Description:
Multiple points fault in the rotary machinery induce complex vibration signatures, which have the tendency to mislead the fault diagnostic models. One of the challenging problems in machine fault diagnosis is to model and study fault signatures dynamics in case of multiple points fault. The existing literature lacks in the study of multiple points fault and the associated vibration signatures. In this study, a multiple-points defect model (MPDM) is proposed which can formulate n-points bearing fault signature's dynamics. Impact of multiple points defect on the well-established state-of-the-art fault diagnostic models is quantified in terms of fault detection accuracy. Results for fault detection accuracy are obtained using Support Vector Machine (SVM) and a modification is recommended in the existing fault diagnostic models in order to nullify the impact of multiple-points fault.
Description:
Inchoate fault detection for machine health monitoring (MHM) demands high level of fault classification accuracy under poor signal-to-noise ratio (SNR) which persists in most industrial environment. Vibration signals are extensively used in signature matching for abnormality detection and diagnosis. In order to guarantee improved performance under poor SNR, feature extraction based on statistical parameters which are immune to Gaussian noise becomes inevitable. This paper proposes a novel framework for adaptive feature extraction based on higher order cumulants (HOCs) and wavelet transform (WT) (AFHCW) for MHM. Features extracted based on HOCs have the tendency to mitigate the impact of Gaussian noise. WT provides better time and frequency domain analysis for the nonstationary signals such as vibration in which spectral contents vary with respect to time. In AFHCW, stationary WT is used to ensure linear processing on the vibration data prior to feature extraction, and it helps in mitigating the impact of poor SNR. K-nearest neighbor classifier is used to categorize the type of the fault. Simulation studies show that the proposed scheme outperforms the existing techniques in terms of classification accuracy under poor SNR.
Description:
Inchoate fault detection for machine health monitoring (MHM) demands high level of fault classification accuracy under poor signal-to-noise ratio (SNR) which persists in most industrial environment. Vibration signals are extensively used in signature matching for abnormality detection and diagnosis. In order to guarantee improved performance under poor SNR, feature extraction based on statistical parameters which are immune to Gaussian noise becomes inevitable. This paper proposes a novel framework for adaptive feature extraction based on higher order cumulants (HOCs) and wavelet transform (WT) (AFHCW) for MHM. Features extracted based on HOCs have the tendency to mitigate the impact of Gaussian noise. WT provides better time and frequency domain analysis for the nonstationary signals such as vibration in which spectral contents vary with respect to time. In AFHCW, stationary WT is used to ensure linear processing on the vibration data prior to feature extraction, and it helps in mitigating the impact of poor SNR. K-nearest neighbor classifier is used to categorize the type of the fault. Simulation studies show that the proposed scheme outperforms the existing techniques in terms of classification accuracy under poor SNR.
Description:
New malware outbreaks cannot provide thousands of training samples which are required to counter malware campaigns. In some cases, there could be just one sample. So, the defense system at the firing line must be able to quickly detect many automatically generated variants using a single malware instance observed from the initial outbreak by tatically inspecting the binary executables. As previous research works show, statistical features such as term frequency-inverse document frequency and n-gram are significantly vulnerable to attacks by mutation through reinforcement learning. Recent studies focus on raw binary executable as a base feature which contains instructions describing the core logic of the sample. However, many approaches using image-matching neural networks are insufficient due to the malware mutation technique that generates a large number of samples with high entropy data. Deriving instruction cognitive representation that disambiguates legitimate instructions from the context is necessary for accurate detection over raw binary executables. In this paper, we present a novel method of detecting semantically similar malware variants within a campaign using a single raw binary malware executable. We utilize Discrete Fourier Transform of instruction cognitive representation extracted from self-attention transformer network. The experiments were conducted with in-the-wild malware samples from ransomware and banking Trojan campaigns. The proposed method outperforms several state of the art binary classification models.
Description:
Trade-off between energy-efficiency and reliability in wireless sensor networks is application dependent. Without the reliability, the extended lifetime of a network is of limited use. Due to the inherent correlation between reliability and congestion, it is necessary to reduce congestion to improve reliability. Existing congestion control algorithms in wireless sensor networks are reactive. They attempt to reduce the congestion only after its detection. In this paper, we present Location Aided Congestion Aware Routing (LACAR) protocol that proactively avoids congestion formation and improves data delivery success rate in data gathering wireless sensor networks. Location, energy and congestion information of neighbours together with the location information of the base station determine appropriate routes. Simulation results show that LACAR achieves high packet success rate in an energy-efficient way.
Description:
The study is focused on estimating the severity level of the bearing faults which helps in determining the residual life of the equipment and planned maintenance. A novel technique, adaptive severity estimation model (ASEM) is proposed based on adaptive selection of wavelet decomposition nodes and support vector machines. Vibration data from multiple severity levels are used to build the fault estimation model. An adaptive criterion for wavelet decomposition node selection is developed which helps ASEM to achieve robustness in estimating fault severity under varying signal to noise ratio (SNR), a key demand in industrial environment. The simulated data with known severity level is used to parameterize the estimation model. The fault severity estimation performance of ASEM is also validated for the real vibration data and its robustness is gauged under varying SNR conditions.
Description:
Early stage faults detection for machine health monitoring demands high level of fault classification accuracy under poor signal-to-noise ratio (SNR). Vibration signal which is used for signature matching in case of abnormality detection and diagnosis, requires robust tools such as wavelet transform (WT) for time-frequency analysis. WT is specifically used to deal with nonstationary signals. In order to guarantee improved performance under poor SNR, this paper proposes a scheme for feature extraction based on fourth-order cumulant and stationary wavelet transform (FoCSWT). Higher order cumulants have the tendency to mitigate the impact of Gaussian noise. Fourth-order cumulant corresponds to the "peakedness" of the random distribution and the fault detection capability quantifies it as the most dominant cumulant among higher order statistics. Stationary wavelet transform is used to avoid down-sampling on the vibration data prior to feature extraction which gives better estimation of statistical parameters of the data distribution and gives performance enhancement in terms of fault classification accuracy. Simulation studies show that FoCSWT outperforms the existing techniques in terms of fault detection accuracies under poor SNR.
Description:
Mining of sensor data for useful knowledge extraction is a very challenging task. Existing works generate sensor association rules using occurrence frequency of patterns to extract the knowledge. These techniques often generate huge number of rules, most of which are non-informative or fail to reflect true correlation among sensor data. In this paper, we propose a new type of behavioral pattern called associated sensor patterns which capture association-like co-occurrences as well as temporal correlations which are linked with such co-occurrences. To capture such patterns a compact tree structure, called associated sensor pattern tree (ASP-tree) and a mining algorithm (ASP) are proposed which use pattern growth-based approach to generate all associated patterns with only one scan over dataset. Moreover, when data stream flows through, old information may lose significance for the current time. To capture significance of recent data, ASP-tree is further enhanced to SWASP-tree by adopting sliding observation window and updating the tree structure accordingly. Finally, window size is made dynamically adaptive to ensure efficient resource usage. Different characteristics of the proposed techniques and their computational complexity are presented. Experimental results show that our approach is very efficient in discovering associated sensor patterns and outperforms existing techniques.
Description:
WSNs generate a large amount of data in the form of data stream; and mining these streams to extract useful knowledge is a highly challenging task. Existing works proposed in literature use sensor association rules measured in terms of occurrence frequency of patterns. However, these rules often generate a huge number of rules, most of which are non-informative or fail to reflect the true correlation among data objects. Additionally mining associated sensor patterns from sensor stream data, which is vital for real-time applications, has not been addressed yet in literature. In this paper, we address these problems and propose a new type of sensor behavioral pattern called associated sensor patterns which capture simultaneously association-like co-occurrence as well as substantial temporal correlations implied by such co-occurrences in sensor data. We propose a novel tree structure, called associated sensor pattern stream tree (ASPS-tree) and a new technique, called associated sensor pattern mining of data stream (ASPMS), using sliding window-based associated sensor pattern mining for WSNs. By capturing the useful knowledge of the data stream into an ASPS-tree, our ASPMS algorithm can mine associated sensor patterns in the current window with frequent pattern (FP)-growth like pattern-growth method. Extensive experimental analyses show that our technique is very efficient in discovering associated sensor patterns over sensor data stream.
Description:
Machine condition monitoring is a process of continuously observing the status of a machine to ensure that proactive measures are taken to prevent damage due to abnormal operating conditions. Generally, industrial units such as mining, oil and gas etc, are located in geographically remote areas, so a large amount of data need to be acquired for fault diagnosis and prognosis remotely. To achieve this, certain resources such as stable communication network and adequate bandwidth are required. Furthermore, it is not always feasible to dispatch human resources simultaneously over large areas of operation to perform on-site maintenance. To overcome these issues, a mobile agent based system architecture is proposed for machine condition monitoring by imitating human immune system (ACMIS), which is also known as artificial immune system. The experiment results are presented to evaluate the performance of the proposed system in terms of fault detection accuracy and bandwidth allocation. Overall performance evaluation of the proposed framework suggests that our proposed scheme not only provides excellent fault detection accuracy but also a flexible and reliable machine condition monitoring system with reduced network and computational resources. Further our approach provides cost effective solution in building a practical machine condition monitoring system.
Description:
The use of smartphones and hand-held devices continues to increase with rapid development in underlying technology and widespread deployment of numerous applications including social network, email and financial transactions. Inevitably, malware attacks are shifting towards these devices. To detect mobile malware, features representing the characteristics of applications play a crucial role. In this work, we systematically studied the impact of all categories of features (i.e., permission, application programmers interface calls, inter component communication and dynamic features) of android applications in classifying a malware from benign applications. We identified the best combination of feature categories that yield better performance in terms of widely used metrics than blindly using all feature categories. We proposed a new technique to include contextual information in API calls into feature values and the study reveals that embedding such information enhances malware detection capability by a good margin. Information gain analysis shows that a significant number of features in ICC category is not relevant to malware prediction and hence, least effective. This study will be useful in designing better mobile malware detection system.
Description:
Due to its widespread use, with numerous applications deployed everyday, smartphones have become an inevitable target of the malware developers. This huge number of applications renders manual inspection of codes infeasible; as such, researchers have proposed several malware detection techniques based on automatic machine learning tools. Deep learning has gained a lot of attention from the malware researchers due to its ability of capture complex relationships among inputs and outputs. However, deep learning models depend largely on several hyper-parameters (i.e., learning rate, batch size, dropout rate). Hence, it is of utmost importance to analyze the effect of these parameters on classifier performance. In this paper, we systematically studied the effect of these parameters along with the effect of network architecture. We showed that building arbitrary deep networks does not always improve classifier performance. We also determined the combination of hyper-parameters that yields best result. This study will be useful in building better deep neural network based model for malware classification.
Description:
Vertical handoff in next generation heterogeneous wireless networks is a multi-dimensional issue. In this article we propose a multi-dimensional adaptive SINR based vertical handoff algorithm (MASVH) which uses the combined effects of SINR, user required bandwidth, user traffic cost and utilization from participating access networks to make handoff decisions for multi-attribute QoS consideration. Simulation results confirm that the new MASVH algorithm improves the system performance in terms of higher throughput and lower dropping probability, as well as reduces the user traffic cost for accessing the integrated wireless networks.
Description:
Features extraction has always been crucial in rotary machines for Condition based machine health monitoring. Time-domain-segmentation being among the preliminary steps for further classification process plays a momentous role. Vibration signals from bearing are quasistationary in nature therefore calculation of constituent frequencies amplitudes in the vibration signal is dependent upon time-segmentation-window size. The proposed research confers the effects of time-segmentation window size on spectral features amplitudes calculation and its impacts on classification accuracy of the Artificial Neural Network (ANN). Using multi-size time-segmentation-window, for comprehensive spectral features calculation, ANN pattern classifier has been trained for enhanced classification. ANN learning assigns importance based relative weights to the links using supervised learning. Experimental results have shown that multi-size-window spectral features for ANN fault classifier perform efficiently for quasi-stationary bearing vibrations.
Description:
The Internet of Things (IoT) has significantly increased the number of devices connected to the Internet ranging from sensors to multi-source data information. As the IoT continues to evolve with new technologies number of threats and attacks against IoT devices are on the increase. Analyzing and detecting these attacks originating from different sources needs machine learning models. These models provide proactive solutions for detecting attacks and their sources. In this paper, we propose to apply a supervised machine learning classification technique to identify cyber-attacks from each source. More precisely, we apply the incremental piecewise linear classifier that constructs boundary between sources/classes incrementally starting with one hyperplane and adding more hyperplanes at each iteration. The algorithm terminates when no further significant improvement of the separation of sources/classes is possible. The construction and usage of piecewise linear boundaries allows us to avoid any possible overfitting. We apply the incremental piecewise linear classifier on the multi-source real world cyber security data set to identify cyber-attacks and their sources.
Description:
Proceedings of the IEEE International Conference on Industrial Technology
Description:
The Internet of Things (IoT) has significantly increased the number of devices connected to the Internet ranging from sensors to multi-source data information. As the IoT continues to evolve with new technologies number of threats and attacks against IoT devices are on the increase. Analyzing and detecting these attacks originating from different sources needs machine learning models. These models provide proactive solutions for detecting attacks and their sources. In this paper, we propose to apply a supervised machine learning classification technique to identify cyber-attacks from each source. More precisely, we apply the incremental piecewise linear classifier that constructs boundary between sources/classes incrementally starting with one hyperplane and adding more hyperplanes at each iteration. The algorithm terminates when no further significant improvement of the separation of sources/classes is possible. The construction and usage of piecewise linear boundaries allows us to avoid any possible overfitting. We apply the incremental piecewise linear classifier on the multi-source real world cyber security data set to identify cyber-attacks and their sources.
Description:
Proceedings of the IEEE International Conference on Industrial Technology
Description:
Condition based maintenance (CBM) in the process industry helps in determining the residual life of equipment, avoiding sudden breakdown and facilitating the maintenance staff to schedule repairs by optimizing demand–supply relationships. One of the prevalent issues in CBM is to predict the residual life of the equipment. This paper proposes a novel framework to predict the remnant life of the equipment, called residual life prediction, based on optimally parameterized wavelet transform and multi-step support vector regression (RWMS). In optimally parameterized wavelet transform, a generalized criterion is proposed to select the wavelet decomposition level which works for all the applications; decomposition nodes are selected by characterizing their dominancy level based upon relative fault signature–signal energy contents. The prediction model is based on multi-step support vector regression to determine the nonlinear crack propagation in the rotary machine according to Paris’s fatigue model. The results both for the simulated as well as the actual vibration datasets validate the enhanced performance of RWMS in comparison with the existing techniques to predict the residual life of the equipment.