enhanced accuracy. Evaluations on real-world data from a delta 3D printer underscore the superiority of our method compared to existing alternatives.
|
|
| 21. |
- Zhang, Liangwei
(författare)
-
Big Data Analytics for eMaintenance : Modeling of high-dimensional data streams
- 2015
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Big Data analytics has attracted intense interest from both academia and industry recently for its attempt to extract information, knowledge and wisdom from Big Data. In industry, with the development of sensor technology and Information & Communication Technologies (ICT), reams of high-dimensional data streams are being collected and curated by enterprises to support their decision-making. Fault detection from these data is one of the important applications in eMaintenance solutions with the aim of supporting maintenance decision-making. Early discovery of system faults may ensure the reliability and safety of industrial systems and reduce the risk of unplanned breakdowns. Both high dimensionality and the properties of data streams impose stringent challenges on fault detection applications. From the data modeling point of view, high dimensionality may cause the notorious “curse of dimensionality” and lead to the accuracy deterioration of fault detection algorithms. On the other hand, fast-flowing data streams require fault detection algorithms to have low computing complexity and give real-time or near real-time responses upon the arrival of new samples. Most existing fault detection models work on relatively low-dimensional spaces. Theoretical studies on high-dimensional fault detection mainly focus on detecting anomalies on subspace projections of the original space. However, these models are either arbitrary in selecting subspaces or computationally intensive. In considering the requirements of fast-flowing data streams, several strategies have been proposed to adapt existing fault detection models to online mode for them to be applicable in stream data mining. Nevertheless, few studies have simultaneously tackled the challenges associated with high dimensionality and data streams. In this research, an Angle-based Subspace Anomaly Detection (ABSAD) approach to fault detection from high-dimensional data is developed. Both analytical study and numerical illustration demonstrated the efficacy of the proposed ABSAD approach. Based on the sliding window strategy, the approach is further extended to an online mode with the aim of detecting faults from high-dimensional data streams. Experiments on synthetic datasets proved that the online ABSAD algorithm can be adaptive to the time-varying behavior of the monitored system, and hence applicable to dynamic fault detection.
|
|
| 22. |
- Zhang, Liangwei
(författare)
-
Big Data Analytics for Fault Detection and its Application in Maintenance
- 2016
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Big Data analytics has attracted intense interest recently for its attempt to extract information, knowledge and wisdom from Big Data. In industry, with the development of sensor technology and Information & Communication Technologies (ICT), reams of high-dimensional, streaming, and nonlinear data are being collected and curated to support decision-making. The detection of faults in these data is an important application in eMaintenance solutions, as it can facilitate maintenance decision-making. Early discovery of system faults may ensure the reliability and safety of industrial systems and reduce the risk of unplanned breakdowns.Complexities in the data, including high dimensionality, fast-flowing data streams, and high nonlinearity, impose stringent challenges on fault detection applications. From the data modelling perspective, high dimensionality may cause the notorious “curse of dimensionality” and lead to deterioration in the accuracy of fault detection algorithms. Fast-flowing data streams require algorithms to give real-time or near real-time responses upon the arrival of new samples. High nonlinearity requires fault detection approaches to have sufficiently expressive power and to avoid overfitting or underfitting problems.Most existing fault detection approaches work in relatively low-dimensional spaces. Theoretical studies on high-dimensional fault detection mainly focus on detecting anomalies on subspace projections. However, these models are either arbitrary in selecting subspaces or computationally intensive. To meet the requirements of fast-flowing data streams, several strategies have been proposed to adapt existing models to an online mode to make them applicable in stream data mining. But few studies have simultaneously tackled the challenges associated with high dimensionality and data streams. Existing nonlinear fault detection approaches cannot provide satisfactory performance in terms of smoothness, effectiveness, robustness and interpretability. New approaches are needed to address this issue.This research develops an Angle-based Subspace Anomaly Detection (ABSAD) approach to fault detection in high-dimensional data. The efficacy of the approach is demonstrated in analytical studies and numerical illustrations. Based on the sliding window strategy, the approach is extended to an online mode to detect faults in high-dimensional data streams. Experiments on synthetic datasets show the online extension can adapt to the time-varying behaviour of the monitored system and, hence, is applicable to dynamic fault detection. To deal with highly nonlinear data, the research proposes an Adaptive Kernel Density-based (Adaptive-KD) anomaly detection approach. Numerical illustrations show the approach’s superiority in terms of smoothness, effectiveness and robustness.
|
|
| 23. |
- Zhang, Liangwei, et al.
(författare)
-
Big Data Mining in eMaintenance : An Overview
- 2014
-
Ingår i: Proceedings of the 3rd international workshop and congress on eMaintenance. - Luleå : Luleå tekniska universitet. - 9789174399738 - 9789174399738 ; , s. 159-170
-
Konferensbidrag (refereegranskat)abstract
- Maintenance related data are tending to be increasingly huge involume, rapid in velocity and vast in variety. Data with thesecharacteristics bring new challenges with respect to data analysisand data mining, which requires new approaches andtechnologies. In industry, related research and applications, somecontributions have been provided to utilize Big Data technologiesfor extraction of information through pattern recognitionmechanisms via eMaintenance solutions. Today, the existingcontributions are not enabling a holistic approach for maintenancedata analysis and therefore are insufficient. However, theimmense value hidden inside the Big Data in eMaintenance isarousing more and more attention from both academia andindustry. Hence, this paper aims to explore eMaintenancesolutions for maintenance decision-making through utilization ofBig Data technologies and approaches. The paper discusses BigData mining in eMaintenance through a general manner byemploying one of the widely accepted frameworks with the nameof Cross Industry Standard Process for Data Mining (CRISPDM).In addition, the paper outlines features of maintenance dataand investigates six sub-processes (i.e. business understanding,data understanding, data preparation, modeling, evaluation anddeployment) of data mining applications defined by CRISP-DMwithin the domain of eMaintenance.
|
|
| 24. |
|
|
| 25. |
- Zhang, Liangwei, et al.
(författare)
-
Sliding Window-based Fault Detection from High-dimensional Data Streams
- 2017
-
Ingår i: IEEE Transactions on Systems, Man & Cybernetics. Systems. - : IEEE. - 2168-2216 .- 2168-2232. ; 47:2, s. 289-303
-
Tidskriftsartikel (refereegranskat)abstract
- High-dimensional data streams are becoming increasingly ubiquitous in industrial systems. Efficient detection of system faults from these data can ensure the reliability and safety of the system. The difficulties brought about by high dimensionality and data streams are mainly the ``curse of dimensionality'' and concept drifting, and one current challenge is to simultaneously address them. To this purpose, this paper presents an approach to fault detection from nonstationary high-dimensional data streams. An angle-based subspace anomaly detection approach is proposed to detect low-dimensional subspace faults from high-dimensional datasets. Specifically, it selects fault-relevant subspaces by evaluating vectorial angles and computes the local outlier-ness of an object in its subspace projection. Based on the sliding window strategy, the approach is further extended to an online mode that can continuously monitor system states. To validate the proposed algorithm, we compared it with the local outlier factor-based approaches on artificial datasets and found the algorithm displayed superior accuracy. The results of the experiment demonstrated the efficacy of the proposed algorithm. They also indicated that the algorithm has the ability to discriminate low-dimensional subspace faults from normal samples in high-dimensional spaces and can be adaptive to the time-varying behavior of the monitored system. The online subspace learning algorithm for fault detection would be the main contribution of this paper.
|
|
